JMIR Publications

ABSTRACT

Background:

The population of adults aged 65 and older is rapidly increasing, while the availability of caregivers is declining. Smart homes that provide unobtrusive, continuous monitoring and alerting on clinically relevant changes in daily activity patterns offer a potential innovative solution for aging in place.

Objective:

To prototype and evaluate a low-cost, community-based smart health system for monitoring the health of older adults with multiple chronic conditions and experiencing poverty, and to identify barriers and facilitators to adoption.

Methods:

Using a prospective, mixed-methods design and iterative community co-design, 46 older adults from 7 different language groups were continuously monitored for 6 months with ambient sensors installed in their homes. Two older adults were monitored for 4 and 5 months respectively resulting in a total sample of N=48. The system generated alerts based on movement pattern changes and escalated notifications to participants, support persons, community health workers (CHWs), and nurses. Sensor data were analyzed descriptively to quantify alert patterns and response rates, while text-based data from in-the-moment surveys, CHW and nurse notes, and semi-structured interviews underwent qualitative descriptive analysis and reflexive thematic coding.

Results:

The system generated 37 million sensor readings condensed into 1.2 million high-level events and 8,927 novel alerts. Qualitative data comprised of 34,086 words of text. Participants responded to 1.57% of initial email alerts and 7.8% of follow-up SMS alerts sent when no email response was received. CHWs responded to 83.6% of escalated alerts, resulting in 1,060 contacts with participants in response to alerts. Clinical contacts resulted in 72 interventions. Three major qualitative themes emerged: (1) mitigating aloneness, (2) building trust in technology and people, and (3) maintaining a human connection. Subthemes included safety, personalization, and digital distress. Participants rated the system highly (Net Promoter Score = 8.48/10) but expressed a strong preference for phone calls over automated alerts. Cultural expectations influenced adoption, particularly in multi-generational households.

Conclusions:

Communities can effectively engage in technology-delivered healthcare. Future research is needed to improve technical aspects of smart health systems, including accurate alerting utilizing machine learning, data visualizations for older adults and healthcare workers, and culturally sensitive features. Additional work should address how and when to communicate automated messaging, engage older adults with their own data, and integrate sensor-based monitoring into healthcare workflows. Research should also explore personalization through advanced computer models such as machine learning and strategies to reduce digital distress. Clinical Trial: None