Psychiatric diagnosis still relies on symptom checklists that were never designed to reflect biology. A peer-reviewed invited review published in Brain Medicine now synthesizes recent advances across four converging domains: conceptual frameworks that move beyond categorical labels, molecular and neurobiological biomarkers, digital phenotyping through smartphones and wearable devices, and machine learning approaches capable of integrating these heterogeneous data streams. The review authors, based at the University of Cambridge, argue that combining objective biological measurement with clinical judgment could yield diagnostic subtypes that predict illness trajectory and guide personalized treatment. They also identify formidable barriers, from data scarcity and algorithmic opacity to regulatory fragmentation and the risk of deepening health inequities.

In their Research review article, “Generative Artificial Intelligence in Medical Imaging: Foundations, Progress, and Clinical Translation,” Hairong Zheng and Shanshan Wang (Paul C. Lauterbur Research Center for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China.) et al. provide a comprehensive overview of recent advances in generative modeling for medical imaging.

Researchers developed a machine-learning workflow that predicts how chemical reactions will form specific “handed” versions of molecules—critical for safe and effective drugs. Trained on small datasets from prior studies, the model screens thousands of reaction components and accurately forecasts outcomes at far lower cost than traditional simulations. By reducing dozens of lab experiments to just a handful, the tool could significantly accelerate and lower the cost of drug discovery and reaction optimization.