Yale and UArizona Researchers Develop Big Data Solution for Measuring Stress in Emergency Healthcare Providers

By automating the processing of massive volumes of biometric data derived from wearable sensors, the HRVEST tool created by Yale University and University of Arizona researchers not only streamlines the extraction of meaningful metrics but also introduces a robust methodology for noise filtration and data quality assurance in emergency healthcare.

 
One of the greatest risks to emergency healthcare is physician burnout. Clinicians face constant high-stress situations, compounded by long hours, sleep deprivation and exposure to traumatic events.

Research is underway to reduce emergency healthcare worker stress, beginning with establishing a data-driven baseline for physiologic stress. But tracking stress in real time—with the goal of improving emotional wellbeing and, subsequently, clinical decision-making and patient outcomes—has been difficult. Data from wearables such as smartwatches may not be accurate or consistent, and when it is, it may be too vast to calculate efficiently or may require cost-prohibitive, proprietary software.

In response, Yale University School of Medicine researchers, in partnership with University of Arizona data scientists from the School of Information, Data Science Institute and SensorLab, the Healthcare Technology Innovation Lab, have developed a groundbreaking tool: the Heart Rate Variability Experimental Sensor Toolkit, or HRVEST. HRVEST is an open-source algorithm that leverages Python programming to batch analyze physiological stress data, automating the detection and removal of noise and artifacts.

Yale researchers recruited 81 emergency room physicians to don smart garments to gauge the autonomic nervous system's response to stress through such indicators as heart rate variability (HRV). Research has found that HRV—the beat-to-beat changes in heart rate over time—is a pivotal physiological stress marker.

The emergency room physicians wore Hexoskin Smart Shirts, which utilize embedded biometric sensors for continuous monitoring. Because they require direct contact with skin, the smart garments were worn under standard medical attire, allowing researchers to capture HRV and other essential data and activity levels without interrupting the physicians' workflow. The Hexoskin Smart Device, paired with the shirt, ensured seamless data transmission and real-time ECG visualization.

The results of the researchers’ 2021 study funded by the Agency for Healthcare Research and Quality was published in February 2024 in Frontiers in Computer Science.

According to iSchool Professor Winslow Burleson, “This research marks a significant advancement in measuring stress among emergency medicine physicians. By utilizing smart garments in a clinical setting, the study not only provided precise and continuous data but also highlighted the importance of technological integration for health monitoring. The HRVEST algorithm facilitated the processing of large datasets, demonstrating its potential for widespread application in both research and clinical environments.”

The study's methodology underscores the significance of multidisciplinary collaboration, bringing together data scientists, clinicians and human factors engineers to refine HRVEST into a user-friendly, widely applicable tool.

The research outcomes of the HRVEST project apply beyond the immediate context of emergency medicine. The tool's adaptability to other wearable sensors and its potential for real-time feedback offer promising avenues for research and clinical practice aimed at improving the wellbeing of healthcare providers across specialties. The open-source nature of HRVEST likewise invites further innovation, setting the stage for a new era of technology-driven solutions to combat burnout and support healthcare provider resilience.
 

Learn more about the dynamic research of the School of Information, or view ways you can support faculty and research in information and data science.