Psychiatric researcher working at the intersection of clinical psychiatry, neurobiology, and artificial intelligence.
Across more than a hundred publications, one question keeps recurring: what happens when you hand a clinical problem to a machine — and what does the machine miss that a clinician wouldn't? Kadir Uludag's work spans schizophrenia biomarkers, tardive dyskinesia genetics, post-earthquake trauma, addiction, and — running through nearly a third of the catalogue — ChatGPT and AI chatbots tested directly against psychological and psychiatric tasks.
The catalogue isn't scattered — it clusters tightly around a handful of concerns, each returned to from a different angle over successive years. Counts and total citations below are drawn directly from the published record.
The largest and most cited thread. Chatbots are treated as test subjects, not tools — interviewed for creativity, probed for plagiarism risk, and evaluated on whether they can distinguish paranoid thought from grandiose delusion or help visually impaired users navigate daily life.
A long-running clinical line on antipsychotic side effects — mapping tardive dyskinesia prevalence, genetic polymorphisms, and superoxide dismutase levels in Chinese patient cohorts, then feeding that data into machine-learning prediction models.
An unusual but persistent sideline: chess as a lens on decision-making, attention, and even criminal-behavior tendencies — plus original chess variants (Crazy Random Chess, Passive Aggressive Chess) designed as psychological instruments in their own right.
Oxidative-stress markers, BDNF levels, and neuroimaging pipelines (Nilearn, NumPy) applied to memory and cognition in schizophrenia — including tutorials on using ChatGPT itself as a coding assistant for imaging analysis.
Mental-health fieldwork following the 2023 Kahramanmaraş earthquake — depression, anxiety, and dissociation in medical students and affected communities, extended to complex PTSD, self-compassion, and the psychology of war and collateral damage.
Genetic and behavioral correlates of heroin dependence, methamphetamine-induced psychosis, smoking in schizophrenia, and newer behavioral addictions — sex addiction and video-recommendation-driven compulsive use.
A reflective, self-aware strand on the research process itself — rethinking the "publish or don't graduate" mandate, transnational leadership in higher education, and the metaverse's role in teaching.
Ranked by citation count as recorded on Google Scholar. The top two — both on AI-supported chatbots in psychology — outpace everything else in the catalogue, a signal of where the field's attention is currently landing.
Publication volume roughly tripled from 2022 to 2023 and has stayed high since, tracking closely with the emergence of ChatGPT as both a research tool and a research subject.
A recurring format: sit ChatGPT down like a subject and test it — for creativity, for diagnostic discrimination between delusion types, for whether it can be trained to "represent" a person after death.
Clinical claims are checked against real cohorts — Chinese schizophrenia patients, Kahramanmaraş earthquake survivors, heroin-dependent populations — rather than resting on the model's output alone.
A parallel stream of practical tutorials — scikit-learn, Nilearn, pandas — shows other researchers how to bring machine learning and brain-imaging analysis into their own psychiatric work.