Understanding Schizophrenia as We Enter an AI Era: The Clinician Knows Best?

SPECIAL REPORT: SCHIZOPHRENIA

My esteemed colleagues published an authoritative series of articles that summarized the state of play in schizophrenia research, with the intriguing and provocative title of “Schizophrenia: Just the facts.”¹ This learned group scanned the literature (perhaps a human pre–artificial intelligence [AI] equivalent) to determine clear signs of truths with respect to the neurobiology and treatment of schizophrenia. Because these senior leaders have such a commanding knowledge of each aspect of schizophrenia research and treatment, this powerful (yet pithy) series of articles—now spanning decades—was very much appreciated by me and others in providing an anchor for our understanding of this enigmatic condition.

Additionally, a veritable cottage industry of meta-analytic reviews on discrete aspects of schizophrenia also helped to provide another lens to schizophrenia research in asking “is this effect present,” “is it reproducible across multiple studies,” and “how impactful (large or small) is this effect?”² Meta-analyses are of course only as informative as the size and methodological rigor of each of the component studies.

So what are the facts? Schizophrenia is a remarkable, heterogeneous condition (a collection of conditions?) with protean clinical manifestations (and large overlap with features of mood disorders) that are shaped into diagnostic elements that aid clinicians in their ability to diagnose and treat patients with psychoses. Curiously (and reproducibly), individuals with schizophrenia are more likely to be born in the first 3 months of the year (the so-called “season of birth” effect) and more likely to have experienced some birth complications.³ The condition runs in families, although how this occurs is uncertain. Although florid psychosis occurs, people with schizophrenia might show even more attenuated features of their psychosis and some decline in function that manifests in advance of their first psychotic break. Also, whether it is at first episode of psychosis or later, their brains are damaged as evidenced by neurotransmitter alterations (of which dopamine dysregulation is the most frequent and the most pronounced) and by alterations in brain size at both overall size (about 6% smaller than normal) and selective regional reductions in brain tissue. Antipsychotic medications help tone down symptoms and improve overall functioning—especially if taken consistently (medication nonadherence is a major factor in undermining therapeutic efficiency).

Presently, there are multiple potential mechanisms of action of these antipsychotic drugs. Within this generation, this difference is reflected more in the patterns of adverse effects among these drugs than in broader variations in efficacy. Medication can help with symptoms, but comprehensive care helps enable real functional improvements and recovery. Collectively, these facts summarize voluminous and complex literature—AI could help here too, going forward.

The search for biomarkers—reproducible biological signals that can meaningfully foretell diagnosis and/or treatment for schizophrenia—has been a major undertaking with a focus on genetics, immunology, neurochemistry or proteins and neurotransmitters, structural and functional neuroimaging, and now, increasingly, digital phenotyping (sleep analysis, smartphone and smartwatch data analyses, and other behavioral dynamics).

As in other areas of medicine (and as we are rapidly learning, for life in general), AI possesses the capacity to bring all these seemingly disparate strands of information into a coherent whole. The data and scientific challenges are huge. The opportunity is great, and the need is compelling. Will AI usher in a new wave of schizophrenia research that closes the gap between science, which affirms how heterogeneous schizophrenia is, and clinicians, who experience wide differences in how patients with schizophrenia present with their illness(es), in how their course plays out, and in their response and tolerance to antipsychotic medications?

In consideration of these themes, I am grateful for each of the contributions in this Schizophrenia Special Report, as they illuminate key clinical challenges set against the backdrop of our contemporary understanding of schizophrenia. I hope that as you read on, you, too, will find this collection of brief articles as interesting and informative as I have.

Dr Buckley is a psychiatrist and serves as chancellor of the University of Tennessee Health Science Center in Memphis, with health science campuses and affiliated partnerships across Tennessee. He is a member of the Psychiatric Times Editorial Board.

References

1. Keshavan MS, Nasrallah HA, Tandon R. Schizophrenia, “just the facts” 6. Moving ahead with the schizophrenia concept: from the elephant to the mouse. Schizophr Res. 2011;127(1-3):1-13.

2. Wei L, Liu W, Li X, et al. Deciphering the heterogeneity of schizophrenia: a multimodal and multivariate neuroimaging framework for unveiling brain-symptom relationships and underlying subtypes. Schizophr Bull. 2026;52(1):sbaf037.

3. McGrath JJ, Murray RM. Risk factors for schizophrenia: from conception to birth. In: Hirsch SR, Weinberger DR, eds. Schizophrenia. Blackwell Press; 2003.