Can the risk of neurological or mental illnesses be prevented using a smartphone or a smartwatch? A team from the University of Geneva (UNIGE) monitored a cohort of "connected" volunteers and analyzed, using artificial intelligence, data including heart rate, physical activity, sleep, and air pollution.
The results show that these devices can predict, with a low margin of error, the affective and cognitive fluctuations of the participants, paving the way for earlier detection of changes in brain health. This work is published in
npj Digital Medicine.
Pixabay illustration image
Brain health, which combines cognitive and affective functions, is one of the major public health challenges of the 21st century. According to the World Health Organization (WHO), more than one in three people are affected by neurological disorders (such as stroke, epilepsy, or Parkinson's disease, for example) and more than one in two people will be affected by a mental illness (depression, anxiety disorders, schizophrenia) during their lifetime. With an aging population, these numbers are constantly increasing.
Even in healthy adults, brain health frequently varies over time, reflecting interactions between different factors, for example environmental or related to individual lifestyles. Analyzing fluctuations in cognition and affect from day to day or week to week is therefore essential for implementing proactive and personalized prevention strategies.
A UNIGE team wanted to know if wearable and mobile technologies can be used to monitor brain health continuously and non-invasively. For this, 88 volunteers aged 45 to 77 were equipped with a dedicated smartphone application and a smartwatch. Over ten months, these devices collected "passive" data, without intervention or changes to the volunteers' habits, such as heart rate, physical activity, sleep, but also weather and air pollution. In total, 21 indicators were selected.
Every three months, volunteers also had to provide "active" data by filling out questionnaires about their affective state and cognitive performance tests.
Data analyzed by AI
"At the end of the experiment, the passive data were analyzed by an artificial intelligence developed for the study. The goal was to verify if the AI could predict fluctuations in the cognitive and affective health of the participants from this data," explains Igor Matias, doctoral assistant at the Research Institute for Statistics and Information Science of the Geneva School of Economics and Management (GSEM) of UNIGE and first author of the study.
The AI predictions were then compared to the results of the questionnaires and tests. "On average, the error rate was only 12.5%, which opens up new prospects for the use of connected devices in the early detection of abnormalities or changes in brain health," the researcher is pleased to say.
Affective states easier to predict
Affective states proved to be the easiest to predict by artificial intelligence, with error rates mostly between 5% and 10%. Cognitive states, on the other hand, were predicted less accurately, with error rates varying between 10% and 20%. In other words, the AI more effectively predicts the results of affective questionnaires than those of cognitive tests.
Regarding the relevance of passive indicators, air pollution, weather, daily heart rate, and sleep variability are the most informative factors for cognition. For affective states, it is mainly weather, sleep variability, and heart rate during sleep that prove to be the most determining.