Early warning signal in the brain

AI model predicts risk for future weight gain among psychiatric patients

Using an AI model, researchers have found a way to better understand the interaction between mental and metabolic disorders—and possibly predict metabolic issues at an early stage.

To the Point

• Prediction: New model estimates a person’s body mass index (BMI) using brain scans.
• Connection with weight gain: In patients at high risk of psychosis, especially younger patients, the AI’s BMI estimates were linked to future weight gain.
• A role in metabolic diseases: These results suggest that disease-related changes in the brain may contribute to metabolic diseases such as obesity and diabetes.
• Support from AI: This approach could help detect and address metabolic risks sooner in patients with psychiatric conditions.

For several years now, AI-based models have been able to reliably estimate a person's biological age using their brain scans. For patients with schizophrenia, such analyses often show a premature aging effect: Their brains are often estimated to be older than they actually are. This so-called “brain age gap” provides valuable clues about disease-related changes in the brain.

In a study, researchers at the Max Planck Institute of Psychiatry and the Ludwig Maximilian University in Munich now turned their attention to the “BMIgap”: Instead of estimating age, the team used an AI model to estimate body mass index (BMI) from brain scans. To do this, they trained the model with nearly 2,000 scans from healthy individuals and patients with depression, schizophrenia, and an increased risk of psychosis.

Link between mental and metabolic illnesses

The idea behind this: Patients with mental illnesses have a significantly higher risk of developing metabolic illnesses, such as obesity or diabetes. This risk can be explained in part by factors such as smoking, alcohol consumption, or side effects of medication. The team around lead authors Adyasha Khuntia and David Popovic from the Max Planck Institute in Munich now wanted to know whether this risk is also due to disease-related biological changes in the brain.

The results showed clear differences between the patient groups. The AI model underestimated the BMI for people with depression: “The brain structure of these patients was similar to those of healthy people with lower body weight,” explains Popovic. The opposite was true for people with schizophrenia and a high risk of psychosis: Here, the model overestimated the BMI.

“BMIgap” as an early warning signal

Particularly promising was the finding that, especially among younger individuals who are at clinically high risk of psychosis, the “BMIgap” was associated with later weight gain. This means that the “BMIgap” could be used as an early warning signal – a tool for identifying at-risk individuals early and providing them with targeted medical support before serious metabolic diseases develop.

These findings help to better understand the complex relationship between mental illness and physical health. They support the assumption that not only external factors but also disease-related changes in the brain can influence the risk of metabolic diseases in psychiatric patients. This knowledge can form the basis for individualized prevention and treatment of psychiatric disorders.