Salih Selek
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Salih Selek has a total of 246 co-authors. Here are the top co-authors by number of shared publications:
Haluk A. Savas
14 shared publicationsMahmut Bulut
Dicle University School of Medicine, Department of Psychiatry, Diyarbakir - Turkey
11 shared publications- Mehmet Yumru
Psychiatry Department, Gaziantep University, Medical Faculty, Gaziantep, Turkey
11 shared publications Hasan Herken
6 shared publicationsEsen Savas
5 shared publicationsFeridun Bulbul
Gaziantep University School of Medicine, Department of Psychiatry, Gaziantep - Turkey
5 shared publications
Latest Publications
- Open Access
Evaluating Diagnostic Accuracy and Clinical Reasoning of Multiple Large Language Models in Psychiatry
Summary: Background: Existing large language model (LLM) evaluations rely on accuracy benchmarks that fail to capture whether models reason well while making diagnoses. Studies that do analyse reasoning focus on post hoc explanations accompanying model outputs rather than distinct, clinician-visible artifacts such as detailed reasoning traces. This creates a translational gap in domains such as psychiatry, where diagnosis relies on narrative interpretation, diagnostic reasoning, and clinical judgment under uncertainty. Methods: We conducted a mixed-methods evaluation of four state-of-the-art LLMs using a clinician-curated dataset of 196 psychiatric case vignettes, including 135 published cases and 61 novel clinician-authored vignettes. Diagnostic accuracy was assessed using multiple metrics (top-1 accuracy, top-5 accuracy, recall@5, and mean reciprocal rank) based on ranked lists of five differential diagnoses per vignette. Clinical reasoning quality was evaluated on a randomly selected subset of 30 vignettes through clinician assessment of model-generated diagnostic reasoning traces alongside qualitative commentary from board-certified psychiatrists. We examined the association between clinician-rated reasoning quality and diagnostic correctness and included an illustrative comparison with psychiatry residents. Findings: Clinician-rated diagnostic reasoning quality was strongly associated with diagnostic correctness in mixed-effects logistic regression analyses (β = 1·80; p < 0·001), whereas data extraction quality alone was not. Across the full vignette set, models demonstrated moderate to high diagnostic accuracy. The highest-performing model achieved a top-5 accuracy of 0·801 and also received the highest clinician-rated reasoning scores. In an illustrative comparison, model diagnostic accuracy fell within the range observed for psychiatry residents. Interpretation: Diagnostic reasoning quality captures clinically meaningful variation in LLM performance beyond accuracy metrics. Psychiatry may represent a stringent testbed for evaluating reasoning in narrative-driven clinical domains. Evaluation frameworks for LLM-based clinical decision support should incorporate structured assessment of reasoning processes, not accuracy alone. Research in Context: Evidence before this study: We searched PubMed and Scopus for studies evaluating large language models for psychiatric diagnosis and/or differential diagnosis from text vignettes. Searches were run from database inception to February 6, 2026, using terms including (“large language model” OR LLM “artificial intelligence” OR “generative AI” OR AI OR ChatGPT OR GPT OR Claude OR Gemini OR DeepSeek OR Llama) AND (psychiatr* OR mental OR DSM OR “differential diagnosis” OR diagnos*) AND (vignette OR case OR “case report”). We included empirical studies that evaluated model diagnostic outputs using psychiatric cases/vignettes and excluded editorials/commentaries and studies that did not report case-level diagnostic performance. We did not formally assess risk of bias/quality; studies were heterogeneous in vignette sources, model access, and outcome definitions, precluding quantitative pooling. Prior studies have typically examined small vignette sets, focused on narrow diagnostic domains, evaluated single models, or relied primarily on outcome-based accuracy metrics. When diagnostic reasoning has been assessed, it has usually been inferred from post hoc explanations accompanying model outputs rather than evaluated as a distinct, clinician-visible artifact. Clinician-grounded evaluations of diagnostic reasoning across multiple contemporary models remain limited. Added value of this study: This study provides a large-scale, clinician-grounded evaluation of diagnostic accuracy and diagnostic reasoning quality across four contemporary large language models using a diverse dataset of psychiatric case vignettes. Rather than relying solely on outcome-based explanations, we directly evaluated model-generated diagnostic reasoning traces as clinician-visible artifacts using structured clinician ratings and qualitative analysis. By integrating multiple accuracy metrics with clinician assessment of reasoning coherence, flexibility, and plausibility, we demonstrate that clinician-rated reasoning quality is strongly associated with diagnostic correctness, whereas data extraction quality alone is not. Our analysis also identifies recurrent reasoning failure modes not captured by accuracy metrics, highlighting psychiatry as a stringent testbed for evaluating reasoning in narrative-driven clinical domains. Implications of all the available evidence: Evaluations of large language models for clinical decision support should extend beyond accuracy to include systematic assessment of clinician-visible diagnostic reasoning. Mixed-methods, clinician-grounded evaluation frameworks that examine both diagnostic outcomes and reasoning artifacts may be critical for responsible assessment of LLMs in psychiatry and other areas of medicine where diagnosis depends on interpretation, judgment, and tolerance of uncertainty.
