Psychiatrists often face a frustrating truth: half of all patients with major depressive disorder don’t respond to their first prescribed treatment. In many cases, it can take months of trial and error to find a therapy that works. But artificial intelligence may soon help change that.
Precision psychiatry in practice
Recent studies are showing how AI can guide antidepressant selection based on patient-specific data, ranging from brain scans and genetic profiles to electronic health records and smartphone data. Tools trained on these datasets are already predicting which medications, like SSRIs or bupropion, may work best for an individual patient, reducing reliance on broad clinical guidelines alone.
Clinical examples are emerging
In one study published in *npj Digital Medicine* researchers trained a deep learning model on more than 100,000 patient records. The AI was able to predict antidepressant response with up to 74% accuracy, significantly better than chance. Another study used EEG data to forecast outcomes with specific drugs, while a third integrated fMRI, EEG, and genetic data to guide choices in neuromodulation therapies like transcranial stimulation.
From experimental to practical use
In Brazil, real-world trials are now testing how AI can be embedded in psychiatric workflows. One ongoing collaboration between universities in Brazil, Canada, and the UK is training an AI platform using clinical data from over 600 patients. The goal: to personalize antidepressant selection, reduce treatment dropout, and minimize side effects.
The role of digital biomarkers
AI models are also drawing on less traditional inputs, like sleep patterns, movement data, or social behavior collected passively from wearables and phones. While still early-stage, these ‘digital phenotypes’ are showing promise in predicting mood swings or early response to treatment. A Brazilian project, SetembroBR, is even training classifiers on public Twitter posts to detect depression risk at scale.
Ethical oversight still matters
To responsibly deploy these tools, safeguards around data privacy, informed consent, and explainability are critical. Regulators in Brazil and elsewhere are now starting to assess how these models should be governed in public health settings.
Conclusion
AI won’t replace psychiatrists. But it can make them more effective, helping match the right treatment to the right patient faster. The future of psychiatry isn’t robotic. It’s just a little more precise.
Photo credit: martin-dm, Getty Images
Dr. Antônio Geraldo da Silva is a psychiatrist, researcher, and president of the Brazilian Psychiatric Association (ABP). He is a nationally recognized voice in Brazil’s mental health field and a leading advocate for ethical innovation and equitable access to care. Dr. Geraldo has co-led several pioneering initiatives that apply artificial intelligence to psychiatric diagnosis and treatment personalization. He is also co-authoring a forthcoming book on the practical use of AI in mental healthcare.
Anderson Gobbi is a senior leader at Microsoft and co-founder of XPDoctor, a Brazilian healthtech company focused on digital mental health innovation. With over 20 years of experience in technology, business continuity, and AI deployment, he leads strategic initiatives to integrate clinical decision support and data science in psychiatry. Anderson also collaborates with academic institutions to promote responsible AI adoption in healthcare across Latin America.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.
