Leveraging Non-Wrist Wearables for Healthcare and Clinical Trials

Contributed Commentary by Geoffrey Gill, MS, Shimmer Americas 

July 29, 2022 | Consumer wearables—Apple Watches, Fitbits, Xiaomi Bands—have shaped the popular imagination of wearables as wrist-worn devices. Still, clinical researchers would be remiss to overlook wearable health sensors worn elsewhere on the body.  

Non-wrist wearables unlock a wealth of opportunities and insights otherwise unavailable. These applications spur us to sound, once again, the clarion call for raw data collection and open-source collaboration.  

Wrist-Worn Wearables  

Most clinical trials employ wrist-worn wearables, which help us glimpse a promising future where patients and their conditions can be better understood. New medicines can be more accessible with more precise and efficacious dosing, and risks can be measured and reduced.  

However, wrist-worn wearables tell only part of the story, as moving our hands to gesticulate, cook dinner, or fold the laundry may induce misleading signals.  

Examples of Non-Wrist Wearables in Clinical Trials  

Employing non-wrist wearables enlarges the scope and quality of mobile health insights across various body areas and characteristics: 

Posture: Posture and activity can be more precisely measured using a combination of sensors worn in the lumbar region (i.e., lower back) and front of the thigh. A sensor's inertial measurement unit can provide its orientation relative to gravity, allowing us to distinguish between standing, sitting, and lying. In addition, walking and running can be inferred from lumbar sensor-derived activity levels during each epoch based on extensive research by the academic community.  

Gait Analysis: Non-wrist sensors can facilitate meaningful medical interventions. For example, Kinesis’ gait and mobility assessment technology uses data from leg sensors to assess a participant’s risk of falling. Healthcare professionals can recommend appropriate exercise and rehabilitative interventions, which reduce the rate of severe falls by roughly half. Furthermore, these leg-worn sensors enable healthcare teams to gauge a participant’s progress objectively.  

Wear Detection: Wearables can improve clinical trial protocol compliance. However, traditional wearable devices must be worn for an extended period to confer clinical benefits, but patients can find them uncomfortable. Analog-based wear assessment approaches such as patient diaries are burdensome and often rely on inaccurate patient memories. In contrast, a wearable—capable of non-wear detection—attached to the device was 99% accurate, allowing a more precise calculation of the impact of wear duration.  

Dosage Refinement: Multiple body-worn sensors are used to determine the correct drug dosage for an individual patient. ClearSky uses sensors on the chest, arms, and thighs to detect dyskinesia induced by levodopa treatment in people with Parkinson’s disease (PD). Data from this group of sensors distinguish between levodopa-induced and PD-caused tremors. This data enables a physician to set a drug dosage tailored to each patient.  

Baseline Investigation: Non-wrist wearables open new vistas of patient research and understanding. Wearables have been placed on newborns' feet to measure galvanic skin response and identify early signs of mental health issues. Wearables also produce baseline data on infant movement when an accelerometer is placed on each limb of participating infants.  

The use of non-wrist wearables provides insights and capabilities otherwise unavailable. However, because non-wrist use cases are less common, the need for raw data collection and open-source collaboration gains greater significance.  

Raw vs. Processed Data in Clinical Trials  

Wearables in healthcare—wrist or non-wrist—should provide raw data in addition to processed data. Raw data is essential when similar wearable research studies are scarce, as is often the case with non-wrist wearables. Raw data facilitates ease of verification (e.g., the first component of The Digital Medicine Society’s V3 framework) and allows for enhanced comparisons among various wearable-related studies. Raw data are algorithm agnostic; therefore, researchers can evaluate the utility of different algorithms or apply future new algorithms retroactively. Raw data maximizes the useful life of a trial or study’s dataset. Collecting raw data also serves as a foundation for open-source collaboration on algorithms used in wearables.  

Open-Source Collaboration  

An open-source approach to algorithm development is crucial, especially when non-wrist wearable use cases are scarce. Wearables enable us to measure and derive any number of digital clinical endpoints, but their value—and challenge—lies in having those endpoints validated and broadly accepted. Validating a single endpoint is time-consuming and expensive, and this difficulty multiplies for each endpoint used. Thus, validating every proprietary algorithm is unfeasible. An open-source approach prizes transparency, cuts through tedium, inspires confidence, and creates a plug-and-play dynamic.  

Geoffrey Gill is President of Shimmer Americas, leading the U.S. operations and the commercial efforts for North and South America for Shimmer Research, a designer and manufacturer of medical-grade wearables. Geoffrey is also a Co-founder of the Open Wearables Initiative, an industry collaboration designed to promote the effective use of high-quality, sensor-generated health measures in clinical research through the open sharing of algorithms and datasets. He can be reached at ggill@shimmersensing.com.