A research team from KAIST, Sungkyunkwan University, and Korea University Anam Hospital has developed an AI framework that analyzes daily lifelog data from older adults at home to identify early warning signs of cerebrovascular disease with 96.5% accuracy. The study, based on real-world data from 1,224 older adults, was published in npj Digital Medicine and could help shift clinical practice toward earlier intervention before symptoms appear.
How the AI spots risk signals in daily life
The researchers used 13,362 two-week lifelog samples collected in real residential environments by LivOn Care Co., Ltd. The AI analyzes daily activity, sleep, circadian rhythm, and indoor environmental information, along with age and chronic disease data. It identifies a prodromal phase-the period when subtle changes begin to emerge before a clinical diagnosis is made.
The system distinguished between data from four weeks before diagnosis (the "imminent diagnostic risk period") and data from 12 weeks before diagnosis with 96.53% accuracy. The team applied explainable AI to surface the specific behavioral and environmental factors behind each assessment.
Behavioral markers linked to cerebrovascular disease
The analysis revealed several patterns that were common among older adults in the prodromal phase:
- Frequent continuous activity between 10 p.m. and 2 a.m., when the body normally prepares for sleep
- Delayed sleep onset and a reduced distinction between day and night activity
- A noticeable drop in continuous activity during the evening window from 6 p.m. to 10 p.m., with more inactive time as diagnosis approached
- Low indoor humidity, indicating a dry indoor environment
These markers capture irregular daily rhythms and environmental conditions that are difficult to detect through standard hospital examinations alone.
From reactive to preventive care
"The key point of this study is not that AI should replace a hospital diagnosis, but that it can first detect risk signals in small lifestyle changes at home and help connect patients to medical care at the right time," said Professor Lisa Lim from KAIST's Department of Civil and Environmental Engineering.
The technology is designed as a supportive tool for older adults who may struggle to describe their own condition clearly. It offers objective health monitoring that can provide early warning indicators to clinicians and caregivers. This approach aligns with broader efforts in AI for Healthcare that emphasize prevention over late-stage treatment.
The team stressed that the system does not predict the exact onset of cerebrovascular disease or replace a clinical diagnosis. Prospective validation in larger patient groups will be necessary before real-world deployment.
Why this matters for healthcare professionals
For clinicians and care teams working with aging populations, this research points toward a practical way to monitor at-risk patients between office visits. Subtle shifts in daily rhythm-like delayed sleep or reduced evening activity-can now be tracked passively at home and flagged before a crisis occurs. That could shorten the window between symptom emergence and medical consultation, and support earlier, less invasive interventions for cerebrovascular disease.
Your membership also unlocks:
