Unlabelled An exploratory survey of health care-related X (formerly Twitter) users identified perceived challenges in digital professionalism, including exposure to misinformation, collegial disagreements, and unprofessional conduct, while highlighting gaps in institutional guidance and the need for clearer policies and digital-professionalism education.
Abstract Background Triple-negative breast cancer (TNBC) is an aggressive molecular subtype of breast cancer that needs to be identified accurately from pathology reports to inform decision-making. Large language models (LLMs) have been reported to show great potential in the automatic extraction of important oncology information from unstructured medical text. Objective We aimed to create and test an LLM-based approach for automatic biomarker extraction and TNBC subtype classification based on breast cancer pathology reports using prompt-engineering techniques. Methods In total, 532 breast cancer pathology reports from The Cancer Genome Atlas (TCGA) were used. Our approach extracted structured oncology features, such as estrogen receptor status, progesterone receptor status, HER2 status, tumor size, tumor grade, and tumor histology, which were used to identify TNBC. We tested 4 frontier models: GPT-4o mini (released July 18, 2024), GPT-4.1 nano (released April 14, 2025), GPT-5 nano (released August 7, 2025), and Gemini 2.5 Flash-Lite. Results The designed framework was able to extract relevant biomarkers and clinicopathological attributes from unstructured pathology reports and allowed for the automated identification of patients with TNBC. Gemini 2.5 Flash-Lite showed superior performance in the extraction of clinicopathological attributes. For all models investigated, chain-of-thought (CoT) prompting was shown to outperform other strategies, such as zero-shot and few-shot prompting, for both biomarker extraction and TNBC classification. GPT-4o mini using CoT prompts reported the highest performance for TNBC classification. Conclusions Using prompt-engineering techniques based on LLMs proved to be an efficient method of translating unstructured breast cancer pathology reports into structured clinical information. Moreover, it can be used to accurately identify the TNBC subtype.
Abstract Background Delirium is a complex neuropsychiatric syndrome whose severe and lasting consequences in hospitalized patients are largely considered preventable. Prevention primarily relies on the early identification of high-risk patients prior to syndrome onset. However, this remains challenging in clinical practice due to the absence of a diagnostic gold standard and the considerable administrative burden associated with established screening tools such as the 4AT score. Objective The aim of this study is to compare the 4AT score with an established AI-supported prognostic tool in applied clinical practice. Our goal is to demonstrate that shifting from a purely diagnostic approach toward AI-supported delirium risk prediction enables more efficient clinical workflows, reduces misclassification rates, facilitates earlier and more targeted interventions, and ultimately improves long-term patient outcomes. Methods In this prospective study, a total of 2002 patients from a trauma ward were screened and assessed for delirium risk using the 4AT score and the AI-based tool at admission, on the evening of admission, and on the following day, with each assessment predicting the risk of delirium occurrence at any point during the remainder of the hospital stay. The AI tool based its predictions on clinical data automatically derived from electronic health records. The AI-based tool applied a previously developed and validated random forest model (not developed as part of this study). Results Of the 2002 processed patients, the AI prediction tool correctly predicted 87.3% (165/189) of patients with delirium onset as at-risk, compared to an early detection rate of 28.0% (5/189) achieved by the 4AT score. Moreover, the AI-based tool correctly classified 73.9% (1340/1813) patients without delirium as not-at-risk, compared to 96.3% (1746/1813) for the 4AT score. Therefore, the AI-based tool achieved very good delirium predictions, with an area under the receiver operating characteristic curve of 0.884 (95% CI 0.8634‐0.9039). Conclusions The AI-based delirium prognostic tool demonstrated superior early predictive capacity for delirium risk at admission, compared with the 4AT score. The AI-based tool enables proactive risk management and targeted risk reduction.
BACKGROUND Major depressive disorder is a leading cause of disability, and access to specialist psychiatric follow-up remains limited in many settings. Telepsychiatry may reduce logistical barriers to care, but data on video-based follow-up under routine outpatient conditions in Türkiye are scarce, and comparisons of delivery modalities are frequently limited by baseline imbalance between self-selected groups. OBJECTIVE The aim of the study is to describe symptom trajectories and attendance over 6 months of routine psychiatric follow-up delivered either face-to-face or online and to examine whether outcomes differed between modalities after adjustment for baseline severity. METHODS In this prospective, nonrandomized observational cohort study, 120 adults with Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) major depressive disorder were followed for 6 months at a university hospital psychiatry department in Ankara, Türkiye (September 2021 to May 2022). Participants entered face-to-face (n=60) or online (n=60) follow-up according to their own preference. Seven sessions were scheduled at approximately 4-week intervals; the Hamilton Depression Rating Scale (HAM-D; primary outcome), the Hamilton Anxiety Rating Scale, and DSM-5 Level 1 and Level 2 measures were administered at baseline and at sessions 1, 2, 4, and 6. Attendance was recorded at all sessions. Between-group comparisons used analysis of covariance adjusted for the corresponding baseline score and for prespecified covariates. Linear mixed models incorporating a group-by-time interaction were fitted to all available observations, and multiple imputation (m=20) was performed as a sensitivity analysis. RESULTS Depressive symptoms decreased substantially in both groups: mean HAM-D fell from 26.6 to 5.0 in the face-to-face group (81.2% reduction; Cohen d=4.86; P<.001; n=43) and from 24.1 to 3.3 in the online group (86.3% reduction; Cohen d=6.06; P<.001; n=49). Response rates were 41 of 43 (95.3%) and 48 of 49 (98%; P=.60), and remission rates were 32 of 43 (74.4%) and 43 of 49 (87.8%; P=.11). Unadjusted between-group differences at sessions 4 and 6 did not persist after adjustment for baseline severity (session 4: -0.47, 95% CI -1.71 to 0.77; P=.46; session 6: -0.79, 95% CI -2.22 to 0.65; P=.28), and the group-by-time interaction was not significant (χ23=1.9; P=.59). Anxiety scores did not differ at any time point. Findings were unchanged under multiple imputation, and prescribing patterns were comparable. Attendance across all 7 scheduled sessions was higher in the online group overall (373/420, 88.8% vs 319/420, 76%; P<.001), with differences at sessions 2 through 5. CONCLUSIONS Both modalities were followed by large reductions in depressive symptoms, and between-group differences did not persist after adjustment for baseline severity. The study was not designed as an equivalence trial and cannot establish equivalence. The most consistent difference favoring online follow-up was attendance, although its reasons could not be determined. These findings characterize online psychiatric follow-up as a feasible means of maintaining continuity of routine care rather than as a more effective one.
