ResearchPortfolioTransformative technologiesfor personalized, vigilanthealth monitoring2022-2023
MISSIONASSIST is developing leading-edge systems for high-value applications such as healthcare andIoT by integrating fundamental advances in energy harvesting, low-power electronics, andsensors with a focus on usability and actionable data. We do this by bringing togethermultidisciplinary researchers, practitioners, and industry partners in a diverse and inclusiveecosystem that encourages innovation with a focus on education and outreach.VISIONASSIST focuses on creating self-powered sensing, computing, and communication systemsto enable data-driven insights for a smart and healthy world.
Engineered Systems333435363738 3940AI-Driven, Resilient and Adaptive Monitoring of Sleep (AI-DReAMS)Bio-Electro-Photonic Microsystem Interfaces for Small AnimalsBody Area Network of Inertial Sensors for Clinical Gait RehabilitationCough Detection using Wearables and Out-Of-Distribution RecognitionEnhanced Detection of Impending Problem Behavior in People with Intellectual and DevelopmentalDisabilities through Multimodal Sensing and Machine LearningIntegration and Demonstration of Wearable Monitoring Systems for Asthma and DiabetesMultimodal Biosensing System for Electrochemical & Biophotonic MonitoringTABLE OF CONTENTSEnergy Harvesting & Storage3Flexible Thermoelectric Generators for Body Heat HarvestingModeling of Thermoelectric Generators for Body Heat HarvestingNew Mode of Mechanical-to-Electrical Energy HarvestingPrinting of Stretchable Conductors Enabled by Highly Tunable Multiphase Liquid Metal PastesSelf-Powered Smart Insoles for Gait and Balance Detection45678Low Power Sensing910 11 1213141516Development of a Wearable Ultrasound Transducer for Sensing Muscle Activities in Assistive RoboticsApplicationsImproving the Performance and Design of Potentiometric Biosensors for the Detection of ExtracellularHistones in Blood with Deep LearningMicroneedle Sensing System for Minimally Invasive Extraction and Analysis of Dermal Interstitial FluidSoft, Skin-mounted, Microfluidics for On-Demand Sweat Sensing Ultra Low Power Metal Oxide Electronic Nose Arrays for Skin, Breath and Environmental MonitoringUltra Low Power PhotoPlethysmoGraphy (PPG) Module for Biomedical ApplicationsZero-Power Wearable Sweat Assays and Long-Term Sensing Device Interfaces Using Osmotic Pumpingand Paper MicrofluidicsForeword217Low Power Systems-on-Chip1819202122Energy-Efficient Circuit Design for Ultra-Miniature, Injectable and Subcutaneous Physiology CapsulesSelf-Powered Bluetooth Backscatter EKG Sensor SystemUltra Low Power SoC and RadiosUltrasonic Energy/Data Transfer for ImplantablesPoly(Octamethylene Maleate (anhydride) Citrate) (POMaC) Circuits23Smart Textiles and Flexible Materials242526272829303132A Biaxially Stretchable and Self-Sensing Textile Heater Using Silver Nanowire CompositeA Wearable Electrocardiography Armband Resilient Against ArtifactsCurvilinear Soft Electronics by Micromolding of Metal Nanowires in CapillariesDesign Factors for High Performance Electrocardiography GarmentsNovel Tactile Sensing Wearable Devices from Hierarchical 3D Printed HydrogelsNovel Textile-Based Sensors for Inner Prosthetic Socket Environment MonitoringTextile Based Soft Actuators for Compliant and Wearable Artificial MusclesUltrasoft Porous 3D Conductive Dry Electrodes for Electrophysiological Sensing and Myoelectric ControlWet Spinning of Advanced Materials into Energy-Storage Yarns for Smart Textiles
Welcome to the ASSIST Center’s second annual research portfolio. We are excited to share our research advances in coreprojects as well as introduce several new research directions. Ten years ago, the ASSIST team set out to create disruptive, always-on wearable devices that would enable continuousmonitoring for chronic disease management. We achieved this through unique co-engineering of energy harvesting, lowpower systems-on-chip, low power sensing and integration on flexible platforms such as textiles. ASSIST built these systems tomeet the requirements of several key chronic health concerns such as asthma, cardiac disease, diabetes, and woundmonitoring. We have built numerous partnerships with clinicians to validate our wearable systems and with industry to translateour intellectual property and technologies. The ASSIST Center is led by NC State University and includes Penn State University,University of Virginia, Florida International University, University of Michigan, University of North Carolina at Chapel Hill, Universityof Utah, and University of Texas at Austin as partners. We maintain our technical leadership in five theme areas: energyharvesting and storage, low power sensing, low power systems-on-chip, smart textiles and flexible materials, and engineeredsystems. This year has been a pivotal point for the Center. In September 2022, after ten productive years, we graduated from theNational Science Foundation. During this time, we have filed 90 inventions, spun off 10 companies, generated 53 patents,graduated 105 Ph.D.s, published over 650 papers and engaged hundreds of students in our educational pipeline. On behalfof the entire Center, we are genuinely grateful to the National Science Foundation for this opportunity. As we celebrate all that we have accomplished, we remain steadily focused on our vision of gaining vigilant data-driveninsights into human health. To this end, in our self-sufficiency phase, we have added several new directions into the Center’smission in addition to our core projects. These new projects, highlighted in this portfolio, range from implantable devices andnovel ultrasound-based sensing and imaging, to the novel use of textile fibers. In addition, while we have primarily focused onpassive monitoring and sensing to date, we are now adding actuation mechanisms through advanced flexible materials anddesigns in wearable systems. Actuation devices are the next frontier for wearable systems, opening new opportunities inhuman comfort as well as pain management and behavioral modification. The ASSIST Center and its team are uniquelypositioned to make disruptive changes in this space. In the past year we welcomed new members to our leadership team who are driving critical programs for the next generationof ASSIST. Dr. Alper Bozkurt has taken over the role of deputy director and is building our core strength in implantables as wellas new use cases such as healthy aging and early detection of cognitive decline. Dr. Bozkurt had already been part of theleadership team as a Testbed Leader since the Center’s inception and will continue in this role to lead our systems integrationefforts. Dr. Ravi Chilukuri has taken the helm of the industry program as our new Innovation Ecosystem Director. He brings hislong track record of industry research and development as well as business development to further expand ASSIST’s industryprogram. Under his leadership we have already welcomed several new members to the ASSIST industry board. Our faculty members are getting international recognition for their efforts. Dr. Ömer Oralkan and Dr. Amay Bandodkarreceived the IEEE Sensors Council research excellence awards. Dr. Douglas Werner was inducted into the national academyof inventors. Drs. Ömer Oralkan, Xiaoning Jing, and Shekhar Bhansali all became members of the 2023 Class of IEEE Fellows, thehighest technical honor given by the society. Our Education Director, Dr. Elena Veety, continues to build a community of all levels of students by offering programs thattarget K-12 students and teachers, and providing undergraduate research opportunities and graduate assistantships. We alsoprovide training for all our students in translational skills not taught in the classroom. In the summer of 2022, we hosted newweeklong summer camps for high school students. Furthering our goal of workforce development for students andprofessionals, we offered a new hands-on data analytics short course in December, 2021. Our first annual ImplantablesWorkshop took place in May 2022, which was a huge success, so much so that we plan to offer this again in May 2023. Wealso plan to host a remote workshop to understand the needs of early detection Alzheimer's disease and related dementias. As we enter our 11th year, we are pushing ahead in our vision to disrupt healthcare through continuous monitoring. As youreview this research portfolio, I invite you to reach out to us for collaborative opportunities. Sincerely,Dr. Veena MisraM.C. Dean Distinguished University ProfessorNC State UniversityFOREWORDVeena
Flexible Thermoelectric Generatorsfor Body Heat HarvestingThe flexible thermoelectric generator with rigid thermoelectric pellets andliquid metal interconnects (left). Mechanical strain testing showing devicesable to withstand 280% strain (right). Principal Investigator: Dr. Mehmet C. Ozturk, Dr. Veena Misra, Dr. Michael Dickey, Dr. Daryoosh VashaeeObjective: The objective of this program is to build high-performance flexible thermoelectric generators, whichcan conform to the human body, providing bothperformance and aesthetics. Thermoelectric generators(TEGs) that can convert body heat to electricity are ofinterest to realize self-powered wearable sensor systems,which can provide hassle-free, long-term, continuousmonitoring. Such devices can significantly improvemanagement of chronic conditions such ascardiovascular diseases and increase patients’ quality oflife. Approach: Our unique patented approach employs industry-standard rigid semiconductor pellets, which areembedded in a flexible elastomer. The pellets areconnected in series using eutectic gallium-indium (EGaIn)liquid metal interconnects, which provide excellentflexibility, stretchability, and electrical conduction. Byemploying rigid pellets currently used in commercial rigidTEGs, this approach eliminates the need for newthermoelectric material development, thereby providinga low-cost-of-ownership flexible TEG option to existing TEGmanufacturers.Key Accomplishments:Our group was the first to propose flexible TEGs that reliedon the same rigid thermoelectric pellets used incommercial TEGs. Our group was also the first to proposethe use of liquid metal interconnects in flexiblethermoelectric generators. To date, our team hasdeveloped several new elastomer composites to improvethe thermal engineering of our devices. Our latest work isfocusing on the reliability of our flexible TEGs. In theseexperiments, the devices are subjected to extrememechanical, thermal and electrical stressors. The figureshows one of our flexible TEGs being subjected to 280%strain. Our devices can withstand such extreme levels ofmechanical strain thanks to the stretchable, self-healingEGaIn interconnects. Impact: Our latest devices provide best-in-class performanceamong published flexible generators and rival theperformance of rigid thermoelectric modules, providing acontinuous supply of energy to realize self-poweredwearable monitoring devices. Funding Sources:ASSIST CenterNSF PFI programStudents/Postdocs: Shima Arab, Tvisha Shete & Farzad Mohaddes, PhD4
Modeling of ThermoelectricGenerators for Body Heat HarvestingExperimental results showing agreement with the model whencomparing measured and calculated output voltages of aflexible TEG during various activities.Objective: The objective of this program is to develop efficientanalytical models to accurately predict the performanceof thermoelectric generators (TEGs) on the human bodyunder various contextual scenarios. TEGs that canconvert body heat to electricity are of interest to realizeself-powered wearable sensor systems, which canprovide hassle-free, long-term, continuous monitoring.Such devices can significantly improve management ofchronic conditions such as asthma, cardiovasculardisease and diabetes, and increase patients’ quality oflife. Approach: Our approach includes the use of both highly-efficientanalytic and 3-D numerical simulations. The analyticmodels that we develop provide the ability to quicklyunderstand the impact of different design parameters ondevice performance, while the 3-D numerical modelsprovide a more in-depth understanding of the TEGoperation. The models include the contributions of thedevice architecture, physical dimensions, thermoelectricmaterials, and parasitic thermal/electrical resistances.The models are also designed to take into account theimpact of the human thermoregulatory system, whichdetermines metabolic rate and core body temperature,allowing time-dependent simulations of the TEG output.Impact: Our latest analytical model can not only guide designengineers in improving their device design, but alsoprovide the ability to factor in the user’s age, sex, weight,and height. As an example, our results indicate that olderindividuals will generate approximately 30% less powerthan younger ones, which is significant because olderusers would likely benefit significantly from self-poweredoperation enabling continuous, long-term monitoring. Themodeling of the impact of physical activity is especiallyimportant for sports performance monitoring.5Principal Investigator: Dr. Mehmet C. Ozturk Funding Sources: ASSIST Center Students/Postdocs:Shima Arab, Tvisha Shete & Farzad Mohaddes, PhDKey Accomplishments:Our model, published in 2016, was the first 3-D analyticalmodel for TEG simulation. These simulations werecomplemented by 3-D numerical simulations using theCOMSOL simulation environment. Our latest model, whichis a first of its kind, includes the impact of the humanthermoregulatory system and physical activity. The modelwas published in Applied Energy in August 2022. Applied Energy
