During Year 2, we successfully optimized all individual sensors with high stability and accuracy (including glucose, lactate, Na, K, pH, blood pulse wave, temperature, and galvanic skin response (GSR)) for continuous monitoring. With materials and chemistry innovation, we have made major progress to improve the long-term stability of our wearable chemical sensors (most of these sensors now have stable performance during multiple-day in vitro operation). We also integrated all the sensors into a multimodal flexible sensor system. A miniaturized iontophoresis module was also integrated into the system that allows stable localized sweat extraction across activities. The entire sensor patch can be prepared with inkjet printing at a large scale and low cost. This multimodal system was validated in human trials against gold standards. We simultaneously monitored the molecular analytes in human sweat including glucose, lactate, uric acid, sodium, potassium, pH, and key vital signs (i.e., skin temperature, pulse wave, GSR) using this wearable multimodal sensing platform. The substantial changes in all sensor responses were observed under physiological and psychological stressors. We also performed machine learning-based data analysis to classify the stress levels based on the data collected. Based on a combination of the physical/molecular data and machine learning model, we have obtained a comprehensive stress assessment system with high accuracy and robustness.

As COVID-19 became a major challenge for us over the past year, the SARS-CoV-2 RapidPlex, developed based on target-specific immunoassays built off laser-engraved graphene, could be used for rapid and remote assessment of COVID-19 biomarkers (i.e., nucleocapsid protein, anti-spike protein IgG and IgM, and C-reactive protein). The SARS-CoV-2 RapidPlex has the potential to quickly and effectively triage patients and track infection progression, allowing for the clear identification of individuals who are infectious, vulnerable, and/or immune. Modification of our platform design may allow for rapid viral antigen and antibody panel testing such that COVID-19 infection could be clearly distinguished from non-specific symptoms of seasonal respiratory infections such as influenza. Additionally, the wireless telemedicine diagnostic platform, when coupled with emerging wearable biosensors to continuously monitor vital signs and other chemical biomarkers, could provide comprehensive information on an individual's health status during the COVID-19 pandemic.

During Year 1, we have successfully developed most of the individual sensors with high performance (including glucose, lactate, Na, K, pH, blood pulse wave, temperature, galvanic skin response (GSR)) for continuous monitoring. With materials and chemistry innovation, we have made major progress to improve the long term stability of our wearable chemical sensors (most of these sensors now have stable performance during multiple-day in vitro operation). We also developed a portable cortisol sensor for rapid stress hormone analysis in human sweat. Our engineering development in Year 1 paves the way for an integrated multimodal system and now we have integrated all these sensors into one multimodal wearable system for physical and chemical sensing on human subjects. We are trying to simultaneously monitor the molecular analytes in human sweat including stress hormone, glucose, lactate, sodium, potassium, pH, and key vital signs (i.e., skin temperature, pulse wave, GSR) using this wearable multimodal sensing platform. In the coming year, we plan to mainly focus on system validation and human studies -- evaluating the multimodal sensor performance in human subjects under different stressors and we will use machine learning to classify the stress levels based on the data collected. Based on a combination of the physical/molecular data and machine learning model, we anticipated that a more comprehensive stress assessment system with significantly higher accuracy and robustness can be achieved.

As COVID-19 became a major challenge for us over the past year, we have also developed, partially supported by the Translational Research Institute for Space Health (TRISH), an ultrasensitive and low-cost telemedicine platform, the SARS-CoV-2 RapidPlex, based on target-specific immunoassays built off laser-engraved graphene for rapid and remote assessment of COVID-19 biomarkers (i.e., nucleocapsid protein, anti-spike protein IgG and IgM, and C-reactive protein). We successfully demonstrated the platform's applicability using COVID-19-positive and COVID-19-negative serum and saliva samples. The SARS-CoV-2 RapidPlex has the potential to quickly and effectively triage patients and track infection progression, allowing for the clear identification of individuals who are infectious, vulnerable, and/or immune.

Our approach is to simultaneously monitor the molecular analytes in human sweat including stress hormones (i.e., cortisol, adrenaline, and noradrenaline), glucose, lactate, sodium, potassium, pH, sweat rate, and key vital signs (i.e., skin temperature, blood pressure, heart rate, and heart rate viability) using the wearable multimodal sensing platform. Based on a combination of the physical/molecular data and machine learning model, a more comprehensive stress assessment system with significantly higher accuracy and robustness can be achieved. This project could provide crucial insight into the stress level of NASA astronauts and result in significant benefits by improving human performance through timely intervention.

To accomplish our goal, we propose the following objectives: (1) Develop a multimodal wearable sensor system for real-time monitoring of physical and molecular parameters. By adapting our existing physical and sweat sensing technologies, we will develop a multimodal platform for both sweat and vital sign analysis. The wearable system will wirelessly communicate with a custom designed user interface. (2) Conduct human trials on dynamic stress response assessment using the multimodal sensing system in both laboratory and real-life settings. We will conduct multiple stress-inducing training sessions to collect large sets of physical and molecular data. (3) Develop the signal processing software and determine the predictive algorithm(s) of stress and anxiety via machine learning. Sensor fusion algorithms will be used on the multimodal signals to extract the key features as well as increase the accuracy of the monitoring process. Stress and anxiety classification algorithms will analyze the derived features to detect stress levels. The combination of the multimodal system (hardware), data processing and machine learning model (software) could provide an attractive solution for accurate deep space stress assessment.