Open AccessEvaluating Diagnostic Accuracy and Clinical Reasoning of Multiple Large Language Models in Psychiatry
Summary: Background: Existing large language model (LLM) evaluations rely on accuracy benchmarks that fail to capture whether models reason well while making diagnoses. Studies that do analyse reasoning focus on post hoc explanations accompanying model outputs rather than distinct, clinician-visible artifacts such as detailed reasoning traces. This creates a translational gap in domains such as psychiatry, where diagnosis relies on narrative interpretation, diagnostic reasoning, and clinical judgment under uncertainty. Methods: We conducted a mixed-methods evaluation of four state-of-the-art LLMs using a clinician-curated dataset of 196 psychiatric case vignettes, including 135 published cases and 61 novel clinician-authored vignettes. Diagnostic accuracy was assessed using multiple metrics (top-1 accuracy, top-5 accuracy, recall@5, and mean reciprocal rank) based on ranked lists of five differential diagnoses per vignette. Clinical reasoning quality was evaluated on a randomly selected subset of 30 vignettes through clinician assessment of model-generated diagnostic reasoning traces alongside qualitative commentary from board-certified psychiatrists. We examined the association between clinician-rated reasoning quality and diagnostic correctness and included an illustrative comparison with psychiatry residents. Findings: Clinician-rated diagnostic reasoning quality was strongly associated with diagnostic correctness in mixed-effects logistic regression analyses (β = 1·80; p < 0·001), whereas data extraction quality alone was not. Across the full vignette set, models demonstrated moderate to high diagnostic accuracy. The highest-performing model achieved a top-5 accuracy of 0·801 and also received the highest clinician-rated reasoning scores. In an illustrative comparison, model diagnostic accuracy fell within the range observed for psychiatry residents. Interpretation: Diagnostic reasoning quality captures clinically meaningful variation in LLM performance beyond accuracy metrics. Psychiatry may represent a stringent testbed for evaluating reasoning in narrative-driven clinical domains. Evaluation frameworks for LLM-based clinical decision support should incorporate structured assessment of reasoning processes, not accuracy alone. Research in Context: Evidence before this study: We searched PubMed and Scopus for studies evaluating large language models for psychiatric diagnosis and/or differential diagnosis from text vignettes. Searches were run from database inception to February 6, 2026, using terms including (“large language model” OR LLM “artificial intelligence” OR “generative AI” OR AI OR ChatGPT OR GPT OR Claude OR Gemini OR DeepSeek OR Llama) AND (psychiatr* OR mental OR DSM OR “differential diagnosis” OR diagnos*) AND (vignette OR case OR “case report”). We included empirical studies that evaluated model diagnostic outputs using psychiatric cases/vignettes and excluded editorials/commentaries and studies that did not report case-level diagnostic performance. We did not formally assess risk of bias/quality; studies were heterogeneous in vignette sources, model access, and outcome definitions, precluding quantitative pooling. Prior studies have typically examined small vignette sets, focused on narrow diagnostic domains, evaluated single models, or relied primarily on outcome-based accuracy metrics. When diagnostic reasoning has been assessed, it has usually been inferred from post hoc explanations accompanying model outputs rather than evaluated as a distinct, clinician-visible artifact. Clinician-grounded evaluations of diagnostic reasoning across multiple contemporary models remain limited. Added value of this study: This study provides a large-scale, clinician-grounded evaluation of diagnostic accuracy and diagnostic reasoning quality across four contemporary large language models using a diverse dataset of psychiatric case vignettes. Rather than relying solely on outcome-based explanations, we directly evaluated model-generated diagnostic reasoning traces as clinician-visible artifacts using structured clinician ratings and qualitative analysis. By integrating multiple accuracy metrics with clinician assessment of reasoning coherence, flexibility, and plausibility, we demonstrate that clinician-rated reasoning quality is strongly associated with diagnostic correctness, whereas data extraction quality alone is not. Our analysis also identifies recurrent reasoning failure modes not captured by accuracy metrics, highlighting psychiatry as a stringent testbed for evaluating reasoning in narrative-driven clinical domains. Implications of all the available evidence: Evaluations of large language models for clinical decision support should extend beyond accuracy to include systematic assessment of clinician-visible diagnostic reasoning. Mixed-methods, clinician-grounded evaluation frameworks that examine both diagnostic outcomes and reasoning artifacts may be critical for responsible assessment of LLMs in psychiatry and other areas of medicine where diagnosis depends on interpretation, judgment, and tolerance of uncertainty.