New Mode of Mechanical-to-Electrical Energy HarvestingSchematic concept for converting mechanical displacement (of a metal alloy, shown here as a sphere in a saline hydrogel) to electricityusing a new soft-stretchable supercapacitor device. The opposite charges in the double layer are shown schematically as red and blue.Multiple modes of deformation change the geometry and thereby force current to move, producing electricity.Objective:Mechanical energy is often wasted (i.e., not harvested) inthe form of engine vibrations, wind, and ocean waves.We seek a new, simple approach by creating a new typeof “variable area soft capacitor” in which anymechanical inputs (compression, stretching, twisting,etc.) can generate electricity. The device is built entirelyfrom soft materials, thus making it compatible with thehuman body. It also is distinguished by the use of saltwater, making it compatible with ocean energyharvesting or on-body sweat.Approach:We are harvesting mechanical energy and converting itto electricity using a completely new soft “variable areaelectrochemical supercapacitor”. Mechanical input fromstretching, deforming, or oscillating the device causescharge to move (i.e., generate electricity) in/out of acircuit. The idea is to utilize a non-toxic, low melting metalalloy as a stretchable electrode to fabricate variablearea capacitors that convert mechanical to electricalenergy. This metal alloy, EGaIn (eutectic gallium indium),is a non-toxic liquid metal at room temperature. Whenmetals are placed in saltwater, they form a so-called“electrical double layer” at their surface (positive andnegative charges). This principle is used commercially insupercapacitors to store large amounts of energy.Whereas conventional supercapacitors use rigid, porous Impact:The most exciting aspect of this approach is the favorablescaling physics that can potentially lead to large levels ofharvested energy. The current scales with the area of theelectrodes, thus the power scales with area squared. Todate, we have increased the area by simply making thedevice larger, but a more promising approach is toincrease the electrode area per volume by percolatingdroplets of the metal.Key Accomplishment:We have published a paper in vanced Materials thatestablishes the concept. carbon electrodes to store energy, we propose to makesuch electrodes out of stretchable liquid metal togenerate energy because of the ability to changegeometry. When the geometry changes (due tomechanical energy input), the capacitance changesand charge moves through a circuit as electricity. The as-fabricated device can generate power from a widerange of frequencies and modes of deformation(compression, strain, twisting, etc.).Advanced Materials Principal Investigator:Dr. Michael DickeyFunding Sources:ASSIST CenterNano-Bio Materials Consortium6
Printing of StretchableConductors Enabled by HighlyTunable Multiphase LiquidMetal PastesDifferences between printing a common 3D printing “ink”(a) and liquid metals (b-c). This project seeks to modifyliquid metals so they can be extruded like a paste (i.e. asshown in (a))Objective:To demonstrate a novel class of multiphase liquid metal(LM) pastes whose properties can be designed throughsystematic incorporation of solid additives and fluidmicro-capsules with nanometer thin oxide shells.Introduction of these easily tunable pastes could enablelarge scale adoption of the materials in stretchableelectronic, thermal management, and medicalapplications. Currently, such deployment is inhibited bymanufacturing issues stemming from the large surfacetension of LMs (making it difficult to pattern and adhereto surfaces).Approach:By incorporating solid and fluid fillers, this project seeks torender LMs easier to pattern by additive manufacturing,broaden their range of physical and chemical properties,and, by decreasing the overall metallic content, increasetheir economic appeal. To date, researchers have shownit is easy to distribute liquid metal droplets in othermaterials (such as polymers), but difficult to do theopposite. The mixing of secondary fluids or solids into LM isa surprisingly non-trivial task due to the high cohesiveenergy density of the metal. Based on our preliminaryresults, we assert that the rapid surface oxidation of LMsenables a general pathway for achieving this task.Key Accomplishments:This is a new project. To date, we have demonstrated theprinciple and have shown it is possible to print liquid metalin this manner.Impact:This project will create soft metallic materials withcompletely new properties while retaining electricalconductivity to make them more manufacturable. Ourinterest is tuning the rheological properties to enablefacile printing of metallic materials at room temperature.Yet, there are many other properties that can be tuned byforming foams and pastes, such as adhesion.Principal Investigator:Dr. Michael Dickey7Funding Source: NSF in collaboration with ASU
Self-Powered Smart Insoles for Gait andBalance DetectionFig. 1 Shoe based energy harvester and example voltageoutput. Harvester generates 9.9 mW when walking at 1 Hz.Principal Investigators:Dr. Shad Roundy, Dr. Susan Trolier-McKinstryFig. 2: Pressure sensing array on metal foil being tested with a heel prototype (left).Example voltage waveforms from four PZT sensors on array (right).Fig. 3: Microfabricated array of spacecharge flexoelectric pillars8Students:Travis Peters, Arash KazemiFunding Sources:ASSIST CenterOffice of Naval ResearchObjective:This project seeks to develop a self-powered pressuresensing array in a shoe insole that can be used in studiestargeting conditions in which gait, balance, and activityare key indicators of disease severity and/or risk.Approach:There are three technical thrusts in the project:development of a shoe insole harvester that iscomfortable and does not affect the gait of the user,development of novel space-charge polarizableflexoelectric transducers, and development of a flexiblelarge-area piezoelectric pressure sensing array. Recently,the energy harvester was redesigned to improve comfort,power generation, and robustness. Several shoe insoleswere characterized by measuring the force-displacement relationship (i.e. stiffness) and the energydissipation. The harvester was then designed to mimic thestiffness and energy dissipation of these standard walkingshoes. This re-design includes a very small ball-screw anda multi-lobe cam which was found to be more robust. Thenew design can incorporate different frequency upconversion factors. With a 3-lobe cam the harvestergenerates 9.9 mW of power at 1 Hz actuation (similar towalking at about 2.5 mph) which nearly meets our targetof 10 mW. The estimated generated power from thisdesign was 19 mW. It is believed that the discrepancy ismostly due to compliance in harvester components otherthan the piezoelectric beams (see Fig. 1). Impact:Falls result in >3,000,000 trips to the emergency room,28,000 deaths, and $50B in costs, annually. This work aimsto enable a technology that will allow long-term, low-burden monitoring of people at potential risk for falls.Key Accomplishments:Foot pressure sensors were prototyped on polished metalfoils passivated with 100 μm LaNiO and a 50 nm HfO_layer. A piezoelectric layer with a 52/48 Zr/Ti ratio with 2%Nb doping, 12% excess Pb and a molarity of 0.6M. Theprocessed foil was withdrawn from the solution bath at arate of 30 mm/min, pyrolyzed and crystallized. Thepressure sensing foil was characterized in tests in which acylinder was rolled over the array to mimic the action ofthe ball of the foot. We also fabricated a squishyprototype heel that was mounted to a material testmachine (see Fig. 2) to mimic the heel strike. Thepiezoelectric thin film produces significant voltages (~ 0.2volts) without amplification. These voltage signals can beused to predict a contact patch area and a center ofpressure. For flexoelectric based harvesters, a process flow wasdeveloped for space charge induced flexoelectric (SCIF)devices shown in Fig. 3. The major challenges areassociated with curvature of the sample on the releaseprocess, which leads to the pyramids delaminating fromthe polymer matrix. Multiple approaches were adoptedto ameliorate this difficulty, resulting in the first devicesthat could be characterized electromechanically. Inaddition, we are working to validate basic constitutiveequations, equivalent circuit models, and finite elementbased computational models. Standard finite elementformulations do not contain strain gradient as afundamental quantity nor coupling between straingradient and polarization. An initial code was developedin Matlab to validate the SCIF pyramid structures. Oncevalidated, this code will be moved to a morecomputationally efficient platform and carrier diffusionequations will be incorporated to more accurately modelSCIF transducers. Finally, the validated models can beused in design optimization to design SCIF transducers forthe energy harvester.3 2
(Left) (a) Schematic demonstration of the wearable US transducer design. (b) The fabrication process for the wearable US transducer. (Right)Demonstration of the wearability and customizability of the transducer: (a) size is about 1.5 cm, which can be used for single muscle measurements;(b) Size is about 10 cm, which can be used for multiple muscles measurements;(c) Demonstration of flexibility for the wearable transducerDevelopment of a Wearable UltrasoundTransducer for Sensing Muscle Activity inAssistive Robotics ApplicationsPrincipal Investigators:Dr. Nitin Sharma, Dr. Xiaoning JiangObjective:Muscle activity sensing plays an important role incontrolling novel assistive robotic (AR) devices formodern medicine. Currently commonly used signals,such as EMG signals and traditional ultrasound (US), arenot suitable for control of modern AR devices due totheir inherent limitations and bulky form factors. Theobjective of this project is to develop a wearableultrasound transducer for sensing muscle activity since,to our knowledge, no other wearable transducer designsfor this application have been reported to date.Approach:We propose to fabricate wearable US arrays using PZT-5A elements embedded in deformable PDMS substrates.PZT is widely used in US transducers with highpiezoelectric and electromechanical properties. Therigidity of PZT led us to select stretchable andbiocompatible PDMS as a flexible substrate for providingmechanical interlinkage between elements. Thewearable US array was constructed after stacking andbonding, dicing, wire connection, and PDMS filling.Wearable US transducers are capable of attaching tothe body area of interest without restricting themovement of the tissue beneath the transducer andpreventing it from shifting. In addition, the sizes of thePDMS substrates, the thickness of the piezo layer, and the Key Accomplishments:To date, our team has fabricated a 4 by 4 US transducerarray consisting of 16 elements, each of which measures1.4mm × 1.4mm in size. On average, the centralfrequency, the -6dB bandwidth, and the electricalimpedance were 10.78MHz, 61%, and 63.85Ω,respectively. Experiments conducted in vitro and in vivodemonstrate the capability of the proposed transducerto monitor muscle activity. Muscle movement imagingwas used to visualize the muscle displacement. Thewearable transducer has a thickness of approximately3mm and is highly stretchable and reliable.Impact:The wearable transducer developed in this project iscapable of accurately monitoring regional muscleactivity. The results demonstrate its high potential forcontrolling AR. Furthermore, customizable devicesenable simultaneous and individual measurements ofmultiple muscle groups, which is important in developingcontrol schemes for AR systems with high degrees offreedom.transducer arrangements can be easily tailored tomatch the location, depth, and shape of the targetmuscle.10Postdocs and Students:Dr. Xiangming Xue, Sunho Moon, Vidisha Ganesh, Krysten LambethFunding sources:NIHNSF
Improving the Performance and Design ofPotentiometric Biosensors for the Detection ofExtracellular Histones in Blood with Deep LearningMeasured SPR titration curve demonstrating the selectiveresponse of the aptamer-functionalized electrodes tocalf thymus (CTH) and human (H4) histones vs. BSA. Theelectrodes are also used for electrochemical sensing.Principal Investigators:Dr. Spyridon Pavlidis, Dr. Edgar Lobaton, Dr. Francis MillerObjective:Out goal is to use standard machine learning techniquesand state-of-the-art deep learning techniques torecognize patterns and relationships in the complex datathat is collected from potentiometric biosensors toimprove their sensitivity and drive design optimization. Thisapproach is being applied to the detection of circulatinghistones in blood, which contribute to the developmentof Multiple Organ Dysfunction Syndrome (MODS), apotentially fatal condition in critically ill patients.Approach:Gold sensing electrodes are functionalized with RNAaptamers to detect extracellular histones with extendedgate sensors. These devices are being evaluated inbuffer, serum, and whole blood. Deep learningtechniques are being leveraged to reveal intricaterelationships and trends to compensate for theconventional losses in sensitivity observed in blood-basedtests. These findings drive the optimal design of thepotentiometric sensors, thus establishing design rules thatcan accelerate the development of these sensors acrossthe community. A major obstacle to the application ofMachine/Deep Learning techniques to biosensing is thegeneration of adequate training data. A multiplexedpotentiometric biosensing platform, made possible by theuse of the extended gate approach, and computersimulations will be developed in order to identify time-and resource-efficient approaches to algorithm training.Key Accomplishments:We have demonstrated physiologically relevant nMdetection of calf thymus histone (CTH) using ourpotentiometric sensors. We have also used surfaceplasmon resonance (SPR) to understand howimmobilization protocols impact sensitivity, selectivity, andstability in response to regeneration. Regression analysis isbeing applied to these experimental data to unveilpatterns, such as drift effects. These analyses have alsobeen done with human histones (H4). Finite elementmodeling is being used to accelerate data generation foralgorithm training.Impact:Most demonstrations of potentiometric sensors stall duringthe translation from testing well-controlled laboratorysolutions to operating in serum or whole blood. Ourintegrated approach aims to overcome this to providebetter sensitivity and reliability. The use of machinelearning remains nascent in this field. Therefore, we willestablish a standardized protocol that other researchers inthe field can leverage in order to accelerate theadoption of potentiometric biosensors in newapplications.Finite element modeling (FEM) is used to generate synthetictraining data at scale for the deep learning algorithms. Here,the impact of mass transport limits on experimental datacan be removed or tuned via simulation. 11Students/Postdocs:Hayley Richardson, Jeffrey Barahona, Joshua KalyanapuFunding Source: NSF
