A person’s daily routine may contain hidden clues about future brain health, according to a new study from the Korea Advanced Institute of Science and Technology (KAIST).
Researchers have created an artificial intelligence system that looks at everyday activities inside the home to identify older adults who may be entering the early stages of cerebrovascular disease.
Their research was published in npj Digital Medicine.
Cerebrovascular disease develops when blood vessels that carry blood to the brain become damaged or blocked. Stroke is one of the best-known examples. Many people recover slowly after a stroke, while others are left with lifelong disability.
Doctors know that early treatment leads to better outcomes, but finding people before major symptoms appear has always been difficult. This is why scientists are searching for safe, simple ways to recognise risk earlier.
The new research takes a different approach. Instead of waiting for patients to report symptoms, the AI quietly studies everyday life. The researchers worked with data collected from 1,224 older adults living in normal homes.
More than 13,000 two-week lifelog samples were analysed. These records described sleeping habits, movement, daily activity patterns, indoor conditions and personal health information such as age and chronic illnesses.
By combining all of these details, the AI learned to recognise subtle lifestyle changes linked with the early phase of cerebrovascular disease. Hospital examinations usually provide only a snapshot of a person’s health, but lifelog information shows how behaviour changes over weeks and months. This long-term picture helped the system identify patterns that might otherwise go unnoticed.
The study found that older adults at higher risk often remained active late at night between 10 p.m. and 2 a.m., suggesting disrupted sleep and body rhythms. As the time of diagnosis became closer, evening activity dropped while periods of inactivity became longer. The researchers also discovered that dry indoor air was associated with a higher likelihood of being close to diagnosis.
To test whether the AI could recognise increasing danger, the researchers compared data collected during the four weeks before diagnosis with information from around 12 weeks before diagnosis. The system separated these periods with an impressive accuracy of 96.53 percent. This indicates that gradual lifestyle changes may become stronger as cerebrovascular disease develops.
Another important feature was the use of explainable AI. Rather than acting like a mysterious black box, the system highlighted which behaviours and environmental conditions influenced each prediction. This makes the technology easier for healthcare professionals to understand and trust.
The researchers hope future versions could support doctors, families and caregivers by encouraging earlier medical assessment for people whose daily routines begin to change. It could be especially useful for older adults who find it difficult to notice or describe subtle health changes themselves.
Overall, the study offers an exciting example of how AI can support preventive healthcare. At the same time, the researchers stress that it is not designed to replace medical diagnosis or predict the exact moment a stroke or other cerebrovascular disease will happen.
More studies involving larger groups of patients will be needed before the technology becomes part of everyday healthcare. Even with these limitations, the findings suggest that carefully analysing ordinary life at home could one day help save lives through earlier detection and treatment.
If you care about stroke, please read studies that diets high in flavonoids could help reduce stroke risk, and MIND diet could slow down cognitive decline after stroke.
For more health information, please see recent studies about antioxidants that could help reduce the risk of dementia, and tea and coffee may help lower your risk of stroke, dementia.
Source: Korea Advanced Institute of Science and Technology (KAIST).
