Leo Sher, M.D. 

Today, we publish an interview with an eminent European psychiatrist, Dr. Zoltan Rihmer.

Zoltan Rihmer, M.D., Ph.D., D.Sc. is Professor of Psychiatry at the Department of Psychiatry and Psychotherapy, Semmelweis University, Budapest, Hungary. His main clinical and research interest is clinical, biological and psycho-social aspects of mood and anxiety disorders and prediction and prevention of suicide. Professor Rihmer is a member of many Hungarian and international scientific organizations, and the recipient of several Hungarian and international scientific awards.

Leo Sher: Do you believe biomarkers or pharmacogenomics will meaningfully change everyday prescribing practices in psychiatry, or are their clinical benefits currently overstated?

Zoltan Rihmer: I am quite skeptical in this regard, and I truly believe that the role of biomarkers, and pharmacogenomics in particular, is currently overrated. Highly sophisticated genetic analyses often examine inappropriate or poorly described clinical phenotypes. Let me illustrate this problem using the example of “depression.” We know that around 20–30 percent of DSM “unipolar” MDD patients exhibit current and/or lifetime clinically significant subthreshold hypomanic symptoms without having had a prior hypomanic or manic episode. These patients are officially diagnosed with “unipolar” MDD. However, family history, drug response, and long-term course (e.g., drug-induced or spontaneous hypomania or mania during follow-up) clearly indicate that these patients do not have “unipolar” (i.e., non-bipolar) MDD, but rather that their disorder falls within the bipolar spectrum. Nevertheless, these patients are routinely considered to have “unipolar” MDD, which compromises the validity of the findings and conclusions of pharmacogenetic studies.

Another example is that affective temperaments, which are genetically and clinically milder (“embryonal” or subaffective) manifestations of major unipolar and bipolar mood disorders, are also not considered relevant subgroups in pharmacogenetic and biomarker studies. However, this distinction is important, as a major depressive episode developing in a patient with hyperthymic or cyclothymic affective temperament indicates bipolar spectrum depression rather than unipolar MDD.

This is important because not only the genetic background, but also the pharmacological treatment of bipolar (including bipolar spectrum) disorders and “unipolar” MDD differs fundamentally. Of course, the use of DSM diagnostic categories is necessary, but it is not sufficient. Taking into account the latest psychopathological findings, beyond DSM categories, will be crucial in future pharmacogenomic studies.

Leo Sher: How will the development of artificial intelligence affect psychiatric practice and research?

Zoltan Rihmer: I believe that artificial intelligence (AI) will not replace psychiatrists, but it will significantly change what it means to be a psychiatrist. In clinical practice, AI will mainly function as a decision-support system: improving diagnostic accuracy, supporting risk prediction (such as suicide risk, risk of relapse, or hospitalization), and helping to tailor treatments more precisely to individual patients. It may also reduce administrative burdens and assist with documentation and monitoring, thereby freeing up more time for direct patient care. At the same time, the core of psychiatry—the therapeutic relationship, empathy, ethical responsibility, and the understanding of subjective experience and meaning—cannot be automated and will remain fundamentally human.

In research, AI is likely to be even more transformative. It will allow us to integrate and analyze large, complex, multimodal datasets (neuroimaging, genetics, clinical records, language data, and digital phenotyping) and to detect patterns that are currently beyond human analytical capacity.

AI also has important implications for education and training. Future psychiatrists will need a new kind of “AI literacy”: understanding what these systems can and cannot do, how to critically evaluate their outputs, and how to use them safely and ethically in clinical practice. Learning to work with AI will become part of professional competence, just like learning psychopharmacology or psychotherapy.

Leo Sher: What advice would you offer to young psychiatrists and researchers who wish to make a meaningful and lasting contribution to the field?

Zoltan Rihmer: I believe that the first step is to acquire solid clinical experience and knowledge. Therefore, I recommend that before embarking on a scientific career, one should spend several years in clinical practice. It is difficult to become a successful young researcher without spending several years working with patients and gaining experience in diagnosis, treatment, and follow-up care. Clinical practice and experience are the best sources for generating meaningful research questions.

Of course, you will need to read extensively, and it is worthwhile to examine contradictory findings in the literature as a potential research topic, or to focus on questions where there appears to be a discrepancy between everyday clinical experience and what is reported in the literature.