Microneedle Sensing System forMinimally Invasive Extraction andAnalysis of Dermal Interstitial FluidObjective:With the increasing interest in improving diagnostic tests,making them portable and more accessible is becominga significant research and commercial effort. Whileconventional blood-based diagnostic tests increase therisk of transmitting blood-borne pathogens and infection,they are also painful, which reduces compliance acrossthe patient population. Sampling interstitial fluid (ISF) usinga nanocomposite microneedle (MN) patch enables anoninvasive, painless alternative to the common finger-stick blood draw. ISF contains comparable analytes toblood, plasma, urine, saliva, and feces and can besimilarly analyzed.Approach:The proposed technology uses a biocompatiblehydrophilic composite of methacrylated hyaluronic acid(MeHA) and TEMPO-oxidized cellulose nanofibers tofabricate MNs that when inserted into the skin, swell withISF that can easily be recovered from the patch foranalysis. The microneedle patches will be integrated intoa wearable health and environmental tracker (HET)platform using an osmotic pump and screen printedelectrochemical sensors for diagnostics and monitoring.Custom fabrication of master molds for the MNs via 3Dprinting has allowed flexibility in the geometry of thepatches in order to optimize insertion and extraction ofISF.Key Accomplishments:The microneedles and osmotic hydrogel pump extractionsystem has allowed continuous extraction and transfer offluids from a skin model to a paper system. Recent efforts have been focused on creating and optimizingbioreceptors and transducers into a lateral flow system fordiagnostics and into an electrochemical sensor formonitoring. When the systems are optimized, the next stepwill be to test the technology in human trials for futuretranslation.Impact:A significant goal of the project is to develop wearabledevices for detection and monitoring which minimallyaffect the users and can provide real time physiologicalstatus for an extended period of time.SEM image of microneedle patchPrincipal Investigators:Dr. Michael Daniele, Dr. Alper Bozkurt, Dr. Orlin Velev, Dr. Michael DickeyHET platform with osmotic pumpand microneedle patch3D printed master mold for microneedle fabrication12Students: Angelica F Aroche, Hannah Nissan, Kaila PetersonDr. Tamoghna Saha, Dr. Tanner Songkakul, Chris Sharkey, Leslie Uy,Funding Sources:DermiSenseNBMCSEMI
Soft, Skin-Mounted, Microfluidics forOn-Demand Sweat Sensing On Body Testing. Photograph ofstationary bicycle set up showing thedevice mounted on the forearm of thecandidate. Scale bar: 5 mmPrincipal Investigators:Dr. Amay J Bandodkar, Dr. Michael D. DickeyObjective:Evolving the field of wearable sweat sensors to a levelthat truly represents ‘lab-on-skin’ technology will requireincorporation of advanced functionalities that give theuser the freedom to: 1) choose the precise time forperforming sample analysis and 2) select sensors from anarray embedded within the device for performingcondition-specific sample analysis. Unfortunately, presentwearable sensors do not offer such capabilities. Here, wehave developed a soft microfluidic patch whichovercomes these limitations and provides the user withon-demand sweat sampling capabilities.Approach:Our soft, wearable microfluidic system includes uniqueuser-activated micro-pumps and micro-valves that allowthe user to perform on-demand sweat analysis.Pneumatic pumping of sweat from inlet to sensingchamber is accomplished via a simple finger actuatedpull tab. Colorimetric assays allow quick measurement of4 common sweat analytes. The entire device is conformalto the skin and features robust attachment via a skin safeadhesive liner.Key Accomplishments:The device’s key components were optimized inbenchtop studies. A pump membrane thickness of 200µm achieved complete pumping and preventedaccidental sample ejection from the device. In thesensing chamber, cellulose membrane volume wastailored to evenly distribute fluid and ensure sufficientcolorimetric assay wetting. The effect of mechanicaldeformation and inertial forces on device performancewas found to be minimal. A two day on-body studydemonstrated the patch’s robust performance in 4consecutive exercise bouts and illustrates the device’sability to offer on-demand, longitudinal, and multi-analytesensing.Impact:This new platform will facilitate advanced sweat handlingdirectly on the body for deeper understanding of howsweat composition changes depending on the person'sphysiological state. Furthermore, the capability to bringsweat to the sensing chamber on-demand will enableincorporation of incubation-based sensors.Wearable microfluidics for on-demand, longitudinal, and multi-analyte sweat monitoring. (A) Representative image of a device for on-demand, longitudinal, and multi-parametric monitoring of sweat using a set of condition-specific assays embedded in each sensingchamber. (B) Close-up image showing activation of the pump by pulling the tab. Scale bar: 5 mm.13Students:Navya Mishra, Nate T. Garland, Krystyn A. Hewett, Mohammad ShamsiFunding Source:ASSIST Center
Key Accomplishments:Our team has made advances in two notable areas: thewearable skin vapor watch system and a monolithic arrayof MOx sensors. Our watch system is a battery-powered, wearable monitoring system to measure the VOCsemanating from human skin. The portable system consistsof gas sensors, Wi-Fi and Bluetooth enabled MCU,temperature, and humidity sensors, all on a 33mm x 30mmPCB board. The developed system includes five sensors todetect and quantify the VOCs from skin as well as tomeasure ambient VOC levels. Principal componentanalysis shows a clear classification among fasting, non-fasting, and alcohol intake when the sensors areoperated at different temperatures. We have alsosuccessfully fabricated a monolithic 4x4 array of ALD MOxsensors that can not only be heated to differenttemperatures but also produce different metal oxidesurfaces through on-chip annealing. We have used thisarray to measure NO , CO and ozone at very low levels.We have also evaluated matrix effects arising fromconnected sensors operating at different temperaturesbased on their baseline resistance changes. Ultra-Low Power Metal OxideElectronic Nose Arrays for Skin,Breath, and Environmental MonitoringWearable skin vapor monitoring system and dashboardshowing multiple gas, temperature, and humidity sensorsPrincipal Investigators:Dr. Veena Misra, Dr. Bongmook LeeObjective:The human body emits volatiles from breath, skin, saliva,urine, blood, sweat, and feces that contain insight intothe health state of the individual such as metabolicactivity, respiratory health, infectious diseases andchronic conditions including cancers. However, thebreath/skin mixtures are complex and contain numerousvolatiles often present in very low concentrations.Furthermore, human emissions change over time,requiring continuous monitoring. The objective of thisproject is to build a dynamic association between volatileorganic compounds (VOCs) and physiological/psychological health, in a continuous, wearable, andnon-invasive manner.Approach:Our approach utilizes multiple metal-oxide sensors madeusing a novel monolithic process based on CMOSprocessing, MEMS, and atomic layer deposition (ALD)that comprises the e-nose platform technology. Our largearray of sensors provides additional specificity to the e-nose analysis. This system will enable passive andcontinuous monitoring of skin vapor over time. The e-nosearray has ultra-low power operation (<1mW) and,through machine learning, can differentiate between avariety of gasses and correlate these to air quality as wellas the metabolic state. Impact:Our wearable and continuous skin vapor sensing systemcan facilitate tracking of the dynamic VOC response toan individual's lifestyle. Since these wearables are alsomonitoring health vitals, this allows a direct correlationbetween health and the environment.Measurements of sensor resistance (left) and corresponding principlalcomponent analysis (right) showing skin vapor classification betweenfasting, non-fasting, and alcohol intake states 14Funding Sources:ASSIST Center NSFStudents and Postdocs: Yilu Zhou, Dr. Farzad Mohaddes, Smriti Rao2
Ultra Low PowerPhotoPlethysmoGraphy (PPG)Module for BiomedicalApplicationsPrincipal Investigators:Dr. Alper Bozkurt and Dr. James DieffenderferObjective:This project focuses on developing ultra-low power andnovel biophotonic techniques for wearable physiologicalsensing that can be incorporated with other sensors forcorrelated sensing experiments. The state-of-the-artcircuit techniques reduce power consumption usingtechniques like logarithmic amplifier, heartbeat lockedloop, etc. However, most of these systems wereevaluated on a benchtop or on the finger. This projectdemonstrated that compressive sensing is one of thelowest power consuming techniques and evaluated itsperformance on the wrist.Approach:This project enabled a compressive sensing based application-specific integrated circuit (ASIC) forphotoplethysmography (PPG). We worked onminiaturization and integration of this novel compressivesensing based ultra-low power PPG ASIC into ASSIST’shealth and exposure tracker (HET) testbed in the formfactor of a wearable wristband and evaluated its usabilityto track heart rate on the wrist.Key Accomplishments:The system miniaturization for a wearable form-factor wasachieved without compromising the performance of theASIC. The ASIC consumes 172 μW of power to extractheart rate from the sparse PPG signal where the wholesystem consumes 1.66 mW of power for continuousstreaming of heart rate data over the commercial off theshelf Bluetooth Low Energy radio of the HET engineeredsystem. We were able to demonstrate a wrist-worn systemas an efficient platform for future evaluation of thecompressive- sampling based PPG technique through thein vivo clinical studies under the HET Testbed environment.Impact:Photonic measurements, such as PPG and pulse oximeter,are the most common methods in wearable systems totrack physiology. On the other hand, these are also someof the most power consuming modalities due to thenecessity of generating a large number of photons.Although most of this is lost due to absorption andscattering in the tissue, important hemodynamicparameters are assessed in return. An ultra-low powerASIC for PPG is required for overcoming translationalbarriers related to use of these systems as a part of self-powered or extended battery life based operation.PPG block diagram (top left).ASIC (top middle) andassembled board (top right),wrist-worn system (bottom).15Postdoc:Dr. Parvez AhmmedFunding Source:ASSIST Center
Zero-Power Wearable Sweat Assaysand Long-Term Sensing DeviceInterfaces Using Osmotic Pumpingand Paper MicrofluidicsPaper-hydrogel sweat collection platform, its principles of operation, and its integration with low-power lactate sensor and wireless data transmitter Student:Sneha MukherjeeObjective:Our team has pioneered a unique sweat extractiontechnique that can operate over periods of multiple daysby a novel, non-invasive osmotic-capillary method thatdoes not require any electrical power. These sweatcollection patches can interface with on-device sensorsor benchtop assays. Approach:Our technique gently extracts sweat from the skin using ahydrogel patch infused with benign solute. The solutecreates an osmotic pressure gradient that pulls sweatfrom the sweat glands (sweat glands naturally expelsweat by osmotic principles; thus, the technique isbiomimetic). The collected sweat is then transportedtoward sensors by simple and reliable paper-basedmicrofluidics, which uses wicking and capillarity totransport fluids without the need for external electricalpower. Engineered evaporation pads placed at the endof the paper strips continually drive the transport of fluidsfor days; in contrast, state of the art paper-based assaysare typically single-use devices and have a shortoperating time. Impact:The osmotic-capillary principles that we have pioneeredto interface the skin could form the basis of a stunningtechnology breakthrough in the field, as they arebiomimetic, non-invasive, do not require any electricalpower, or any active sweating. The ability to continuouslyharvest sweat in a non-invasive and non-irritating mannerfor long-durations enables performing a variety ofbioassays in a non-invasive, user-friendly manner, andwithout an external power source.Key Accomplishments:We focus, as the first major outcome of the project, onthe development of simple, inexpensive, non-invasive skinpatches for testing of sweat for cortisol levels, interleukins,other stress biomarkers, ionic balance, and toxins. Thisplatform can be used in integrated wearable systems forlong-term sweat and interstitial fluid [ISF] sampling andanalysis of hormones, glucose and other ISF biomolecules.We have published our latest findings in onintegrated wireless devices.ACS Sensors16Funding Source:ASSIST CenterPrincipal Investigators:Dr. Orlin D. Velev and Dr. Michael D. Dickey, Dr. Alper Bozkurt, Dr. Michael Daniele.
