Brazilian Researcher Wins Award in Germany for AI-Based Mental Health Diagnosis

A Brazilian researcher has been honored in Germany for pioneering work using artificial intelligence (AI) to diagnose mental disorders. Francisco Rodrigues, from the University of São Paulo (USP), is leading studies that show AI algorithms can identify psychiatric conditions with over 90% accuracy, using data from laboratory exams.

The research, published in prestigious scientific journals such as Nature and PLOS One, uses magnetic resonance imaging (MRI) to generate data that trains AI systems. These systems can detect brain alterations associated with disorders like epilepsy, autism, and schizophrenia. According to Rodrigues, the technology allows researchers to map affected brain regions and link these changes to specific conditions.

“We are able to identify which brain regions have been altered in a person with epilepsy, autism, or schizophrenia, for example, and understand which changes are related to that specific disorder,”

– Rodrigues

AI as a Diagnostic Tool

Currently, mental health diagnoses rely on patient history and clinical tests, with no definitive biological markers—unlike diseases such as diabetes. Rodrigues envisions a future where a simple brain exam could indicate the presence of depression or other conditions.

While still in its early stages, the technology could prove invaluable for psychologists and psychiatrists, especially in cases with overlapping symptoms or in the early phases of illness.

“With traditional methods, a psychiatrist can’t predict if you’ll develop schizophrenia in ten years. That’s the breakthrough we’re aiming for,”

– Rodrigues

According to Brazil’s 2022 Census, at least 2.4 million Brazilians have been diagnosed with autism spectrum disorder (ASD). Additionally, 1.6 million people aged 15–44 live with schizophrenia, and 1.7 million over 60 suffer from dementia, including Alzheimer’s and Parkinson’s.

Challenges and International Collaboration

The São Paulo-based research uses MRI and electroencephalogram (EEG) data from both healthy individuals and those diagnosed with mental disorders. However, data collection faces challenges: EEGs can be imprecise, and MRIs require patients to remain still for over 40 minutes, limiting participation.

To expand the dataset, Rodrigues has also incorporated data from the United States. “The larger the dataset, the more accurate the algorithm,” he notes.

In January, Rodrigues was among 20 scientists awarded the Friedrich Wilhelm Bessel Prize by the Alexander von Humboldt Foundation in Germany. The award includes €60,000 (about R$370,000) for foreign researchers with significant scientific impact.

Next Steps: Mini-Brains and Future Applications

In Germany, Rodrigues plans to work with “mini-brains”—organoids grown from animal cortical cells—that generate electrical signals captured by chips. While these models don’t fully replicate the human brain’s complexity, they provide valuable data for AI training.

Rodrigues, who holds a degree in physics from USP, has maintained academic ties with Germany since 2006, when he was a visiting student at the Max Planck Institute. In 2011, he began collaborating with Professor Cristiane Thielemann at Aschaffenburg University of Applied Sciences, who nominated him for the award.

By the end of 2026, Rodrigues will relocate to Frankfurt for a year, where he will continue his research and teach courses on complex systems and machine learning at the Humboldt Foundation. He estimates that a general, automated diagnostic method could be available within about a decade, pending regulatory approval from Brazil’s National Health Surveillance Agency (Anvisa).

Source: Globo