Feasibility of assessing cognitive impairment via distributed camera network and privacy-preserving edge computing

Chaitra Hegde ^1,2, Yashar Kiarashi ¹, Allan I Levey ³

⁰Department of Biomedical Informatics Emory University Atlanta Georgia USA.

Alzheimer's & Dementia (amsterdam, Netherlands) +

|

February 25, 2025

View abstract on PubMed

Summary

Automated monitoring of mild cognitive impairment (MCI) using privacy-preserving cameras accurately detects behavioral differences. This technology differentiates cognitive function levels through movement and social interaction analysis.

Area Of Science

Gerontology
Artificial Intelligence
Computer Vision

Background

Mild cognitive impairment (MCI) is characterized by cognitive decline exceeding normal aging.
MCI is associated with reduced social engagement and increased undirected movement.
Automated behavioral analysis can enhance longitudinal monitoring of MCI.

Purpose Of The Study

To develop an automated system for detecting behavioral changes in individuals with MCI.
To differentiate between higher and lower cognitive functioning MCI groups using machine learning.
To assess the feasibility of using a privacy-preserving camera network for MCI monitoring.

Main Methods

A distributed, privacy-preserving camera network was deployed in an indoor setting.
Movement and social interaction features were extracted from collected video data.
Machine learning algorithms were trained to classify MCI patients into distinct cognitive functioning groups.

Main Results

Significant differences were observed in movement and social interaction patterns between high- and low-functioning MCI cohorts.
Key differentiating features included path length, walking speed, direction changes, and group formation frequency.
A machine learning model achieved 71% accuracy in differentiating cognitive levels, even without individual identity data.

Conclusions

Edge computing and privacy-preserving camera networks can effectively differentiate MCI levels.
Behavioral analysis of movement and social interaction provides valuable insights into cognitive status.
This approach offers a non-invasive method for objective MCI assessment and monitoring.

Keywords:

computer vision edge computing gait mild cognitive impairment social interaction spatial navigation