AI for mental health and neuroscience: key insights from leading experts

Anxiety has surged since 2016, impacting people’s ability to work. AI applications can play various roles in promoting mental health, such as detecting and predicting anxiety, supporting interventions, and assisting in decision-making. In his presentation, Simo Levanto, psychologist and advisor at Ilmarinen, emphasizes that "the biggest challenge is not building the tools but maintaining the motivation to follow the advice given and ensuring companies allocate enough resources to support."

Senior scientist Johanna Kallio from VTT presented the results from the Mad@Work EU project where they used real-life computer usage data and self-reported stress perceptions to develop machine learning technologies for assessing stress. “Our case study indicated that perceived stress can be detected automatically from mouse usage data," said Kallio. A tool was created to visualize stress data while preserving privacy, achieving 72% accuracy in daily stress detection. Based on this survey, employees are generally open to digital behavior data collection if privacy is preserved.

The DIGIMIND project tries to understand the significance of psychophysiological signals in identifying mental state and in monitoring treatment effects The focus of the project is on diagnosing anxiety. Current data often lack useful information, necessitating laboratory measurements and experimental manipulations. The goal is to enrich physiological recordings with relevant mental health indicators through AI-based analytics. "Building on existing technological expertise and integrating different data sources will result in useful indicators of mind/mental state," professor and project lead Tiina Parviainen from the University of Jyväskylä summed up the aims of the project.

Hanna Renvall, head of HUS BioMag Laboratory and assistant professor at Aalto University, talked about AI and neuronal activity tracking for predicting dementia risk. Over 10 million Europeans suffer from mild cognitive impairment, with 20 to 50 per cent progressing to dementia. AI combined with neuroimaging can detect early brain hyperactivity, a precursor to amyloid accumulation in the brains of Alzheimer’s patients. The AI-Mind project involves extensive data collection and analysis to predict dementia risk. “We use AI to find those features in EEG and MEG signals that are the most predictive of the brain network disturbance. Then we combine that with other biomarkers, such as genes, blood biomarkers and cognitive tests. As an outcome we get the dementia risk assessment result,” explained Renvall.

Professor Lauri Parkkonen from Aalto University spoke about normative modelling of brain activity using artificial intelligence. Mental health problems cost the EU more than 600 billion euros a year. Normative models using large brain imaging datasets can capture individual variation and help create effective biomarkers for brain diseases, enabling early diagnosis and cost savings in healthcare. "Our research has also been able to show that brain connectivity changes with age. This needs to be taken into account when creating normative models," says Parkkonen. 

These brief highlights from the webinar presentations demonstrate the transformative potential of AI in mental health and neuroscience and highlight the importance of integrating advanced technologies into clinical practice to improve mental wellbeing and diagnostic accuracy.

This news item was originally published on the FCAI website on 6.8.2024