Smartwatches and AI: wearable data may help predict changes in brain health

Researchers at the University of Geneva (UNIGE) have shown that data from smartwatches and smartphones can help predict changes in emotional and cognitive health. Artificial intelligence analyzed physiological and environmental data to identify early indicators of fluctuations in brain health.

Their findings, published in npj Digital Medicine, suggest that connected devices can predict changes in emotional and cognitive functioning with considerable accuracy.

According to the World Health Organization, more than one in three people worldwide live with neurological conditions such as stroke, epilepsy, or Parkinson’s disease. In addition, more than half of all individuals will experience a mental disorder during their lifetime. Because brain health can fluctuate over time due to environmental influences and lifestyle factors, monitoring these changes may help support early detection and prevention strategies.

To explore this possibility, a research team at UNIGE equipped 88 volunteers aged 45 to 77 with a smartphone app and a smartwatch. Over a period of ten months, the devices continuously collected passive data without requiring participants to alter their daily routines.

The collected information included:

• heart rate patterns
• physical activity levels
• sleep patterns
• weather conditions
• air pollution exposure

In total, researchers analyzed 21 different indicators.

Every three months, participants also provided active data by completing questionnaires about their emotional state and performing cognitive tests.

After the data collection period, researchers applied artificial intelligence models developed within the project to analyze the passive data. "The aim was to determine whether AI could predict fluctuations in participants' cognitive and emotional health based on these data," explains Igor Matias, a doctoral assistant at the Research Institute for Statistics and Information Science at the Geneva School of Economics and Management and lead author of the study.

The AI-generated predictions were then compared with the results from questionnaires and cognitive tests. The results showed that the models were able to predict changes with an average error rate of 12.5 percent.

According to the research team, this level of accuracy suggests that wearable technologies could contribute to early detection of changes in brain health.

The study found that emotional states were predicted more accurately than cognitive performance.

• Emotional states: error rates typically between 5 percent and 10 percent
• Cognitive states: error rates between 10 percent and 20 percent

Researchers also identified the most informative passive indicators. For cognitive functioning, the strongest predictors included: air pollution exposure, weather conditions, daily heart rate, and variability in sleep patterns. For emotional states, the most influential indicators were weather conditions, sleep variability, and heart rate during sleep.

The study forms part of the interdisciplinary project Providemus alz, which focuses on identifying early indicators of cognitive decline. The next phase of the project will extend data collection to 24 months and examine which individual characteristics influence the performance of the AI models. The goal is to better understand how wearable-based monitoring could be applied in personalized health monitoring and real-world prevention strategies.

According to the University of Geneva, the results highlight the potential of combining wearable devices with artificial intelligence to support continuous and non-invasive monitoring of brain health in everyday life.

MEDICA-tradefair.com; Source: Université de Genève