Block diagram of the ASICObjective:Monitoring of animal health is crucial for variousapplications from agriculture to working animalapplications, and from pharmaceutical industries tosports. In this project, our major aim is the development ofultra-miniature, injectable and subcutaneous physiologycapsules. Approach:Since existing capsule devices suffer from limitations onpower budget and device size, we are designing andtaping out an application-specific integrated circuit(ASIC) to enable measurements of heart rate, respiratoryrate, and core body temperature using power and areaefficient circuit designs. These capsule devices will beinserted into the animal body via needles typically usedfor conventional RFID microchip implants for identifyinglost pets or indexing farm animals. We will develop awirelessly-powered animal homecage to eliminatetethers, batteries, and limitations on the duration of in vivostudies for small, freely moving animals.Key Accomplishments:We have successfully taped-out our first ASIC chip andtested its functionality. In PPG measurement, the LEDdriver delivered current pulses to the LED, with currentadjustable within 1 mA–15 mA, pulse width adjustablewithin 50 µs–150 µs, and frequency adjustable within 50Hz-–200 Hz, while the PPG sensing frontend (SFE)converted the photodiode current to a voltage signal. InECG measurement, the ECG SFE amplified and filteredthe input signals with gain adjustable within 45dB - 80dB.In the body temperature measurement, the temperatureSFE showed a linear output voltage as a function oftemperature within 20°C-45°C. The outputs of the ECG,PPG, and body temperature SFEs were digitized by ADC.The successfully reconstructed ECG, PPG, and bodytemperature signals indicate the functionality of the ADCand data packetizer. Now, our team is designing thesecond ASIC. In addition to the functions realized in thefirst ASIC, the second ASIC will include an on-chipmemory controller that interfaces directly with an externalmemory chip (e.g., Winbond W25Q80EW) to remotelystore ADC samples.Impact:Successful completion of this project will result in energy-efficient circuit designs, such as AC-DC converters, DC-DCconverters, LED driver and readout circuits for PPGsensing, ECG readout circuits, and body temperaturereadout circuits. These energy-efficient circuit IPs can bereused for self-powered or ultra-low-power wearables andIoTs. Energy-Efficient Circuit Design forUltra-Miniature, Injectable andSubcutaneous Physiology CapsulesPrincipal Investigators:Dr. Yaoyao Jia, Dr. Alper BozkurtPrototype of the wirelessly-powered animal homecageMicrograph of the ASIC18Funding Source:ASSIST CenterStudents/Postdocs: Raymond Stephany, Yiming Han, Linran Zhao, and Parvez Ahmmed
Self-Powered BluetoothBackscatter EKG Sensor SystemObjective:Electrocardiograms (EKGs) are commonly performedusing up to 12 long leads attaching a patient to a largemachine. These tethered measurements are problematicas the long leads are a source of failure, do not allow thepatient freedom of motion, and they commonly gettangled with themselves and other sensor wires causingincorrect readings and disconnections. Mostproblematic is the time it takes in emergency situationsto remove the EKG tethers to move a patient or to re-connect them when time is critical and seconds canmake a difference. Current wireless approaches thatremove the need for tethers suffer from large powerrequirements and depend on large and heavy batteries,creating bulky and uncomfortable devices.Approach:Our proposed solution is a wireless EKG system thatutilizes backscatter communication – communication viareflections – in the form of Bluetooth packets. Byleveraging backscatter communication, the powerconsumed by the EKG sensor is driven extremely low (~10uW), allowing the sensor to be self-powered by humansweat. This approach doesn’t require a battery and itsfootprint can be made extremely small, on the order of apostage stamp.Key Accomplishments: We have fabricated and tested prototypes of the body-worn EKG sensor, the RF source board, and a customphone app that displays the recorded EKG. We havealso tested this prototype with a sweat-powered“battery” with no drop in performance. We are currentlyoptimizing the design of the EKG sensor board on aflexible substrate and integrating it with EKG electrodesand a sweat-powered “battery.”Impact:The test results of the wireless Bluetooth backscatter EKGsensor system show its potential to revolutionize EKGmeasurement methods in a hospital setting and as aplatform for next generation body-area sensing networksand the future of wireless health monitoring.Typically, backscatter systems suffer from overhead andinfrastructure problems as they require RF sources tocover the desired area of communication, but thisapproach solves this conventional issue with aninnovative and small RF source that plugs into thephone/tablet/PC that is displaying the EKG information.This approach has the potential to create multi-sensorbody-area networks with low-cost and infinite shelf-lifesensors.Self-powered Bluetooth backscatter EKG sensor systemdiagram showing body-worn EKG sensor block diagram andplug-in RF sourceBluetooth backscatter EKG sensor board prototype on the leftwith major areas highlighted, and plug-in RF source connectedto a phone showing EKG recording app on the right.Principal Investigators:Dr. Jordan Besnoff, Dr. David Ricketts, Dr. Amay Bandodkar19Funding Source:Chancellor's Innovation Fund (CIF) 2022 - 2023
Demonstration of SoC with analog front end chip and PPG power as low as 9.35 μW (left)and close-up view of SoC and AFE chips (right) Objective:This project seeks to develop a core electronics platformfor integrating the technologies from other ASSIST thrustsinto a unified self-powered sensing system with a totalpower sustainable by energy harvesting, targeting abudget less than 50 uW, and having flexible multimodalcapabilities for the second generation self-poweredadaptive platform (SAP 2.0). Approach:The multi-chip platform, centered around a system onchip (SoC), includes circuits for data collection, datastorage, data processing, node control, powermanagement, power harvesting, power delivery, andwireless communication. This project plays a vital role inthe strategic plan of the center, since SAP 2.0 cannotoperate from body-harvested energy without the lowpower electronics from this project. This project is the hubaround which many of the other projects across differentthrusts are arranged, and it is driving the core capabilityfor self-powered operation.Demonstrated SoC (with RISC-V MCU, 8kB memory,boot ROM, on-chip clocks) integrated with flexibleanalog front end chip. Demonstrated a new flexible analog front end (AFE)with photoplethysmography (PPG) power as low as9.35 μW.Built printed circuit boards for integrated wearabledesign using custom ASSIST chips and components.Python system model for duty-cycled, hierarchical self-powered systems.Built a 300uW Bluetooth receiver that meets all specsof a BT receiverKey Accomplishments:Impact:This work is enabling the ASSIST SAP 2.0 as well as futuregenerations of the SAP systems to operate entirely fromharvested energy, due to the low power operation andflexible functionality of the custom chips.SAP 2.0 multi-chip system block diagramUltra Low Power SoC and RadiosPrincipal Investigators:Dr. Ben Calhoun, Dr. David Wentzloff20Students/Postdocs:Xinjian Liu, Sumanth Kamineni, Shourya Gupta, Natalie Ownby, Katy Flynn, Peter Le, Suprio Bhattacharya, Omar Faruqe, DaeHyun Lee,Akiyoshi Tanaka, Anjali Agrawal, Nugaira Mim, Omar Abdelatty, Abdullah Alghaihab, Yaswanth K. CheriviralaFunding Sources:ASSIST Center
Ultrasonic Energy/Data Transfer forImplantablesPrototype ultrasound power receiver including a custom ICand a pre-charged CMUT.Principal Investigators:Dr. Ömer Oralkan, Dr. F. Yalcin Yamaner, Dr. Yaoyao Jia2Pre-charged CMUTs fabricated ona 4-inch borosilicate glass wafer.21Students:Muhammetgeldi Annayev, Linran ZhouFunding Sources:ASSIST Center (C2C Program)NSFObjective:Power management in implantable devices is a criticalrequirement for successful adoption and impact.Ultrasonic power transfer provides substantially higherpower density and reaches much deeper in tissuecompared to alternative sources using inductive couplingor radio-frequency (RF). A key advantage of ultrasonicenergy transfer over the competing RF technology is thatthe maximum allowed power level in tissue for diagnosticultrasound is 7.2 mW/mm , which is about 70x highercompared to RF energy limits. Furthermore, attenuationof ultrasonic signals in tissue is far less than RF, and thewavelength of the ultrasonic energy in tissue is on theorder of millimeters. These advantages translate to asmall device size and excellent range in biologicalsystems. In this project, we aim to develop a miniatureultrasonically powered device integrated into an endo-vascular aneurysm repair (EVAR) stent-graft that couldprovide on-demand diagnostic information about thepresence of endoleak (a condition leading to pressurebuildup in the aneurysm sac), based on measurements ofthe aneurysm sac dimensions, and of the stent-graftinside the vessel lumen.Approach:Our approach to implement the described implantabledevice relies on using a capacitive micromachinedultrasonic transducer (CMUT) with integrated electroniccircuits to function as an ultrasonic power receiver, adistance measurement sensor, and a transmitter forwireless data transfer to an external unit.Impact:The results accomplished to date show the potential ofCMUT-based powering, sensing, and wirelesscommunication for implantables in a broad range ofapplications ranging from cardiovascular health to neuralsensing and stimulation. Key Accomplishments:To date, our team has demonstrated the feasibility of thepresented approach which includes key concepts for animplantable intravascular ultrasound device tomonitor/diagnose endoleak in endovascular aneurysmrepair stent-grafts. In early benchtop studies usingexternally biased CMUTs and off-the-shelf discretecomponents, we demonstrated greater than 1 mWpower recovery from a 3-mm device with incidentultrasound intensity of 5 mW/mm , which is less than thespatial-peak temporal-average ultrasound intensity(ISPTA) limit of 7.2 mW/mm set for diagnostic devices.Ultrasonic biphasic communication concepts withpotential for high data rate and pulse-echo ranging fromsensor to EVAR structures have also been shown. Mostrecently we have demonstrated a custom integratedcircuit (IC) that interfaces with a pre-charged CMUTdevice for ultrasonic energy harvesting. We implementedan adaptive high voltage charge pump (HVCP) in theproposed IC, which features low power, overvoltagestress (OVS) robustness, and a wide output range. Theultrasonic energy harvesting IC is fabricated in the 180-nmHV BCD process and occupies a 2 × 2.5 mm silicon area.We have also demonstrated a novel device structure toimplement pre-charged CMUTs.2222
