Can Wearable Tech Help Detect Early Signs of Mental Illness?
Researchers at the University of Geneva (UNIGE) are exploring whether smartphones and smartwatches can detect early signs of neurological or mental health issues. The team monitored a group of participants wearing connected devices and applied artificial intelligence (AI) to analyze data like heart rate, physical activity, sleep patterns, and environmental factors such as air quality. Their study highlights that these devices can effectively predict changes in emotional and cognitive health, paving the way for earlier detection of brain health issues.
Brain health is a significant public concern, with over a third of the global population affected by neurological disorders, including conditions like stroke and Parkinson’s disease. Furthermore, more than half of adults will experience some form of mental health issue—such as anxiety or depression—over their lifetime. As the population ages, these statistics are expected to rise.
Even in healthy individuals, brain health can vary significantly, influenced by environmental factors and lifestyle choices. Understanding daily or weekly shifts in emotional and cognitive functions is vital for creating tailored prevention strategies.
To investigate the potential of wearable technology, a study involved 88 participants, aged between 45 and 77. For ten months, participants used a specific smartphone app and a smartwatch to gather “passive” data, which did not require any adjustments to their daily routines. This data included heart rate, physical activity, sleep patterns, and even local weather and pollution levels. In total, the researchers analyzed 21 different indicators.
Every three months, participants were also asked to provide “active” feedback through questionnaires about their emotional well-being and cognitive performance tests.
AI in Action
After gathering the data, researchers used the AI tools developed for the project to analyze it, aiming to see if AI could predict changes in cognitive and emotional health based on the collected information.
“The goal was to see if AI could accurately forecast fluctuations in participants’ mental state,” said Igor Matias, a doctoral assistant on the project.
The predictions made by AI were then compared to the responses from the questionnaires and cognitive tests. Remarkably, the AI predictions had an average error rate of just 12.5%, indicating promising potential for using wearable devices in early detection of brain health changes.
Predicting Emotional States
The results showed that emotional states were the easiest for the AI to predict, with error rates between 5% and 10%. Cognitive states were predicted with less accuracy, showing an error margin of 10% to 20%. This means that AI is more effective at deciphering emotional feedback than cognitive assessments.
In terms of the passive data indicators, factors such as air quality, weather conditions, daily heart rate, and sleep variability were found to be the most influential for cognitive performance. For emotional health, weather and sleep conditions, as well as heart rate during sleep, were key indicators.
Supervised by Professors Katarzyna Wac and Matthias Kliegel, this research is part of a longer-term project aimed at studying brain health. The next phase will involve data collection over a 24-month span while identifying individual characteristics that influence the effectiveness of the AI models, enhancing their practical applications in personal health monitoring.
