A State-of-the-Art Review of Computational Models for Analyzing Longitudinal Wearable Sensor Data in Healthcare
Organizations: Department of Electrical and Computer Engineering, PERFORM Center, Concordia University, Montreal, QC, Canada
Abstract
Wearable devices are increasingly used as tools for biomedical research, as the continuous stream of behavioral and physiological data they collect can provide insights about our health in everyday contexts. Long-term tracking, defined in the timescale of months of year, can provide insights of patterns and changes as indicators of health changes. These insights can make medicine and healthcare more predictive, preventive, personalized, and participative (The 4P's). However, the challenges in modeling, understanding and processing longitudinal data are a significant barrier to their adoption in research studies and clinical settings. In this paper, we review and discuss three models used to make sense of longitudinal data: routines, rhythms and stability metrics. We present the challenges associated with the processing and analysis of longitudinal wearable sensor data, with a special focus on how to handle the different temporal dynamics at various granularities. We then discuss current limitations and identify directions for future work. This review is essential to the advancement of computational modeling and analysis of longitudinal sensor data for pervasive healthcare.
Figures & tables
| Metric | Description | Formula |
|---|---|---|
| Rhythm-based metrics | ||
| Interdaily Stability (IS) [ 41 ] | A measure of how stable a person’s rhythm is over multiple days, calculated from samples of activity levels , aggregated hourly hour-bins. represents the average activity during hour . IS scores range from 0, indicative of a total lack of rhythm, to 1, indicative of perfectly stable rhythm | |
| Intradaily Variability (IV) [ 41 ] | A measure of the fragmentation of activity or how activity level shifts between two consecutive time slots calculated from samples of activity levels , aggregated hourly hour-bins. IV scores range of 0 to 2 with higher values indicating higher fragmentation | |
| Rest-activity Amplitude [ 93 ] | Measures the difference between a person’s daily activity (measured as the mean activity level over the most active 10 consecutive hours (M10)) and nocturnal activity (measured as the mean activity during the least active 5 consecutive hours (L5)) | |
| Regularity Index [ 25 , 91 , 27 ] | Compares the uniformity of a person’s schedule, comparing hourly (t) data from two days (a,b). Sensor measurements ( ) are normalized to (-0.5, 0.5) | |
| Stability Index [ 46 ] | Measures the inverse of the median across a period of multiple days P of the distances between the normalized cumulative activity distributions ( ) on two different days. It characterizes the degree to which behavioral patterns varied across the period. A higher index means higher stability. | where |
| Feature | Routines | Rhythms | Regularity Metrics |
| Focus | Sequences and co-occurrence of | Periodicity and regularity of | Quantifying the strength of |
| discrete events | continuous processes | repetitiveness in behavioral or | |
| physiological patterns without | |||
| explicitly modeling the patterns | |||
| themselves | |||
| Modeling | Frequent sequences, probabilistic | Sinusoidal functions | Statistical summaries or indices |
Wearable Sensing
Wearable sensing uses body-worn devices to measure motion, physiology, and interaction for health, activity, and assistive applications.
See all 143 papers