Poly(Octamethylene Maleate(Anhydride) Citrate) (POMaC) CircuitsPrincipal Investigator:Dr. Michael DanieleA POMaC circuit composed ofresistor and LED operating in PBS.Scale bar is 9 mm.A. Stress strain curves of POMaC polymer achievable withchanging processing conditions. B. Resistance under strain forencapsulated POMaC circuit over 1000 cycles. Image showsthe circuit tested.Example of photopatterning mechanism and resulting proof-of-concept pattern produced. Scale bar is 500 μm.22Funding Sources:ASSIST Center (Center-to-Centre Grant)Student:Brendan Turner Objective:This work seeks to develop an elastic, bioresorbablesubstrate capable of supporting stretchable electronicsalong with cell payloads with the goal of providing bothmonitoring and therapeutic capabilities. There is a needfor development of implantable medical devices (IMDs)capable of integration with host tissue. Biointegration ofdevices with tissue has been limited by non-stretchable,non-degradable materials utilizing a non-interactivematerial design paradigm. These non-interactivematerials require removal surgery at end of implant lifeand typically result in additional local inflammation dueto non-matching physical and biological properties.Impact:Successful development of the POMaC circuit system willlead to implantable devices capable of providing boththe therapeutic and diagnostic functions offered bycurrent IMDs along with specific biotherapeuticpayloads (cells or biomolecules). For example, a devicecapable of targeted myocardial regeneration whileproviding transient pacing along with diagnostics tomonitor healing would be an exciting application.Further, soft, degradable and stretchable electronicswould be desirable in fields like wearables, soft robotics,and green technology.Approach:We have developed a citric acid-based elastomer as asubstrate for flexible and stretchable electronics. Thepolymer is made up of human metabolites and can bedegraded in vitro into its substituent monmers. Thematerial has a history of usage as cell scaffolds inregenerative medicine and can be tuned to matchphysical properties of various native tissues. Devicesfabricated using this technology could be used fortransient implants, single-use degradable biosensors,tissue regeneration, and combinations thereof. As part ofthis work, we plan to fabricate a degradable, wirelesspower module that could be used to power a variety ofsystems.Key Accomplishments:The mechanical properties of POMaC films can betuned with moduli ranging from 0.4-2.3 MPa (matching awide range of human tissue). A fabrication methodologyfor POMaC circuits has been developed and used toproduce circuits capable of operating in simulatedbiofluid. POMaC circuits have been characterized understrain showing R/R ~2.3 after 1000 strain cycles to 20%strain. Proof-of-concept photopatterning has beenshown and illustrates the potential for biomolecularpatterning of the material to enhance cell applications.0
Key Accomplishments:Relatively stable resistance values were obtained under100% uniaxial tensile strain in x- and y- direction (ΔR/R-within 0.1%) and 50% biaxial tensile strain (ΔR/R within0.15%). The heater was able to achieve a hightemperature of ~140 ºC with a low current of 0.125A andfast heating and cooling rate of ~16.5 and ~14.1 °C/s,respectively. Additionally, the heater showed outstandingrepeatability over 400 heating cycles. A feedback controlsystem with fast response (0.13s) was developed tocontrol the temperature on the heater by using theheater itself as a thermometer.00A Biaxially Stretchable and Self-Sensing Textile Heater Using SilverNanowire CompositeBiaxial heater when undeformed and deformed, and heating performance of the heaterPrincipal Investigator:Yong ZhuObjective:Thermotherapy is the use of heat in therapy, commonlyused for rehabilitation purposes. The therapeutic effectsof heat include increasing extensibility of collagen tissues,decreasing joint stiffness, reducing pain, and increasingblood flow, among others. Hence, the development offlexible and stretchable heaters is in great need. Silvernanowire (AgNW) is a promising conductive material forflexible and stretchable electrodes. However, most of theheaters are typically uniaxially stretchable (i.e.,stretchable in only one direction) and/or not gaspermeable. More importantly, their resistance usuallychanges with the applied strain, vitiating the heatingperformance. Approach:We fabricated a novel biaxially stretchable, yet strain-insensitive and self-sensing heater based on a compositematerial comprised of an AgNW network and polyimide(PI) matrix. With the designed Kirigami pattern, local strainwas effectively minimized due to the out-of-planedeformation when a large tensile strain was applied. Impact:This work can be of promising potential for wearableapplications. For example, thermotherapy at thecurvilinear surface of the knee using the stretchableheater. Also, the wearable heater can be used in otherparts of the human body especially those with largebiaxial deformations (shoulders, ankles).24Funding Source:NSFStudents/Postdocs:Shuang Wu, Zheng Cui, G. Langston Baker, Siddarth Mahendran, Ziyang Xie
Objective:Electrocardiography (ECG) on the upper left arm can bea useful technique to assess cardiovascular health whilestill maximizing user comfort when compared to the goldstandard of chest ECG. However, maintaining goodsignal quality of arm ECG in the presence of motionartifacts is still a major barrier. In this work, we havefocused on developing an ECG armband that is resilientto motion artifacts while providing user comfort. Approach:We developed our ECG armband using smart textilesintegrated with compression properties, printed dryelectrodes, miniaturized hardware, data storage, andwireless communication. We evaluated differentelectrode configurations by conducting ECGmeasurements at both static and motive states and usedimproved algorithms to quantify data quality and assessthe agreement between the proposed new techniqueand the gold standard. The optimal electrode positionwas determined by balancing wearable suitability andsignal quality. Key Accomplishments:We conducted measurements with three differentconfigurations under static, motion, and muscle activities,respectively. We determined an optimal configuration forthe highest peak-to-peak amplitude and the highest SNRunder all test conditions. Bland-Altman plots showed thatour selected position on the armband has a closeagreement with the gold standard. The compressionapplied by the elastic armband enabled intimatecontact between the dry electrodes and skin, reducingthe contact impedance that typically reduces dryelectrode signal quality. Our embedded long and slimrectangular electrodes in the armband with an area of9.6 cm compared to 7.07 cm of circular electrodes fromour prior work not only reduced skin-electrodeimpedance but also minimized the impact of electrodeposition shift that occur due to anatomical differencesacross individuals. We have shown that the ECG armbandis resilient against noise from common physical activitiesand can enable continuous cardiac monitoring withoutinterrupting daily life activities. Further, the horizontallyaligned configuration of electrodes in the armbandreduces electrode spacing and makes the armbandslimmer, which improves the armband's appearance. Impact:This all-in-one ECG armband prototype suitable for routineuse and daily wear includes innovations that haveproduced a systemwide improvement. Its contactpressure is measured to get a better picture of intimacyand clothing comfort. The system provides real-time andnoise-resilient ECG data without interrupting daily life andcan be implemented in use cases that warrant continuousECG monitoring.A Wearable ElectrocardiographyArmband Resilient Against ArtifactsPrincipal Investigators:Dr. Veena Misra, Dr. Bongmook Lee, and Dr. Amanda Mills25Funding Sources:ASSIST CenterStudents/Postdocs:Yilu Zhou, Farzad Mohaddes, PhD, Smriti Rao, Courtney LeeComparison of armband and BIOPAC at rest (top), and running (bottom); plots on leftshow ECG signal and plots on right show Bland-Altman with mean reference.(a) hardware design, (b) armbanddesign, and (c) subject wearingthe dry electrode armband22
Curvilinear Soft Electronics byMicromolding of Metal Nanowiresin CapillariesPrincipal Investigator:Dr. Yong ZhuObjective:Two challenges exist in the printing of nanomaterials foradvanced applications – patterning of grid structureswith uniform thickness and direct printing on curvedsurfaces. In this work, we report scalable printing of silvernanowires for soft electronics using micromolding incapillary (MIMIC) to address these two challenges andexplore the application of printed AgNWs. Approach:A polydimethylsiloxane (PDMS) microfluidic device is usedas a mold and AgNW ink is dropped at the inlet of amicrochannel. Capillary pressure arising at the liquid-airinterface in the microchannel pulls the ink to fill themicrochannel. After drying and removing the mold, thepattern is successfully printed on the substrate.Compared to other printing techniques, MIMIC printing ofAgNWs has the following advantages: (1) The ink isformulated by AgNWs and solvent only without polymerbinders, leading to high electric conductivity; (2) Thismethod can print grid structures with uniform thickness;(3) Direct printing on curved surfaces can be achieved.Key Accomplishments:In this work, we demonstrated the capabilities of theMIMIC method for scalable printing of AgNWs on a varietyof substrates for soft electronics applications. Complexpatterns can be printed on non-flat surfaces includingcylinder, hemisphere, saddle, step, corrugated surface,etc. Taking advantage of these features, deformablehybrid transparent conductive electrodes (TCEs),distributed pressure sensors on a glove, and smart contactlenses with an intraocular pressure sensor are directlyprinted. The printed hybrid TCEs can achieve sheetresistance as low as ~2.5 Ω/sq with an opticaltransmittance of ~82%, and the intraocular pressure of anartificial eyeball can be measured by wirelesslymonitoring the impedance of the pressure sensor printedon the smart contact lens.Impact:This work reports direct printing of complex and highlyconductive patterns on soft curvilinear and unevensubstrates with high resolution and uniformity. The methodsignificantly expands the manufacturing capability ofprinting nanomaterials on non-flat surfaces. The fabricatedsoft electronics show promising potential as next-generation 3D electronics. 26Students:Yuxuan Liu, Michael ZhengFunding Source:NSFDemonstration of scalable printing ofAgNWs on various substrates and non-flatsurfaces for soft electronics applications.
