Data-driven systems to detect physical weakening from daily routine: A pilot study on elderly over 80 years old

Area of Science:

• Gerontology
• Biomedical Engineering
• Data Science

Background:

• Telemonitoring with wearable sensors shows promise for preventing and treating physical decline in older adults.
• Existing research often lacks detail on study cohorts, protocols, and relevant features for assessing frailty.
• Characterizing physical function in older adults requires robust, data-driven approaches.

Purpose of the Study:

• To investigate the efficacy of data-driven systems for characterizing physical function in individuals over 80 with impaired mobility.
• To develop and validate an automated process for feature extraction and machine learning classification of frailty.
• To assess the feasibility of unsupervised, daily routine monitoring for identifying frailty in the elderly.

Main Methods:

• An automated process was developed to extract heterogeneous time-domain features from 24-hour acceleration and barometric data.
• Statistical testing was employed to identify the most discriminant features for frailty assessment.
• Machine learning classifiers were trained using these features to differentiate between frail and non-frail subjects.

Main Results:

• The proposed system achieved an accuracy of up to 93.51% in distinguishing frail from non-frail individuals.
• Analysis of 570 days of recordings highlighted the importance of longitudinal data for specific frailty diagnosis.
• The study successfully characterized older individuals over 80 with impaired physical function during daily routines.

Conclusions:

• Data-driven telemonitoring systems can effectively characterize physical function and identify frailty in older adults.
• The proposed automated feature extraction and machine learning approach offers a promising tool for geriatric care.
• Longitudinal monitoring using wearable sensor data is crucial for accurate and specific frailty diagnosis in the elderly.