3 AI Innovations in Diagnosing Mental Illness

3 AI Innovations

Artificial intelligence (AI) is redrawing the boundaries of what we thought was possible in the medical field, at a dizzying speed. In the mental health space, AI innovations are improving how we diagnose and treat mental illness, by enabling earlier, more accurate detection of symptoms. Here are just three examples of this application of AI for three different mental health conditions.

AI Analysis of Brain Scans for ADHD

Attention deficit hyperactivity disorder, characterized by difficulties paying attention, restlessness, trouble sitting still, and impulsivity, occurs in around 5.7 million children and adolescents in the U.S.. When untreated, the disorder can significantly impair a person’s daily functioning. It also is becoming more common due to smartphone use, researchers say.

An early diagnosis can ensure more effective treatment for ADHD. However, the condition often goes undiagnosed or is misdiagnosed, because of subjective self-reporting of symptoms.

This has prompted scientists to look for more objective diagnostic measures of ADHD—and, they seem to have found one. With the help of “deep learning” AI, researchers were able to analyze the brain MRI scans of over 11,000 adolescents with and without ADHD and find “significant differences in nine brain white matter tracts in  the individuals with ADHD,” according to results published November 2023. When used alongside routine questionnaires, these objective markers for ADHD will improve the accuracy of its diagnosis.

AI Language Model for Screening CB-PTSD

“CB-PTSD,” or “Childbirth PTSD,” is post-traumatic stress disorder that results from experiencing serious or life-threatening birth complications. It is surprisingly common, affecting as many as 17 percent of postpartum parents, according to a 2020 study in the Journal of Reproductive and Infant Psychology, cited by the American Psychiatric Association.

Symptoms of CB-PTSD, such as trouble sleeping, hypervigilance, and intrusive thoughts or flashbacks, mirror typical PTSD and can interfere with healthy bonding between a parent and their infant. Diagnosing the condition can also be troublesome because it depends on a clinician’s subjective observations.

AI may change that. When researchers at Massachusetts General Hospital tested several models from OpenAI, including Chat GPT, to see how they analyzed the accounts of women who had just gone through childbirth, the researchers were able to identify a model that provided detailed insights into postpartum maternal mental health. The model can also be incorporated into routine obstetric care to detect not just CB-PTSD but other mental health issues, too, the researchers said.

AI Language Model for Identifying Schizophrenia

Schizophrenia is a severe mental illness that can be characterized by psychotic symptoms like hallucinations and delusions, as well as disorganized speech (and thought) and other behavioral issues that negatively and often significantly impact a person’s daily life. First onset typically occurs between the ages of 16 and 30.

Meanwhile, research has shown that early treatment of schizophrenia can improve outcomes for those who live with the condition, especially when the treatment is long-term, but diagnosis has largely depended on a clinical evaluation and bloodwork.

Now, a new AI-powered tool is inserting more objectivity into the process, thanks to the work of scientists at the University College London’s Institute for Neurology. They explored how the automated analysis of language might help to identify schizophrenia via subtle differences in people’s speech. To test their inquiry, they used an AI model trained to represent huge amounts of internet text as words that have similar meanings to humans.

What the researchers found was that the AI model was indeed able to accurately predict speech patterns when test subjects were administered a verbal test, and that this predictability decreased among test subjects with schizophrenia, a symptom of which is disorganized speech. AI enabled the researchers to catch even subtle differences in this department, so that if a person’s speech patterns fell well outside of the more predictable range, they were a marker of schizophrenia.

ADHD, CB-PTSD, and schizophrenia are just three examples of how AI can be harnessed to improve the diagnosis and treatment of mental health issues. They help to illustrate the rapid evolution in AI technology that is transforming psychiatric care for the better and has only really begun to gain momentum.


To Top

Pin It on Pinterest

Share This