Sports-bra design (left, center) highlighting the electronics placement and non-intrusive electronics housing designand slim armband (right) electrode configuration with curved housing for a low profile design.Design Factors for HighPerformanceElectrocardiography GarmentsPrincipal Investigators:Dr. Amanda Mills, Dr. Adam Curry, Dr. James Dieffenderfer, Dr. Veena MisraObjective:The objective of this research is to improve the comfortand data quality of garment-based electrocardiography(ECG) monitoring. This includes exploring the inevitabletradeoffs associated with garment design and e-textileperformance. For example, one area of research is in slimform factors for electronic housings that will not interferewith integrated sensor performance.Approach:The arrangement of electrodes and their connectionmethod to a data aggregator are the key variables toconsider when trying to improve the signal to noise ratio(SNR) for an ECG garment. Our team uses screen-printed,Ag/AgCl dry electrodes that can be added to garmentsduring or after production which provides the opportunityfor modular and customized design approaches. Key Accomplishments:Our team has designed and fabricated two garments forECG monitoring: a sports bra and slim armband. Indeveloping the bra, we investigated the design andmaterials used in interconnecting the dry electrodes tothe electronic housing location. Copper foil cut in asinusoid pattern demonstrated the lowest interconnectresistance without adding any bulk to the garment. Thearmband design looked to reduce motion artifacts whilemaintaining a slim profile. The in-line electrodearrangement reduced the armband footprint andshowed the highest SNR of the electrode configurationstested. Impact:Biometric data quality of an e-textile smart garment isheavily reliant on the garment design and sensorintegration strategy and can still vary from user to user. Thisresearch explores and tunes the key factors in achieving ahigh performing smart garment for ECG monitoring. Theoutcomes of which can be applied to other on-bodysensing methods. 27Students:Yilu Zhou, Courtney Lee Funding Sources:ASSIST CenterNSF
Novel Tactile Sensing WearableDevices from Hierarchical 3DPrinted HydrogelsPrinciple, material composition, and preliminary data on the new HHG sensors, demonstrating good sensitivity and low creep.Principal Investigators:Dr. Lilian Hsiao and Dr. Orlin D. Velev28Objective:This project will introduce a new generation ofmechanically sensitive and sustainable tactile sensors forwearable devices which will allow haptic feedback,virtual key input, as well as prosthetic feedback andcontrol. They will be based on a new class of touch andhaptic interaction sensing wearable patches, which willbe completely made of benign and flexible hydrogels.Approach:Emerging technologies for touch-based skin-mountedsensors require the development of wearable sensingpatches that are inexpensive and based on sustainablesources, thus reducing the problem of e-waste. Thisproject is based on a transformative approach and set oftechniques for making broadband piezoresistive andpiezo-ionic patches. The new patch-like sensors will use anumber of breakthrough inventions and scientificadvances made by the dynamic collaborative team ofthe PIs Hsiao and Velev. These innovations include thesynthesis of new fibrillar networks from biopolymers,forming hierarchical piezoresistive hydrogel networks thatgenerate changes in electrical resistance through thereversible breakage and adhesion of nanofibrils within a3D network. Impact:The making of soft sensing patches by using safe,printable, and sustainable hydrogel materials represents asignificant breakthrough in the field of haptic wearablesensors. This advancement will enable us to develop aclass of hydrogel haptic wearables that could potentiallydetect changes in tactile forces (~ 0.1 N) down to thenanoscale (~ 10 nm).Key Accomplishments:The use of hydrogel medium poses several challengessuch as mechanical stability, fracture, creep, and drying.Our team has thus far successfully increased the overallFunding Source:NSFsensitivity and resiliency of biocompatible hydrogels andpublished the results in Nature Communications Wesolved many of these problems by introducing a newclass of printable touch sensors made completely out ofhydrogels with hierarchical network microstructure. Onebreakthrough in this project is the formulation of newclasses of hydrogels with double network – colloidal andmolecular, which are synergistically reinforced whilemade of the same material. These “homocomposite”hydrogels (HHGs) serve as an excellent base material for3D-printable piezo-conductive sensors operating onchanges in ionic conductance. The HHGs sensors couldbe easily 3D printed, had higher ionic conductance,much improved mechanical stability, and much smallermechanical and electrical creep. The hierarchicalmicrostructure of the ionic hydrogels provides superiortactile sensitivity as compared to conventional hydrogel.Our team will seek to prototype and demonstrate, incollaboration with other ASSIST investigators, wirelessinterfacing/readout of the sensor patches and use ofmachine learning methods for multi-touch dataprocessing.Nature Communications.
Novel Textile-Based Sensors forInner Prosthetic Socket EnvironmentMonitoringObjective:This project aims to develop a novel Flexible InneR-socketSensing Technology (FIRST) that is seamless, unobtrusive,and elegantly integrated into the lower-limb prosthesissocket. FIRST is based on an electronic-fabric structurewhere the fibers of the fabric act as sensory elementsand can simultaneously track tactile forces,moisture/wetness, electromyography and bodytemperature at multiple sensing points around theresidual limb. The major challenge is to develop afundamental understanding of the coupling andinteraction between multi-component fiber cross-sectional architecture, fabric structure, and its electro-mechanical response to achieve a multimodal sensorthat can be unobtrusively integrated into 'textile-based'sensory devices in general. The interpretation of the datais to identify locations of skin problems to enable patient-self management and allow for more objective clinicalevaluation to avoid the occurrence of potential skinbreakdown and the resulting complications.Approach:Our collaborative research team works on melt-extrudedmulti-component fiber and seamline based sensordevelopment where we carefully engineer the fibercross-section, fabric structure, and its electrical response.This targets a sensitive and specific multimodal responseusing microfabricated and, ultimately, textile-basedpolymeric fibers with ordered segments of conductingand insulating areas in the fiber cross-sectional structure.We aim to unobtrusively integrate these into manyelectronic small- or large-area textile-based sensorydevices and systems of the future especially for healthmonitoring. Key Accomplishments:We manufactured arrays of multi-component fiber andseam-line based sensors connected to a wireless high-speed data recording and transmission system via textileinterconnects. We tested the sensors on an in-vitroartificial limb testing setup and two in-vivo experimentsinvolving an able-bodied subject donning a bent-kneeadapter and a bilateral transtibial amputee participant.In all these cases, the sensor array successfully detectedpressure changes within the inner- socket during weight-shifting and walking experiments.Impact:Amputation is one of the major causes of disability.Sockets are the important prosthesis components andphysical interface to integrate the prosthetic limbsmechanically with the amputee's residual limb to replacelost function. Objective monitoring of the inner socketenvironment (i.e. pressure, temperature, and humidity)and residual muscle activity during daily prosthesis userequires flexible, unobtrusive and multi-modal sensors thatcan be integrated into the socket structure withoutcausing subject discomfort. The lack of such an inner-socket sensor technology has been a long-standingproblem for evaluating the prosthesis socket, preventingthe complications elicited by poor socket design and fit,and advancing the socket technologies. Therefore,advanced socket technologies are urgently needed andwill be developed under this project to significantlyreduce the number of clinic visits, lower the healthcarecosts for amputees, and ultimately improve their quality oflife.Schematic depictions of the fiberand seam-line sensor arrays,images of the integrated textilesensors, and examples ofintegration and testing with humansubjects, demonstrating successfuldetection of pressure changes.Principal Investigators:Dr. Alper Bozkurt, Dr. Tushar Ghosh, Dr. Helen Huang29Student:Brendan ThompsonFunding sources:NSF
Textile Based Soft Actuators forCompliant and WearableArtificial MusclesUltrathin pneumatic fiber-shaped robotics and theirapplications in orthosis devices and tissue engineering.Principal Investigator:Dr. Xiaomeng FangObjective:Textile-based robotics are fibers, yarns, and fabrics thathave comparable properties to traditional textiles, allwhile possessing “smart” shape-changeable features. Thisproject aims to design fabrics using fiber-shaped roboticsto achieve targeting forces and motions. It will inevitablypromote the development of advanced wearableelectronics, biomedical devices and tactilecommunication systems. Until now, the development oftextile artificial muscles is still at the early stage. There is aknowledge and technology gap that urgently needs tobe filled, which is to understand the influence of fabricpatterns and selected actuator technology oncorresponding shape changeable fabric motions. Thisknowledge should serve as the foundation to unravel theactuation mechanisms, and as the basic theory to guidethe future development of novel devices to fulfill variousrequirements in real-world applications.Approach:Our lab developed ultrathin fiber-shaped robotics,including pneumatic-driven, electromagnetic-driven, andelectric-driven types. They are inherently compliant andcan be integrated into the woven and knitted fabric, withor without blending with regular textile yarns. Withdifferent fabric patterns and number of active fibers, wecan develop fabric robotics having target performance.Muscle-like fibers and fabrics would be of enormous andwidespread benefit for many emerging technologies,including microrobotics, medical implants, hapticdevices, and responsive prosthetics. Key Accomplishments:We have demonstrated the following features of ourapproach: our minimum fiber diameter is less than 1mm;the highest force generated by a single fiber is up to 50N;the shape-changing ratio is up to 20%; fabric robotics canbend, or form a roll upon actuation. Currently, thedeveloped fiber robotics have been used in poweringassistive knee orthosis to help users’ to bend the lower legto ~30 degrees.Impact:Our technology offers unique solutions for current urgentchallenges related to our life quality and well-being. Theadvanced devices developed in this project will be abreakthrough for the future development of advancedbiomedical, military, tactile communication and wearabledevices driven by soft, conformable, and controllabletextile artificial muscles. Fabric robotics using fiber-shaped actuators (left two) and therolling motion upon actuation (right).Textile based soft robotics and their applications.30Students:Sen Zhang, Muh Amdadul HoquesFunding Sources:NC State University