- 28 April 2022
- Published by Elsevier in Biological Psychiatry
- Vol. 91 (9) , S240
P378. Blue Light Photic Stimulator for Augmentation of Electroconvulsive Therapy
- 28 April 2022
- Published by Elsevier in Biological Psychiatry
- Vol. 91 (9) , S240
P378. Blue Light Photic Stimulator for Augmentation of Electroconvulsive Therapy
- Open Access
- 17 March 2022
- Published by Elsevier in Psychiatry Research Case Reports
- Vol. 1 (1) , 100005
13-Month maintenance ECT treatment in a 66-year-old male patient with pacemaker
- 17 March 2022
- Published by Elsevier in Psychiatry Research Case Reports
- Vol. 1 (1) , 100005
Open Access13-Month maintenance ECT treatment in a 66-year-old male patient with pacemaker
- Open Access
- 4 March 2022
- Published by Wolters Kluwer Health in The Journal of ECT
- Vol. 38 (3) , 159-164
National Network of Depression Centers' Recommendations on Harmonizing Clinical Documentation of Electroconvulsive Therapy
Electroconvulsive therapy (ECT) is a highly therapeutic and cost-effective treatment for severe and/or treatment-resistant major depression. However, because of the varied clinical practices, there is a great deal of heterogeneity in how ECT is delivered and documented. This represents both an opportunity to study how differences in implementation influence clinical outcomes and a challenge for carrying out coordinated quality improvement and research efforts across multiple ECT centers. The National Network of Depression Centers, a consortium of 26+ US academic medical centers of excellence providing care for patients with mood disorders, formed a task group with the goals of promoting best clinical practices for the delivery of ECT and to facilitate large-scale, multisite quality improvement and research to advance more effective and safe use of this treatment modality. The National Network of Depression Centers Task Group on ECT set out to define best practices for harmonizing the clinical documentation of ECT across treatment centers to promote clinical interoperability and facilitate a nationwide collaboration that would enable multisite quality improvement and longitudinal research in real-world settings. This article reports on the work of this effort. It focuses on the use of ECT for major depressive disorder, which accounts for the majority of ECT referrals in most countries. However, most of the recommendations on clinical documentation proposed herein will be applicable to the use of ECT for any of its indications.
- 4 March 2022
- Published by Wolters Kluwer Health in The Journal of ECT
- Vol. 38 (3) , 159-164
Open AccessNational Network of Depression Centers' Recommendations on Harmonizing Clinical Documentation of Electroconvulsive Therapy
Electroconvulsive therapy (ECT) is a highly therapeutic and cost-effective treatment for severe and/or treatment-resistant major depression. However, because of the varied clinical practices, there is a great deal of heterogeneity in how ECT is delivered and documented. This represents both an opportunity to study how differences in implementation influence clinical outcomes and a challenge for carrying out coordinated quality improvement and research efforts across multiple ECT centers. The National Network of Depression Centers, a consortium of 26+ US academic medical centers of excellence providing care for patients with mood disorders, formed a task group with the goals of promoting best clinical practices for the delivery of ECT and to facilitate large-scale, multisite quality improvement and research to advance more effective and safe use of this treatment modality. The National Network of Depression Centers Task Group on ECT set out to define best practices for harmonizing the clinical documentation of ECT across treatment centers to promote clinical interoperability and facilitate a nationwide collaboration that would enable multisite quality improvement and longitudinal research in real-world settings. This article reports on the work of this effort. It focuses on the use of ECT for major depressive disorder, which accounts for the majority of ECT referrals in most countries. However, most of the recommendations on clinical documentation proposed herein will be applicable to the use of ECT for any of its indications.
- Top Cited Papers
- 25 February 2022
- Published by Springer Nature in Psychiatric Quarterly
- Vol. 93 (1) , 249-253
Artificial Intelligence and Chatbots in Psychiatry
The utilization of artificial intelligence (AI) in psychiatry has risen over the past several years to meet the growing need for improved access to mental health solutions. Additionally, shortages of mental health providers during the COVID-19 pandemic have continued to exacerbate the burden of mental illness worldwide. AI applications already in existence include those enabled to assist with psychiatric diagnoses, symptom tracking, disease course prediction, and psychoeducation. Modalities of AI mental health care delivery include availability through the internet, smartphone applications, and digital gaming. Here we review emerging AI-based interventions in the form of chat and therapy bots, specifically conversational applications that teach the user emotional coping mechanisms and provide support for people with communication difficulties, computer generated images of faces that form the basis of avatar therapy, and intelligent animal-like robots with new advances in digital psychiatry. We discuss the implications of incorporating AI chatbots into clinical practice and offer perspectives on how these AI-based interventions will further impact the field of psychiatry.
