Can Machine Learning Diagnose Mental Illnesses?

by Kareena Dodeja
Can Machine Learning Diagnose Mental Illnesses?

March 16, 2021

Mental health issues are at an all time high, thanks to the pandemic. Medical aid is available but not to everyone. What if machine learning could lend us a hand and help us identify mental health issues? A team of researchers from the University of Birmingham has cracked the code! Technology could actually help us diagnose mental illnesses.

Region of anterior thalamic radiation that revealed significantly lower axial diffusivity and mean diffusivity values in those with major depressive disorder relative to healthy control participants. From the top left clockwise — coronal, sagittal, 3-D render and axial projections.

Mental Health Diagnosis Conundrum

Our brains are as fascinating as they are mysterious. Mental health clinicians often find it a challenge to make an accurate diagnosis. Normally, the clinicians give a diagnosis based on primary symptoms, but secondary illnesses often take a back seat in such scenarios. For example, clinicians who diagnose psychosis would regard depression as a secondary illness as the focus is on psychosis symptoms such as hallucinations or delusions.

There have been initiatives taken to solve the symptoms of depression through neuromodulation which is an implanted machine that can change the way your brain works. But with patients who have depression and psychosis, it can be challenging to find an accurate diagnosis-which is the primary or secondary symptom.

Recently research was conducted at the University of Birmingham’s Institute for Mental Health and Centre for Human Brain Health. They published their findings in the ‘Schizophrenia Bulletin.’ The researchers worked along the PRONIA consortium to explore the use of machine learning in creating a model where it could investigate the diagnosis of mental health illnesses.

Many indicators, including mortality, disease, health behaviors, and self-rated health, but things are murkier when it comes to mental health.

Depression Or Psychosis- Differentiating it Right

The lead author of the study, Paris Alexandros Lalousis, explained that patients who have co-morbidities like people with psychosis would have depressive symptoms. There is a big challenge for clinicians to diagnose patients and deliver treatments designed for patients without co-morbidity. She explained further that it is not about the misdiagnosis, but diagnostic categories do not accurately reflect the clinical and neurobiological reality.

A European-funded cohort study conducted the PRONIA study across seven European research centres of 300 patients. They went through questionnaire responses, interviews, and data from the structural magnetic resonance imaging of all the patients. They identified a sub-group of patients divided into those who are suffering from psychosis with depression or from depression without psychosis symptoms.

The team took to machine learning models to understand what is ‘pure’ depression and ‘pure’ psychosis. The data proved useful as it helped them apply the models to patients who have both illnesses. They wanted to create an accurate disease profile for every patient and test it out to see if it is accurate.

49997851407 9d4568d415 b

Their results highlighted the need to find out more about mental health. They found out that patients with depression as a primary illness were diagnosed accurately but patients with psychosis had symptoms that tended to depression. Their results indicated that depression plays a bigger role in mental illnesses than what they were taught.

Lalousis mentioned in an interview that there is a pressing need for better treatments for psychosis and depression as they constitute a major role in mental health illnesses. The study highlighted the need for clinicians to study in depth the co-morbid symptoms and the complex neurobiology of the symptoms. The role of depression in illnesses needs to be addressed, as well.

In the study, they used machine learning algorithms to understand the clinical, neurocognitive, and neurobiological factors of mental illnesses. He added that machine learning could become a critical tool for diagnosis in the future. The data-driven diagnosis can someday make mental illnesses keep up with physical health, which is the need of the hour.

The Misdiagnosis Of Bipolar Disorders And Major Depressive Disorders

Bipolar disorders are a huge part of depressive episodes but sometimes it can lead to misdiagnosis as many clinicians cannot diagnose which is a primary or second symptom. According to statistics, 40% of patients who have been diagnosed with bipolar disorder are diagnosed with depression.

The machine learning algorithm Extreme Gradient Boosting (XG Boost) can help distinguish between bipolar and depressive disorder. Artificial intelligence has taken strides in recent years, especially in understanding mental illnesses. A study published in Translational Psychiatry talks about how AI machine learning can distinguish between disorders.

Bipolar Disorder and Major Depressive Disorders can be challenging to diagnose as some misunderstand BD with MDD because of similar symptoms. The patients ranging from 18-45 years-old were analysed in this study. Through Extreme Gradient Boosting (XGBoost), researchers could differentiate the participants with bipolar disorder from a major depressive disorder. Using the machine learning algorithm with blood biomarker data and the questionnaires could help reduce the misdiagnosis.

The evidence shows that machine learning can help diagnose mental illnesses precisely which is important for clinicians. AI has time and again proven to play a vital role in every field and now in mental health. This move could be just what we need to stop the stigma surrounding mental health.


Recommended for you

Leave a Comment

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More