Ultrasoft Porous 3D Conductive DryElectrodes for ElectrophysiologicalSensing and Myoelectric ControlFabrication process of the 3D porous dry electrode and itsapplication in Electrophysiological Sensing and Myoelectric ControlPrincipal Investigator:Dr. Yong ZhuObjective:In personal healthcare, continuous monitoring ofbiopotential signals, such as the electrocardiogram(ECG), electromyogram (EMG), electroencephalogram(EEG), and electrooculogram (EOG), could greatlybenefit clinical physiological tracking and medicaltreatment. In these applications, the most widely usedelectrodes are pre-gelled and disposable. Severalconcerns have arisen with pre-gelled electrodes thathinder them from long-term use, including bulkiness, skinirritation, and poor robustness due to gel dehydration.Most electrodes are limited to 2D planar designs.However, 3D biopotential electrodes with interconnectedconductive networks are less explored.Approach:In this work, we present fabrication and applications ofultrasoft, 3D conductive, breathable dry electrodes forelectrophysiological sensing and neural–machineinterface (NMI). The high gas permeability and ultralowelastic modulus improve skin compatibility forapplications requiring long-term skin contact. Owing tothe 3D interconnected conductive network embeddedin a polymer matrix, the electrode remains robust againstdeformations and sweat during daily activities. Thefabricated electrodes can acquire high-fidelityelectrophysiological signals (i.e., ECG, EMG, EEG) in atruly unobtrusive manner. The EMG-driven controlinterface can be used for prosthesis control,neurorehabilitation, teleoperation, gaming, and virtualreality.Key Accomplishments:The developed porous dry electrodes addressed severalchallenges faced by the commercial electrodes for long-term wearable applications: 1) the electrodes are gel freeand breathable. They rely on porous nanocomposites asthe compliant electrical interface to overcome the skinirritation issue of the conductive gel. 2) The electrodes areultrasoft, stretchable, and compressible, which aresignificantly less bulky on the skin than the commercialelectrodes. 3) The electrodes are robust for long-term use,owing to the encapsulation of nanomaterials in thepolymer matrix.Impact:All the above features allow for the accurate andunobtrusive acquisition of ECG/EMG/EEG signals. Basedon the real-time tracking of EMG signals and thedeveloped musculoskeletal model for interpretinghand/wrist kinematics, an EMG-driven NMI for driving thevirtual hand/wrist can be of great potential in prosthesiscontrol.31Funding Sources:NSFNIHStudents:Shanshan Yao, Shuang Wu
Wet Spinning of AdvancedMaterials into Energy-StorageYarns for Smart TextilesFigure 1. Schematic demonstration of the fabrication processfor our YSCs.Principal Investigators:Dr. Wei Gao, Dr. Feng ZhaoObjective:The advancement of smart textiles requires textilecompatible power sources for sensing, computing, anddata transmission, etc. Energy storage devices that canbe efficiently and reliably integrated with other electroniccomponents, while maintaining the breathability andflexibility of the resulting fabrics, are favorable. Incomparison to their film counterparts, yarn-shapedenergy storage units can be incorporated into fabrics viaweaving or knitting in diverse sizes and shapes, and arethus readily deployable to different parts of clothes.However, a robust energy-storage yarn configuration thatcan withstand rigorous textile fabrication processes andvarious end uses, is yet to be established.Approach:To fabricate yarn-shaped supercapacitors (YSCs),conventional textile techniques, including wet spinning,yarn twisting/plying, sizing, etc., are employed toengineer active materials, current collectors, separator,and electrolyte, into a single yarn structure (Fig. 1). Aunique yarn twisting protocol is designed to induce betterinterfacial contact between each component in yarnelectrodes, improving the performance consistency andlinear power density of YSCs. Separator threads areinnovatively introduced in-between yarn electrodes,leading to reliable power units in woven or knit fabrics,bearing repetitive mechanical deformations alongalmost all directions.Impact:The fabrication process of our YSCs are highly scalable byutilizing traditional textile techniques. Their power output isadjustable and customizable through tuning the diameterand length of YSCs to meet various power requirements.The current yarn fabrication protocol is also applicable forlithium-ion battery yarns, thus broadening the scope ofreal-life applications. Figure 2. A photographic image of a woven fabric, where YSCsare incorporated to light up five LEDs.32Funding Source:US Army STTRStudents:Nanfei He, Junhua Song, Jinyun LiaoKey Accomplishments:Tens of meters long mechanically flexible and robust aswell as electrochemically reliable YSCs can beincorporated into fabrics to power electronics (Fig.2),bearing mechanical deformation along differentdirections. To meet various power requirements, YSCs canbe tailored to desired energy and power densities bytuning their diameters and lengths. High power density of46.5 mW/cm (18.7 mW/cm) can be achieved at anaverage energy density of 0.65 mWh/cm (0.26 mWh/cm)in our unique YSC design.33
AI-Driven, Resilient andAdaptive Monitoring of Sleep(AI-DReAMS)Objective:This project investigates the use of an artificial intelligencedriven, reconfigurable sleep monitoring system totransform sleep research in the clinic and at-home. Asensor fusion strategy backed by artificial intelligence toultra-miniaturize the sleep assessment instruments andexplore novel sleep-related biomarker features have thetransformative potential to invigorate sleep research formore efficient and accurate diagnosis and treatment ofsleep disorders. There is a need for combining lower costwith better ergonomic comfort, and more efficient dataanalysis to pave the way for rapid translation, adoption,and effective deployment of sleep technologies.Approach:This project integrates two parallel efforts combininginnovations in hardware and data analytics: 1) enablingan adaptable and reconfigurable embedded systemplatform in the form factor of an adhesive patch, and 2)developing state-of-the-art machine learning techniquesincorporating the data-driven models necessary forimproving sleep monitoring system resilience. Thehardware system fuses multimodal wearable sensors,combining near infrared spectroscopy with othertraditional sleep related signal sensors, in skinconformable substrates, to collect data on multiple bodylocations. The data analytics platform includes 1) signalprocessing to enable data-driven metrics for signalquality assessment for a given inference task, 2) inferencemodels based on transfer learning techniques anddiverse datasets for detection of sleep events anddisorders using new sensing modalities, and 3) BayesianNeural Network supported sensor selection for improvingthe resilience and adaptability of sleep sensor systems.Key Accomplishments:This recent project stems from an earlier clinical studyfunded by National Institute of Health to explore the useof near infrared spectroscopy and machine learning tobring a new perspective to sleep studies. The teamdemonstrated flexible devices to perform near infraredspectroscopy and electroencephalography in the formfactor of a flexible bandage. The current efforts focus onconstructing a reconfigurable version of the hardwareplatform to collect data at-home and in sleep clinicstudies and support the development of the proposedartificial intelligence techniques for studying sleep moreefficiently.Impact:A considerable amount of the population in the US andaround the world suffers from a chronic sleep disorder.However, the majority of these are not diagnosed ortreated. There is a vital need for new wearabletechnologies to increase the capacity of sleepresearchers to make further advances in investigatingsleep, understanding sleep pathologies, and to improvethe ability of clinicians to reliably detect and treat sleepdisorders. The research results from this award have thepotential to positively influence the continuous monitoringinstrumentation required for other chronic conditions suchas heart diseases. In addition to allowing a novel, artificialintelligence-driven and reconfigurable tool design forsleep research, this effort will also shed light into novelmultimodal biomarkers assessed noninvasively in wearableform factors for detection of sleep stages and disorders.34Principal Investigators:Dr. Alper Bozkurt, Dr. James Dieffenderfer, Dr. Michael Daniele, Dr. Edgar Lobaton, Dr. Vladimir PozdinStudents/Postdocs:Dr. Parvez Ahmmed, Kaila Peterson, Devon Martin, Alec BrewerFunding Source:NSFPhotonic sensor and circuit for sleep studies, in an adhesive bandage form factor with wireless recharging capabilities.
Bio-Electro-PhotonicMicrosystem Interfaces forSmall AnimalsKey Accomplishments:The capsule system has been evaluated in clinicalenvironments for tracking the physiological signals in ratsand chickens. The collar and harness system have beendeployed in the field with guide dog puppies forimproving puppy training program outcomes. Recentefforts focus on training dogs to follow or interact withunmanned-air-vehicles towards deploying this in workingdog applications such as search and rescue operationsand agricultural pest detection. Impact:The microsystem under development enablesphysiological monitoring of small animals in their naturalenvironment. This system would be impactful for thewelfare of farm, companion, working, and wildlife animalsin addition to providing new bi-directional channels tocommunicate with them.Objective: This project targets a wirelessly powered injectablecapsule capable of wirelessly monitoring biophotonicand bioelectrical physiological signals in small animals.This microsystem responds to the critical need for a novelminimally invasive class of devices for continuousrecording of key physiological parameters of animals innatural environments without disturbing natural behavior.Approach:This project develops two parallel physiological andbehavioral sensing platforms on two different formfactors: an injectable subcutaneous capsule and awearable harness system for animals. The capsuleplatform provides photoplethysmography, electrocardio-graphy, accelerometry, and thermometry performedunder the skin. This is used to calculate heart rate,respiration rate, oxygen saturation, pulse transit time andcore body temperature. The harness system is for moresimple wearable applications with electrocardiography,photoplethysmography, inertial and environmentalmeasurements integrated into a standard dog harnessand collar.35Principal Investigator:Dr. Alper BozkurtPostdocs and Students:Dr. Parvez Ahmmed, Dr. James Reynolds, Devon Martin, Colt Nichols, Maxwell NoonanFunding source:NSF
Representative data from on-body trials,showing distinct differences in measuredgait data between overloading andunderloading conditionsBody Area Network of InertialSensors for Clinical GaitRehabilitationPrototype modular WBAN peripheral(right) and miniaturized version 2device (left)Principal Investigators:Dr. Michael Daniele, Dr. Jason Franz, Dr. Brian Pietrosimone, Dr. Adam Kiefer Objective:Monitoring of gait and lower limb function is of greatinterest to clinicians overseeing patients recovering frommusculoskeletal injuries and surgeries, and those at risk oflower limb dysfunction due to advanced age or overallhealth status. Existing inertial sensors capable ofobtaining this information are prohibitively expensive,making widespread monitoring of these patientsimpractical. Our novel system of inertial sensors isintended to provide a low-cost option for monitoring ofpatient gait using a wireless body-area network (WBAN)of self-contained sensor peripherals, capable ofrecording movement and sending these data to acentral hub for analysis using machine learningtechniques.Approach:Our system consists of up to eight wearable devicesinterfaced wirelessly with an off-body control unit. Eachperipheral device contains an inertial measurement unit(accelerometer, gyroscope, and magnetometer) andsupporting circuitry needed to enable continuous on-body operation for up to 4 hours. Data from each sensorchannel is sent to the control unit in 44-byte packets at arate of ~50Hz/device, sufficient to capture detailed gaitinformation. Additional planned peripheral sensors aim tocapture other data streams, such as heel drop pressure,to improve monitoring of patient status.Key Accomplishments:At present, we have validated the use of theseperipherals for collecting gait data as part of a body-area network. Human trials involving deliberate over- orunderloading of participant vertical ground reactionforces have demonstrated our system’s ability to capturesubtle changes in gait with high resolution. Additionally,an add-on circuit for heel pressure sensing has beensuccessfully demonstrated and will be incorporated intoupcoming trials. Ongoing work will focus on integratingthe WBAN with machine learning to provide real-timefeedback, as well as continued miniaturization of theindividual sensors.Impact:Our WBAN system will enable clinicians to monitorpatients for minute changes in gait during rehabilitation,as well as providing prophylactic surveillance for harmfulchanges in gait within vulnerable patient populations(e.g., post-surgery). Importantly, the low cost of thissystem will facilitate detailed monitoring in real-worldsettings, ensuring a complete picture of the user’s lowerlimb function is obtained during the performance of dailytasks, rather than stereotyped movements in a labsetting.WBAN peripheral interfaced withprototype heel drop pressure sensor36Students/Postdocs:Jack Twiddy, Kaila Peterson, Grace Maddocks, Ryan MacPherson,Ricky Pimentel, Max Yates, Cortney Armitano-LagoFunding Sources:ASSIST Center, NSF,Eshelman School of Pharmacy Institute for Innovation
