Buy or Build: Challenges Developing Consumer Digital Health Interventions.

BACKGROUND: Digital health interventions offer opportunities to improve collaborative care between clinicians and patients. Designing and implementing digital health interventions requires decisions about buying or building each technology-related component, all of which can lead to unanticipated issues. OBJECTIVES: This study aimed to describe issues encountered from our "buy or build" decisions developing two digital health interventions over different timeframes, designed to use patient-generated health data to: (1) improve hypertension control and (2) measure and improve adherence to HIV-related medications. METHODS: CONDUIT-HID (CONtrolling Disease Using Information Technology-Hypertension In Diabetes) was developed during 2010 to 2015 to allow patients receiving care from a multispecialty group practice to easily upload home blood pressure readings into their electronic health record and trigger clinician action if mean blood pressure values indicated inadequate control. USE-MI (Unobtrusive SEnsing of Medication Intake) was developed from 2016 to 2022 to allow entry of patients' HIV-related medication regimens, send reminders if patients had not taken their medications by the scheduled time(s), attempt to detect medication ingestion through machine learning analysis of smartwatch motion data, and present graphical adherence summaries to patients and clinicians. RESULTS: Both projects required multiple "buy or build" decisions across all system components, including data collection, transfer, analysis, and display. We used commercial, off-the-shelf technology where possible, but virtually all of these components still required substantial custom development. We found that, even though our projects spanned years, issues related to our "buy or build" decisions stemmed from several common themes, including mismatches between existing and new technologies, our use case being new or unanticipated, technology stability, technology longevity, and resource limitations. CONCLUSION: Those designing and implementing digital health interventions need to make numerous "buy or build" decisions as they create the technologies that underpin their intervention. These "buy or build" decisions, and the ensuing issues that will arise because of them, require careful planning, particularly if they represent an "edge case" use of existing commercial systems.

iLid: Low-power Sensing of Fatigue and Drowsiness Measures on a Computational Eyeglass.

The ability to monitor eye closures and blink patterns has long been known to enable accurate assessment of fatigue and drowsiness in individuals. Many measures of the eye are known to be correlated with fatigue including coarse-grained measures like the rate of blinks as well as fine-grained measures like the duration of blinks and the extent of eye closures. Despite a plethora of research validating these measures, we lack wearable devices that can continually and reliably monitor them in the natural environment. In this work, we present a low-power system, iLid, that can continually sense fine-grained measures such as blink duration and Percentage of Eye Closures (PERCLOS) at high frame rates of 100fps. We present a complete solution including design of the sensing, signal processing, and machine learning pipeline; implementation on a prototype computational eyeglass platform; and extensive evaluation under many conditions including illumination changes, eyeglass shifts, and mobility. Our results are very encouraging, showing that we can detect blinks, blink duration, eyelid location, and fatigue-related metrics such as PERCLOS with less than a few percent error.