- 25 February 2022
- Published by Springer Nature in Psychiatric Quarterly
- Vol. 93 (1) , 249-253
Top Cited PapersArtificial Intelligence and Chatbots in Psychiatry
The utilization of artificial intelligence (AI) in psychiatry has risen over the past several years to meet the growing need for improved access to mental health solutions. Additionally, shortages of mental health providers during the COVID-19 pandemic have continued to exacerbate the burden of mental illness worldwide. AI applications already in existence include those enabled to assist with psychiatric diagnoses, symptom tracking, disease course prediction, and psychoeducation. Modalities of AI mental health care delivery include availability through the internet, smartphone applications, and digital gaming. Here we review emerging AI-based interventions in the form of chat and therapy bots, specifically conversational applications that teach the user emotional coping mechanisms and provide support for people with communication difficulties, computer generated images of faces that form the basis of avatar therapy, and intelligent animal-like robots with new advances in digital psychiatry. We discuss the implications of incorporating AI chatbots into clinical practice and offer perspectives on how these AI-based interventions will further impact the field of psychiatry.
- 7 June 2021
- Published by Springer Nature in Psychiatric Quarterly
- Vol. 92 (4) , 1541-1548
A Novel Refractory Mood Disorders and Electroconvulsive Therapy Elective for Medical Students: Fighting Stigma through Experience
Treatment resistant mood disorders (TRMD) have a significant impact on patients and society. Electroconvulsive therapy (ECT) has been shown to be effective for treatment resistant depression (TRD). Despite the effectiveness and safety of ECT, there remains significant stigma surrounding its use. Studies worldwide have shown that many medical students receive their knowledge from the media, which often portrays ECT in a negative light, and very few have exposure to ECT prior to their psychiatric clinical rotations. In this article we highlight the importance of medical education and introduce a novel approach in helping to fight the stigma of ECT through educational intervention for medical trainees that combines active, longitudinal and theoretical learning. By ensuring that trainees have a robust education in this arena, we can help them educate patients about treatment options, improve confidence in prescribing and administering these therapies, and ultimately improve patient and societal outcomes.
- 7 June 2021
- Published by Springer Nature in Psychiatric Quarterly
- Vol. 92 (4) , 1541-1548
A Novel Refractory Mood Disorders and Electroconvulsive Therapy Elective for Medical Students: Fighting Stigma through Experience
Treatment resistant mood disorders (TRMD) have a significant impact on patients and society. Electroconvulsive therapy (ECT) has been shown to be effective for treatment resistant depression (TRD). Despite the effectiveness and safety of ECT, there remains significant stigma surrounding its use. Studies worldwide have shown that many medical students receive their knowledge from the media, which often portrays ECT in a negative light, and very few have exposure to ECT prior to their psychiatric clinical rotations. In this article we highlight the importance of medical education and introduce a novel approach in helping to fight the stigma of ECT through educational intervention for medical trainees that combines active, longitudinal and theoretical learning. By ensuring that trainees have a robust education in this arena, we can help them educate patients about treatment options, improve confidence in prescribing and administering these therapies, and ultimately improve patient and societal outcomes.
