An Information-Driven Approach for the Early Health Technology Sustainability Assessment and the Frugal Design of the Internet of Medical Things: An Exploratory Study of Wearable Activity Monitoring Devices (Preprint)

BACKGROUND: Wearable Activity Monitoring (WAM) devices have become increasingly prevalent in the last decade for improving quality of life, and the prevention and day-to-day management of a wide range of health disorders. WAM devices based on the Internet of Medical Things (IoMT) paradigm offer a practical means of tracking Physical Activity (PA), but their widespread use raises sustainability concerns. Meanwhile, established methodologies such as Health Technology Assessment (HTA) and Life Cycle Assessment (LCA) are typically applied at very advanced stages of development, when empirical certainty about the final design and operating conditions is available, leaving little room for subsequent improvements. OBJECTIVE: This exploratory work aims to provide the empirical foundations for an information-driven approach that addresses this paradox in the development, early evaluation, and conception of wrist-worn step counters, for which recent evidence suggests overdimensioned and unsustainable electronic designs. Specifically, we seek to identify optimal resource-performance trade-offs in critical electronic components of frugal smartbands, based on the amount of data they can collect and the information they preserve for step counting. In this manner, we explicitly account for uncertainties in device reliability, variability in users' gait speeds, and material and energy consumption in final products. METHODS: We proceeded in three steps. First, we conducted a secondary analysis of an existing accelerometer dataset characterizing wrist motion in healthy individuals walking at different speeds. The original sampling rate of the x-, y-, and z-axis signals was progressively reduced using cubic spline interpolation and then discretized to quantify the information preserved in the downsampled signals, first as a function of frequency variations and then with respect to changes in motion velocity. Additionally, we assessed preliminarily the viability of the downsampled signals for step detection by estimating percentage errors in peak and valley counts. Based on this analysis, we constructed and evaluated four design archetypes for frugal smartbands, linking energy consumption and sampling frequency for four widely used accelerometers, and combining the capabilities of essential electronic components (transceivers and microcontrollers) required to implement a suboptimal asynchronous First-In-First-Out (FIFO) algorithm operating at different sampling rates. In a third step, we evaluated the environmental impact and circularity of these components through a streamlined analysis focused exclusively on their raw materials. RESULTS: Between 70% and 90% of the information is lost in signals downsampled at very low frequencies (between 2 Hz and 5 Hz), whereas moderate losses below 24% are observed from 20 Hz onwards. Additionally, substantial information loss occurs when individuals walk briskly or jog (≥ 8 km/h), or when they walk at normal speeds below 8 km/h with sampling frequencies below 7 Hz or above 25 Hz. Step-counting accuracy is expected to be acceptable from approximately 11 Hz onwards. Conversely, higher sampling rates rapidly saturate FIFO buffers and increase energy overhead, particularly when implemented in memory-dense components handling both data transfer and processing. Finally, gold and silver in transceivers and microcontrollers contribute significantly to resource depletion, while copper content remains relevant for potential material recovery. CONCLUSIONS: These findings provide preliminary insights toward the concurrent assessment and development of frugal WAM devices. This work extends current understanding in step detection under knowledge-constrained conditions and provides concrete mechanisms to reduce uncertainty during early Health Technology Assessment and eco design.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.