Diagnostic parity

A diagnostic review from the Department of Chemical Engineering and Biotechnology (CEB) explores how psychiatry could move towards more biologically grounded diagnosis, bringing it closer in line with other areas of medicine.

Published in Brain Medicine, the invited review by Dr Jakub Tomasik, Jihan K. Zaki and Professor Sabine Bahn – of the Cambridge Centre for Neuropsychiatric Research at CEB – synthesises research across four converging domains. These include conceptual frameworks that move beyond current categorical labels, molecular and neurobiological biomarkers, digital phenotyping through smartphones and wearables, and machine learning approaches capable of integrating these diverse data streams.

Today, most psychiatric diagnoses rely on conversations and symptom checklists. While these frameworks provide a shared clinical language, they do not capture the underlying biology of mental illness. People with the same diagnosis can present very differently or respond differently to treatment.

The review highlights several emerging scientific approaches that could help change this. Molecular and neurobiological biomarkers, including genomics, epigenetics, proteomics and metabolomics, are beginning to identify patterns associated with conditions such as schizophrenia, depression and bipolar disorder.

Digital phenotyping using smartphones and wearable devices can track changes in sleep, activity, speech and social behaviour over time, capturing aspects of mental health that static assessments miss. Machine learning models are being developed to integrate these diverse data streams, offering the potential to define diagnostic subtypes that predict illness trajectory and guide personalised treatment.

Professor Sabine Bahn said: “While the DSM and ICD provide an essential framework for psychiatric classification, they fall short of capturing the true nature of mental illness.”

“These systems lack a firm biological basis, yield highly heterogeneous and partly overlapping categories, impose arbitrary thresholds, and rely on subjective judgments that vary across clinicians. Perhaps most critically, diagnostic labels often do not predict prognosis or guide effective treatment.”

The authors note that important challenges remain, including limited data, algorithmic transparency, regulatory differences, and the risk of deepening health inequities. They emphasise that advances in biomarkers, digital phenotyping, and AI should support clinical judgement rather than replace it. Together, these approaches may help move psychiatric diagnosis towards biologically informed frameworks, bringing the field closer in line with how diseases are diagnosed across the rest of medicine.