A student wearing some of the sensor equipment in the lab. Photo credit: Fred Zwicky.
Imagine you’re at the store, and suddenly you start to get really anxious. The moment passes, but your doctor has told you to keep track of these events and report them so you can manage your anxiety better. The only problem is that the details of the moment get fuzzy, and you can’t say exactly what triggered the moment. You might have been thinking about something at work or at home. Or maybe it was a stray conversation you heard that reminded you of something you dread. You’re not even really sure how anxious you were.
But what if you’d been wearing something that monitored and recorded your body’s response to stress?
Manuel Hernandez, teaching associate professor in biomedical and translational sciences at the Carle Illinois College of Medicine, is part of a team of NCSA affiliate researchers, including Richard Sowers, professor in industrial and enterprise systems engineering, and Elizabeth Hsiao-Wecksler, Grayce Wicall Gauthier professor in mechanical science and engineering, who have been utilizing Center resources, such as the Delta supercomputer, to develop wearable technology that can capture a person’s physiological state in real time. Perhaps, by having one of these wearable sensors, they’d be able to deliver “just-in-time interventions” – those reminders you might get from your smartwatch to sit down and take a break if your heart is racing.
“This work stemmed from observing the impact of the COVID pandemic on mental health in society, and the need to better monitor mental health changes and to aid in the delivery of just-in-time interventions in adults across the lifespan, and particularly those in high-stress environments,” said Hernandez.
Hernandez’s team identified a key issue with treating anxiety – the information doctors mainly have to work from is reported to them by patients, well after an event occurs.
“Anxiety is a common mental health condition that can significantly impair daily functioning, especially for university students in STEM,” said Hernandez. “State anxiety is a situational emotional response and is typically assessed through self-reported questionnaires and clinical interviews. These traditional methods only capture discrete snapshots of an individual’s emotional state and rely heavily on retrospective reporting. To overcome the limitations of self-reporting, we use wearable and contactless sensors for continuous monitoring and the development of objective markers of mental health status in our research.”
A significant question surrounding the use of wearables to detect anxiety was whether they would remain accurate outside the lab. A lab is a highly controlled environment, but a person’s day-to-day life isn’t. In one of the team’s papers published in the journal Sensors, the team demonstrated that despite the randomness often found in real-life situations, their wearable technology could still accurately identify anxiety.
Having a real-time report of things like blood pressure and heart rate isn’t just something that can be applied in a general health care setting. Certain professions are associated with higher stress levels, and understanding precisely what a person’s body is going through during their job could help prevent someone from reaching a critical point before permanent damage is done.
“This research has implications for the devices, sensors and approaches we could take to discreetly monitor changes in mental health status in stressful environments, such as first responders or health care systems,” said Hernandez.
This type of work greatly benefits from access to high-performance computing (HPC) resources, such as the GPUs and CPUs available on NCSA’s Delta supercomputer. With HPC resources, researchers can process months of data in minutes.
Hernandez’s team was able to easily get an allocation for compute resources through the U.S. National Science Foundation (NSF) ACCESS program. “NCSA resources helped our team fast-track the algorithm development and data analysis of both machine learning and deep learning models for use in state anxiety detection,” said Hernandez.
NCSA resources on campus provide an incredible opportunity for researchers to leverage state-of-the-art computing resources and technical expertise with emerging research questions with the potential to transcend disciplines and address significant societal challenges.
Teaching associate professor, Carle Illinois College of Medicine
Their research involved processing “multimodal” data (multiple types of signals) from wearables, such as electrocardiograms (ECG) and respiration monitors. These sensors generate large volumes of data; for instance, some signals were recorded at 256 hertz (256 data points per second), which requires significant power to clean and analyze.
Another useful function of the HPC resources used was in simulating conditions for the machine learning required to create their models. Hernandez and his team utilized artificial intelligence to simulate real-world variables, such as increased movement or sweat, to ensure their model was as accurate as possible.
Hernandez’s team has made significant progress with their research. In addition to their published work in Sensors, they’ve also published findings in IEEE Transactions on Affective Computing and Applied Sciences. But the team’s work isn’t finished – they have plans to keep evolving the research going forward.
“Our next steps are to evaluate what interventions may be most effective at alleviating the negative impacts of stress on mental health, and to evaluate how effectively we can monitor changes in anxiety while we carry out everyday activities,” said Hernandez.
You can find further reading about this research in the following publications:
A Scoping Review of ML Approaches in Anxiety Detection from In-Lab to In-the-Wild in the journal Applied Sciences
Objective Anxiety Level Classification Using Unsupervised Learning and Multimodal Physiological Signals in the journal Smart Health
Physical, Social and Cognitive Stressor Identification using Electrocardiography-derived Features and Machine Learning from a Wearable Device in the Proceedings of the 2024 of the IEEE-Engineering in Medicine and Biology Society (EMBC) International Conference
ABOUT DELTA AND DELTAAI
NCSA’s Delta and DeltaAI are part of the national cyberinfrastructure ecosystem through the U.S. National Science FoundationACCESS program. Delta (OAC 2005572) is a powerful computing and data-analysis resource combining next-generation processor architectures and NVIDIA graphics processors with forward-looking user interfaces and file systems. The Delta project partners with the Science Gateways Community Institute to empower broad communities of researchers to easily access Delta and with the University of Illinois Division of Disability Resources & Educational Services and the School of Information Sciences to explore and reduce barriers to access. DeltaAI (OAC 2320345) maximizes the output of artificial intelligence and machine learning (AI/ML) research. Tripling NCSA’s AI-focused computing capacity and greatly expanding the capacity available within ACCESS, DeltaAI enables researchers to address the world’s most challenging problems by accelerating complex AI/ML and high-performance computing applications running terabytes of data. Additional funding for DeltaAI comes from the State of Illinois.