- Open Access
- 13 May 2020
- Published by Manuscript Technomedia LLP in Journal of Scientometric Research
- Vol. 9 (1) , 77-81
Comparison of h and m Indices among Departments in McGovern Medical School
- 13 May 2020
- Published by Manuscript Technomedia LLP in Journal of Scientometric Research
- Vol. 9 (1) , 77-81
Open AccessComparison of h and m Indices among Departments in McGovern Medical School
- 1 February 2020
- Published by Elsevier in Journal of Affective Disorders
- Vol. 263, 289-291
Cannabis use: A co-existing condition in first-episode bipolar mania patients
- 1 February 2020
- Published by Elsevier in Journal of Affective Disorders
- Vol. 263, 289-291
Cannabis use: A co-existing condition in first-episode bipolar mania patients
- 22 January 2020
- Published by Wolters Kluwer Health in The Journal of ECT
- Vol. 36 (3) , e19-e21
Image of Electroconvulsive Therapy in YouTube Videos
Objectives Image of electroconvulsive therapy (ECT) in TV shows and movies have been studied before. However, there is no study about image of ECT on YouTube as one of the most commonly used social media platforms for disseminating health information. Our study aims to study the image of ECT in YouTube videos. Methods Key word “Electroconvulsive therapy and ECT” were used to search on Youtube.com and only videos over 50,000 view counts and in English were selected. Above videos were reviewed by an ECT physician and were classified into neutral, negative, and positive groups based on the image of ECT. Results There were 41 YouTube videos selected based on the criteria as stated above, among which 14 were unrelated with ECT treatment, 8 were neutral, 9 were positive, and 10 were reflecting negative image about ECT. There was no significant difference among view counts of the positive, negative and neutral videos (χ2 = 2.746, P = 0.253). Furthermore, the most viewed 3 videos showed negative image of ECT. Additionally, only 6 videos showed the modified ECT and one showed both modified and nonmodified ECT. Conclusions This is the first study conducted on image of ECT in YouTube videos. We found the stigma against ECT may persist in digital video platforms, such as YouTube.
- 22 January 2020
- Published by Wolters Kluwer Health in The Journal of ECT
- Vol. 36 (3) , e19-e21
Image of Electroconvulsive Therapy in YouTube Videos
Objectives Image of electroconvulsive therapy (ECT) in TV shows and movies have been studied before. However, there is no study about image of ECT on YouTube as one of the most commonly used social media platforms for disseminating health information. Our study aims to study the image of ECT in YouTube videos. Methods Key word “Electroconvulsive therapy and ECT” were used to search on Youtube.com and only videos over 50,000 view counts and in English were selected. Above videos were reviewed by an ECT physician and were classified into neutral, negative, and positive groups based on the image of ECT. Results There were 41 YouTube videos selected based on the criteria as stated above, among which 14 were unrelated with ECT treatment, 8 were neutral, 9 were positive, and 10 were reflecting negative image about ECT. There was no significant difference among view counts of the positive, negative and neutral videos (χ2 = 2.746, P = 0.253). Furthermore, the most viewed 3 videos showed negative image of ECT. Additionally, only 6 videos showed the modified ECT and one showed both modified and nonmodified ECT. Conclusions This is the first study conducted on image of ECT in YouTube videos. We found the stigma against ECT may persist in digital video platforms, such as YouTube.
- Open Access
- 31 December 2019
- Published by Online Turk Saglik Bilimleri Dergisi in Online Türk Sağlık Bilimleri Dergisi
- Vol. 4 (4) , 464-475
The Evaluation of Mental Health of Turkish Immigrants Living in Germany
This study aimed to identify if there is a difference between the Turkish people living in Turkey (TPTs) and the Turkish immigrants living in Germany (TIGs) regarding the social anxiety level. It is hypothesized that the TIG group would score higher on the subscales avoidance and anxiety level of the Liebowitz Social Anxiety Scale (LSAS) compared to the other group. For this study, 120 subjects were recruited through the Facebook official website of the Psychiatry department of the Harran University. In this study, participants fill out a questionnaire on sociodemographic properties and score on LSAS. The results of this data displayed a non-significant outcome on subscales, social anxiety (t (n) = 1.15, p = 0.25) and social avoidance (t (n) = 1.67, p = 0.1). It is demonstrated that TIGs had higher scores in the items related to telephoning in public, speaking up at a meeting, expressing a disagreement or disapproval to foreign, giving a report to a group, but these findings were not significant. Further studies with larger sample size are needed.
- 31 December 2019
- Published by Online Turk Saglik Bilimleri Dergisi in Online Türk Sağlık Bilimleri Dergisi
- Vol. 4 (4) , 464-475
Open AccessThe Evaluation of Mental Health of Turkish Immigrants Living in Germany
This study aimed to identify if there is a difference between the Turkish people living in Turkey (TPTs) and the Turkish immigrants living in Germany (TIGs) regarding the social anxiety level. It is hypothesized that the TIG group would score higher on the subscales avoidance and anxiety level of the Liebowitz Social Anxiety Scale (LSAS) compared to the other group. For this study, 120 subjects were recruited through the Facebook official website of the Psychiatry department of the Harran University. In this study, participants fill out a questionnaire on sociodemographic properties and score on LSAS. The results of this data displayed a non-significant outcome on subscales, social anxiety (t (n) = 1.15, p = 0.25) and social avoidance (t (n) = 1.67, p = 0.1). It is demonstrated that TIGs had higher scores in the items related to telephoning in public, speaking up at a meeting, expressing a disagreement or disapproval to foreign, giving a report to a group, but these findings were not significant. Further studies with larger sample size are needed.