Transforming mental health assessment with Artificial Intelligence

mental health assessment
© iStock/Chinnapong

A pioneering study has revealed that Artificial Intelligence (AI) can achieve a high-quality mental health assessment, potentially enhancing the future of diagnostics for mental health disorders.

The research, led by Denis Engemann from Inria, illuminated that machine learning from substantial population cohorts can yield “proxy measures” for brain-related conditions without requiring a mental health assessment from a specialist. The researchers employed the UK Biobank for their study, one of the most comprehensive and sizeable biomedical databases on the planet, comprised of extensive and health-related data on the UK population.

The study’s findings are published in the journal GigaScience.

The mental health crisis

Around the world, mental health issues have been steadily increasing, with the World Health Organization (WHO) estimating that between 2007 and 2017, mental health conditions and substance abuse disorders have risen by 13%. The knock-on effects are considerable, impacting society in nearly all aspects of life, such as school, work, family, friends, and community engagement.

One of the most paramount issues in accurately and efficiently addressing these disorders is that a traditional mental health assessment requires an expert in the field, which is problematic due to the small proportion of them globally in comparison to mental health sufferers. Now, the development of a mental health assessment that utilises AI looks to provide an accessible means of detecting, preventing, and treating such health issues.

Machine learning diagnostics

To produce their AI models, the Inria team and their colleagues’ employed data from the UK biobank, which in addition to biological and medical datasets, includes questionnaire data about personal circumstances and habits, such as age, education, tobacco and alcohol use, sleep duration, and physical exercise. For this specific investigation, the questionnaires also included sociodemographic and behavioural data, including moods and sentiments of the individual; the biological data also included Magnetic Resonance (MR) images of 10,000 participants’ brain scans.

The team amalgamated the data sources to create models that approximate measures for brain age and subsequently defined intelligence and neuroticism traits scientifically. They act as “proxy measures,” which are indirect measurements that correlate significantly with specific mental health conditions or outcomes that cannot be measured directly. This method of producing approximations has been employed successfully in the past to predict brain age from MR images. The researchers validated their proxy measures by demonstrating the same results in a separate subset of UK Biobank data.

Engemann said: “In this work, we generalised this methodology in two ways. First, we demonstrated that, beyond biological ageing, the same proxy measure framework is applicable to constructs more directly related to mental health. Second, we showed that useful proxy measures can be derived from other inputs than brain images, such as sociodemographic and behavioural data.”

The results of the research signify that AI and specialists can work together to achieve an accurate and personalised mental health assessment. Although human interaction is still a crucial aspect of a mental health assessment, an individual could allow a machine learning model to securely access their social media account to attain proxy measures that can be useful to the client and their mental health professional.

Engemann commented: “What is not going to change is that mental health practitioners will need to carefully interpret and contextualise test results on a case-by-case basis and through social interaction, whether they are obtained using machine learning or classical testing.”

LEAVE A REPLY

Please enter your comment!
Please enter your name here