Brain scan as BMI oracle for people with mental disorders

With a simple standard MRI scan of the brain, doctors could soon be able to predict which people suffering from mental disorders will put on weight after their initial diagnosis – and thus increase their risk of physical illness – and which will not. “This would allow us to initiate targeted prevention to combat the frequently observed weight gain in these patients,” says Professor Nikolaos Koutsouleris from the Chair of Precision Psychiatry at LMU and the Department of Psychiatry and Psychotherapy at LMU University Hospital. These remarks were occasioned by the results of a new study, in which numerous partners from Germany and abroad were involved. The findings have just appeared in the journal Nature Mental Health.

Almost 18 million people in Germany suffer from a mental disorder: depression above all, but also anxiety disorders, for example, or schizophrenia. It is a little known fact that people with severe mental disorders die on average 10-15 years before the rest of the population. This is largely accounted for by physical ailments, especially cardiovascular diseases, which disproportionately affect people with severe mental disorders. “And so,” explains Koutsouleris, “it makes sense for patients to be conscious of risk factors such as lack of exercise, smoking, and being overweight or obese.”

Speaking of which, experts do not yet know with certainty why so many people with mental disorders put on weight. “In addition to the known side effects of particular drugs, we suspect based on certain medical findings that it is connected to changes in the brain, which in turn are related to the mental disorder,” says Koutsouleris. Could these brain alterations be used as a sort of oracle to predict at the time of initial diagnosis which patients will subsequently experience an increase in body mass index (BMI)?

Step by step to prediction

Step one: To build such an oracle, an international team of researchers started out by creating a machine learning model. That is to say, the scientists fed this artificial intelligence with MRI scans of the brains of healthy people. The model was tasked with using the scans to independently learn to determine the individual weight of the people. “And our algorithm does this really well,” notes the psychiatrist.

Step two: The researchers applied their system to the MRI brain scans of patients with mental disorders. “Here, our prediction model made systematic errors,” explains Koutsouleris. “It miscalculated the weight of the patients in question.” In the case of schizophrenia, for example, it overestimated the weight of patients, because certain brain regions – such as the frontal cortex, where parts of the reward system are located – are smaller than usual in these people.

This system plays a major role in controlling our eating behavior,” adds Koutsouleris, “and our prediction model had previously learned from healthy people that less volume in these brain regions means more body weight.” Although schizophrenia patients have smaller brain volumes upon initial diagnosis, they do not necessarily have a higher body mass index (BMI).

Step three: The researchers tracked the BMI of the patients for a year after their first diagnosis and initial weight estimate: “And then we saw that those patients whom our AI model incorrectly estimated to have a higher BMI had in fact put on a lot of weight.” This was the case particularly for schizophrenia patients, but also for depression patients. As Koutsouleris concludes: “The difference between the estimated and the actually observed BMI, the so-called BMI gap, has predictive power for the further weight development of the patients.”

The utility for patients

This oracle provides an opportunity for the targeted prevention of future weight gain. “We can try to motivate the affected people to live a healthier lifestyle, tell them things like: Why don’t you try a weight loss program, play more sport, eat healthier,” says the psychiatrist. “Or we can prescribe drugs like metformin, which reduce or prevent the risk of metabolic diseases. This would bring a lot of benefits, especially as there are indications that inflammatory processes in the brain are less active when people put on less weight. This would also mean fewer psychiatric symptoms in the course of the illness.”

As soon as the new tool is refined by additional parameters such as the individual genetics of patients or blood values like cholesterol, making it even more precise, the plan is to provide it to all doctors so that they can measure the BMI gap.

Adyasha Khuntia et al.: The BMIgap tool to quantify transdiagnostic brain signatures of current and future weight. Nature mental Health 2025