Cough Detection using Wearablesand Out-Of-DistributionRecognitionDetailed Pipeline of the ACDA with OOD DetectionPrincipal Investigators:Dr. Edgar Lobaton, Dr. Alper Bozkurt, Dr. Michelle Hernandez, Dr. James Dieffenderfer, Dr. Tahmid LatifObjective:The objective of this project is to develop an AutomaticCough Detection Algorithm (ACDA) for wearabledevices that meets clinical monitoring requirements byfusing multimodal sensor data. This ACDA should be ableto extract features from a wearable, process the data,and detect coughs in a smart device. Furthermore, wewish to ensure that privacy is maintained so no speech isrecognizable from the features while maintaining enoughdetails in the signal to detect and characterize differenttypes of coughs.Approach:Our ACDA is implemented using a convolutional neuralnetwork (CNN). This system is enhanced by incorporatingOut-Of-Distribution (OOD) Detection to recognize OODdata that is unfamiliar to the system (e.g., anenvironmental sound such as an engine sound that is notpart of the training of the system), which can be a sourceof confusion for Artificial Intelligence systems. A realizationof this solution is shown in the figure in which an acousticsignal is filtered on the embedded device and is used tocome up with cough detection in the smart device. Weare extending this cough detection system by developinga wearable system that combines multiple sensingmodalities (e.g., ECG, PPG, audio, and inertial). Bycombining data from a wearable, we would be able toseparate interfering sounds from individuals other thanthe main user. In addition, we would be able to correlateand fuse these measurements with the other modalities.Part of our cough detection efforts focus has also beenintegrated with the embedded hardware, in which weare trying to minimize the amount of data to betransmitted and processed for power consumption andprivacy concerns.Key Accomplishments:Our novel ACDA meets clinical monitoring requirements,was developed using publicly available data, reliablyoperates at a low sampling frequency, and maintainsuser privacy. Our CNN-based ACDA achieves a sensitivityof 92.7%, a specificity of 92.3%, and an accuracy of 92.5%using a sampling frequency of just 750 Hz. A low samplingfrequency allows us to preserve patients' privacy byobfuscating their speech. We have analyzed the trade-off between speech obfuscation for privacy and coughdetection accuracy and realized that the 750 Hzsampling rate is optimal. We further improve our CNN-based ACDA by incorporating OOD detection algorithms.We found that the new algorithm produces trustworthyresults when the sampling rate is greater than 750 Hz andthe window size is between 4 - 10 seconds. At 750 Hz,ACDA with OOD detection can keep accuracy higherthan 80% even when more than 50% of the input is OODdata, and its performance surpasses the ACDA withoutOOD detection when only 15% of the input is OOD data.Impact:Cough detection can serve as an important biomarker tomonitor chronic respiratory conditions. However, manualtechniques which require human expertise to countcoughs are both expensive and time-consuming. RecentACDAs have shown promise to meet clinical monitoringrequirements, but due to the required portability of sensingtechnologies and the extended duration of datarecording, only in recent years have they made their wayto non-clinical settings. More precisely, these ACDAsoperate at high sampling frequencies, which leads to highpower consumption and computing requirements, makingthem difficult to implement on a wearable device. Havingan ACDA capable of continuous monitoring whilemaintaining privacy would allow for new types ofdiagnosis and prevention of the spread of disease incommunities.37Students/Postdocs:Yuhan Chen, Pankaj Attri, Jeffrey BarahonaFunding Source:NSF
Enhanced Detection of Impending ProblemBehavior in People with Intellectual andDevelopmental Disabilities ThroughMultimodal Sensing and Machine LearningComparison of the heart rate detection between the BioPacBioNomadix and our physiological sensing wristwatch.Principal Investigators:Dr. Nilanjan Sarkar, Dr. James Dieffenderfer, Dr. Amy WeitlaufObjective:Severe problem behavior can put individuals withintellectual and developmental disabilities (IDD) at risk forharm and it is advantageous to mitigate this behavior byidentifying precursors. This prediction of problem behavioris currently done using human observation by acaretaker, and is prone to bias. This project is focused onusing wearable sensors and machine learning in order tocreate a more objective approach to predictingproblem behavior.Approach:This prediction system comprises a wearable sensornetwork, which feeds information via Bluetooth LowEnergy to a machine learning algorithm. The wearablesensor network consists of five posture sensing nodes(which include an accelerometer, magnetometer, andgyroscope) as well as one physiological sensingwristwatch (which measures heart rate, skin temperature,galvanic skin response, and steps).Key Accomplishments:We have completed the design of the hardware andsoftware for the wearable sensor network. We havebenchmarked the physiological sensing watch against agold standard (BioPac) and are currently in the process ofbenchmarking our physiological sensing nodes. We haveadditionally completed our initial experimentation ofcomfort comparison between different fabrics andwristband materials for our target population. This resultedin the selection of a nylon wristband and two types ofpolyester based textile garments (selected from a list of 20fabrics).Impact:If this work is successful, we will have created a systemthat is able to predict and allow for behavioralintervention. This could have potential uses in other areaswhere cognition and motion are needing to besimultaneously observed.Different components of the posture sensing node and physiological sensing wristwatch. Both these devices transmitdata wirelessly using Bluetooth Low Energy.Different styles and materials of wristwatch band evaluated forcomfort and usability. 38Funding Source:NSFStudents: Nibraas Khan, Abigale Plunk, Madeline Smith
Integration and Demonstration ofWearable Monitoring Systems forAsthma and DiabetesPrincipal Investigators:Dr. Alper Bozkurt, Dr. James Dieffenderfer, Dr. Michael Daniele, Dr. Edgar Lobaton, Dr. OrlinVelev, Dr. Vladimir Pozdin, Dr. Michelle Hernandez, Dr. Michael Dickey, Dr. Veena Misra39Funding source:ASSIST Center Students:Kaila Peterson, Yi Chen, Brendan Thompson, Maxwell Noonan, Devon MartinObjective:The health and environmental tracker (HET) system is aunique system that has a modular structure to test andevaluate various ASSIST technologies (from electrodesand optical devices to ultra-low power electronics andsensors) to drive the vision of ASSIST for correlated sensingof health and exposure. In addition to physiologicalsensors (electrocardiography, pulse oximetry, photo-plethysmography) and behavioral sensors (inertial meas-urement units), this system also supports environmentalsensors (ozone, VOC, ambient temperature, and relativehumidity), and novel biochemical sensors (lactate,glucose, and pH) for a more comprehensive andcorrelated analysis. No existing system can provide themulti-sensing capability of HET prototypes as well as anopen platform where the next generation sensingdevices can be integrated.Approach:The HET engineered system is composed of a modulararchitecture with various electrophysiological, bio-photonic, inertial, potentiostatic, amperometric, andenvironmental sensor front ends connected to a system-on-chip combining a microcontroller with a bluetooth lowenergy transceiver. This circuit architecture has beenpackaged in the form factors of a wristband, chest-patch, and a flexible patch that can be attached to anybody location. Using HET, sweat is collected through azero-power osmotic pumping scheme connected to ascreen printed enzymatic sensor. Collecting sweat andanalyzing the glucose and lactate concentrations help Key Accomplishments:HET systems have been tested in various clinicalexperiments related to asthma exacerbation prediction,sweat analysis, and wound sensing. This past year the HETprototypes were brought to a technology readiness levelof 5/6 while also lowering the power consumption to sub-milliwatt levels.Impact:The HET system strategically targets asthma as the medicalcondition of interest due to its high prevalence anddependence on environmental factors. HET brings theunique potential of continuously measuring local ozoneconcentration around the patient while also assessingheart rate, heart rate variability, respiratory rate, arterialoxygenation, and coughing frequency. This is combinedwith the next breakthrough in wearables with the analysisof biochemical markers such as sweat and wound fluidsanalysis for diabetic management. These capabilitiescould allow medical professionals to track asthmaexacerbations, diabetic metabolism and wound healingremotely and enable advanced data analytics togenerate automated feedback for the patient.assess metabolic state and support diet management fordiabetic patients to provide automated and actionablefeedback. The wound monitoring system monitors uricacid levels to ultimately track healing. The modularstructure of the HET engineered system enables it to bepowered by inductively rechargeable lithium batteries orenergy harvesting structures such as flexible solar cells orASSIST’s thermoelectric generators.
Multimodal Biosensing Systemfor Electrochemical &Biophotonic MonitoringMain ModulePrincipal Investigators:Dr. Michael Daniele, Dr. Alper BozkurtObjective:Non-invasive wearable technology has become readilyadopted by consumers, but has been limited tomeasuring physiological signals such as heart rate. Beingable to also measure biologically relevant analytes fromfluids such as sweat could increase the knowledge of anindividual's health and help inform related decisions.Therefore, our objective is to create a fully wearablemultimodal biosensor platform that simultaneouslymonitors both electrochemical and physiological signalsto provide a holistic picture. The current system measuressweat lactate, sweat pH, skin temperature, and tissueoxygenation.Approach:The main module of the system is a custom, wearablepotentiostat that interfaces with a flexibleelectrochemical sensor for monitoring sweat analytesand a biophotonics sensor for monitoringphotoplethysmography (PPG) and temperature, all ofwhich are transmitted over bluetooth to an app. Themain module is intended to be able to interface withother electrochemical sensors with minimal changes tomeasure a range of different analytes for various healthapplications. Moving forward, transitioning a version ofthis system from a wearable platform to a transcutaneousand potentially subcutaneous format will be researched.Key Accomplishments:The platform’s sensing ability was validated againststandard benchtop equipment in laboratory settings. Theplatform's ability to continuously and in real time monitorchanges in sweat lactate and pH during high intensityexercise in humans has also been demonstrated. Thissystem is also being interfaced with sensors that containosmotic pumps, hydrogels, and microneedles to extractinterstitial fluid during rest or low activity for analytemonitoring.Impact:This system allows for multimodal biosensor integration intoone wearable platform, allowing for a comprehensivephysiological picture of an individual. The tailorability ofthe system by interfacing the main module with differentsensors opens up the possibility for sensing very differentbiomarkers with the same hardware. The groundwork hasalso been laid in the current wearable format to beadapted into a transcutaneous and subcutaneousformat.Module with Electrochemical Sensor and Biophotonics Sensor40Funding Source:NSFStudents: Dr. Tanner Songkakul, Kaila Peterson, Hannah Nissan,Angelica Aroche, Misk Hussain, Grace Maddocks
ASSIST Startups / Licensees$50M invested through NSF,industry, and other fundingagencies90 inventions, 53 patents filed, 10 startups 105 Ph.D., 27 MS, and 96undergraduate degreesconferred658 publicationsH-index of 5040 capstone projects 117 REUs (Research Experiencefor Undergraduates)>500 K-12 participants in thewearable device competition120 Young Scholars (high schoolstudent researchers)117 RETs (Research Experiencesfor Teachers)ASSIST by the Numbers41
www.assistcenter.orgTo find out more about Industry Partnerships,contact us at email@example.com.