Inferring Destinations and Activity Types of Older Adults From GPS Data: Algorithm Development and Validation

Area of Science:

• Gerontology
• Geographic Information Systems (GIS)
• Health Informatics

Background:

• Outdoor mobility is crucial for assessing older adults' functional status.
• Existing Global Positioning System (GPS) applications focus on spatiotemporal mobility aspects.
• A gap exists in classifying older adults' outdoor mobility based on semantic dimensions (intentions, motivations).

Purpose of the Study:

• To assess the feasibility of using GPS to determine semantic dimensions of older adults' outdoor mobility.
• To identify destinations and infer activity types (e.g., food, shopping, sport) from GPS data.

Main Methods:

• Five healthy participants (≥65 years) used GPS devices for 4 weeks.
• Participants maintained travel diaries to record excursion details.
• An algorithm was developed to extract destinations and infer activity types from GPS data, validated against diaries.

Main Results:

• GPS successfully detected stop locations with an 87% F1 score.
• Home locations were identified with an average accuracy of 40.18 meters.
• Activity inference achieved an 86% F1 score, with accuracy varying by neighborhood Walk Score.

Conclusions:

• GPS technology demonstrates accuracy in determining semantic dimensions of outdoor mobility.
• Further refinement is necessary for robust clinical adoption of GPS-based mobility assessment systems.