Behind the Screen: A Narrative Review on the Translational Capacity of Passive Sensing for Mental Health Assessment

Abstract

Mental health disorders—including depression, anxiety, trauma-related, and psychotic conditions—are pervasive and impairing, representing considerable challenges for both individual well-being and public health. Often the first challenges to treatment include financial, geographic, and stigmatic barriers, which limit the accessibility of traditional assessment measures. Further, compounded by frequent misdiagnosis or delayed detection, there is a need for effective, accessible, and scalable approaches to identification and management. Considering advances in computing and the ubiquitous nature of personal mobile and wearable technology, this narrative review examines the utilization of passive sensor data as a screening and diagnostic tool for mental disorders. As an alternative to traditional screening measures, passive sensing offers a tool to overcome barriers that prevent many from seeking services. We critically assess the literature up to September 2023, exploring the use of passive data—such as heart rate variability, movement patterns, and geolocation—to predict mental health outcomes across a spectrum of disorders. From a translational perspective, our review explores the state of passive sensing science, with special emphasis on the capacity for the science to be implemented in real world clinical and general populations, a novelty specific to this review to the best of our knowledge. Toward this aim, we consider multiple study factors, including participant demographics, data collection methods, sensor modalities, outcome measures, and analytic modeling approaches. We find that passive sensing features, such as GPS, heart rate, and actigraphy offer promise for enhancing early detection and improving the diagnostic process for mental disorders. Despite this promise, however, our findings highlight important limitations in passive sensing research including (1) a trend toward smaller, specialized samples, (2) a predominance of data collection apps built on the Android operating system, and (3) a reliance on self-reported measures as proxies for important clinical outcomes. These limitations ultimately stymie efforts to implement and scale important research findings in larger and more heterogeneous populations. With future translational research in mind, we emphasize the importance of validating passive sensing findings with larger, more diverse samples and ensuring assessment tools can be deployed across multiple device types and operating systems. Further, where possible, we emphasize the need for robust, objectively validated outcome measures, such as by clinician assessment. We conclude that careful consideration of translational factors in the design of future research will aid in enhancing the impact of future passive sensing studies, ultimately enhancing mental health outcomes on a broad scale.

Michael V. Heinz, M.D.

Research Psychiatrist

Michael V. Heinz is a practicing psychiatrist and researcher at Dartmouth College interested in scalable digital methods for assessment and intervention in mental health.