Mustafa Hüseyin Temel

Objectives: Generative artificial intelligence (AI) models capable of producing photorealistic medical images are increasingly proposed for patient education, clinical illustration, and trainee instruction. However, their ability to accurately represent anatomically distinct disease subtypes remains unclear. This study evaluated the diagnostic accuracy of a widely used generative AI model in producing images corresponding to the five anatomical lipedema types defined by the Schmeller classification. Methods: In this prospective audit, ChatGPT’s image-generation interface was prompted to create 60 images for each lipedema type (Types I–V),yielding 300 images. Prompts were standardized and limited to the subtype label without additional descriptors. Two clinicians independently classified each image into one of the five lipedema types or as indeterminate, blinded to the original prompt; disagreements were resolved by a third clinician. Diagnostic performance was assessed using a confusion matrix and per-type sensitivity, specificity, positive predictive value(PPV), negative predictive value (NPV),F1-score,and one-vs-rest receiver operating characteristic area under the curve (ROC AUC). Overall accuracy and Cohen’s κ statistics were also calculated. Results: All 300 images were evaluable. The model generated anatomically consistent images for Types I,II, and III (sensitivity = 1.00 for each). Specificity was 1.00 for Types I and II but 0.50 for Type III because all images requested as Types IV and V were classified as Type III. Consequently, the model failed to generate any images consistent with Type IV(arm-predominant) or Type V(calf-isolated) lipedema (sensitivity = 0.00 for both). Overall accuracy was 0.600. Unweighted and quadratic-weighted Cohen’s κ values were 0.500 and 0.667, respectively. Micro- and macro-averaged ROC AUC were both 0.750. Conclusion: The model reproduces severity gradients within lower-extremity lipedema but systematically collapses anatomically distinct subtypes into the dominant Type III phenotype, failing to depict arm-predominant and calf-isolated disease. Current generative AI systems may therefore encode lipedema as a single visual phenotype rather than a distributed anatomical entity, limiting their reliability for medical education and clinical communication.

Background: This study aims to systematically evaluate the current landscape of artificial intelligence (AI) and machine learning applications in lymphedema research by employing bibliometric and altmetric analyses. The goal is to identify major trends, research focuses, and influential contributors in this rapidly evolving field. Method: A total of 43 AI-related articles on lymphedema published between 1975 and 2025 were retrieved from the Web of Science Core Collection. Bibliometric indicators such as publication years, journals, countries, authorship, and citation metrics were analyzed. Altmetric scores were also assessed. Each study was classified by study type and thematic focus. Result: Original research articles constituted the majority (n = 26), with clinical studies being the most common subtype. The United States and China led in publication output. Most studies were published in Q1-Q2 journals, indicating high scientific quality. Scientific Reports was the most productive journal. General AI applications and risk prediction emerged as dominant themes. A moderate positive correlation was found between average annual citations and altmetric scores (r = 0.470, p = 0.039), suggesting consistency between academic impact and online visibility. Conclusion: This is the first study to comprehensively map AI-based research in the field of lymphedema using bibliometric and altmetric methods. The findings reveal increasing global interest and high-impact publications, particularly in the domains of risk prediction and early diagnosis. These insights may guide future methodological frameworks and interdisciplinary collaborations in this emerging field.

Background: Dual‐energy X‐ray absorptiometry (DXA) is the gold standard for assessing bone mineral density (BMD) and diagnosing osteoporosis (OP). However, variation in DXA scanner measurements may affect clinical interpretation. This study compared variability in BMD and T ‐score among DXA scanners in widespread clinical usage: Stratos DR, Lunar DPX‐ NT and Osteosys Primus. Methods: A retrospective analysis was conducted with 64 patients who underwent BMD measurements using Stratos, Lunar, and Osteosys scanners within 1 year at the same hospital. Paired comparisons of T ‐score and BMD were made between devices. Intra‐class correlation coefficients (ICCs) and a Bland–Altman analysis were used for device reliability and device‐to‐device biases. The McNemar test was used to compare differences in OP and severe OP classifications. Results: The mean age of participants was 65.2 ± 9.83 years (range, 40–83). ICCs for T ‐scores were 0.782 between Stratos and Lunar and 0.791 between Stratos and Osteosys, indicating strong inter‐device reliability. BMD measurements showed ICCs ranging from 0.317 (L1) to 0.956 (femur total). Bland‐Altman plots revealed minor biases between scanners, with a slight positive bias for femoral neck T ‐scores. No significant differences in OP and severe OP diagnoses were found across the scanners. Conclusion: DXA scanners showed substantial concordance in T ‐score and BMD, particularly in femoral measurements. The slight differences between these measurements did not have a major effect on OP diagnosis. Greater variability in lumbar spine results warrants caution. Further studies should aim to minimize inter‐scanner differences and assess their long‐term clinical implications.

Knee osteoarthritis (OA) is a prevalent source of pain and disability. As YouTube becomes increasingly used for health information, the nature and impact of OA-related content warrant systematic evaluation. Using the YouTube Data Application Programming Interface, 327 videos were initially retrieved with the terms “knee osteoarthritis,” “gonarthrosis,” and “degenerative joint disease.” After applying exclusion criteria, the 100 most-viewed English-language videos were selected. Metadata and 60,888 user comments were extracted. Videos were thematically categorized into 6 domains: treatment, disease education, symptoms, diagnosis, patient experience, and uncategorized. Sentiment analysis was performed using the TextBlob library, and statistical trends were assessed via Statistical Package for the Social Sciences (SPSS) 29.0. Treatment was the dominant theme (38%), with nonsurgical options like exercise and rehabilitation comprising over half of these videos (55.3%). Educational (24%) and symptom-related content (12%) were also frequent. Viewer engagement peaked in 2022, with the highest number of views and comments. Over time, video uploads and comment activity increased, best modeled by quadratic trends. Seasonal variation in engagement was not significant. Commonly used words included “exercise,” “knee,” “thank,” and “pain.” Sentiment analysis revealed a decline in positive comments and a rise in neutral sentiment, while negative sentiment remained stable. YouTube serves as a growing platform for knee OA information, emphasizing conservative treatments. Although engagement and content volume have increased, sentiment has shifted toward neutrality. Given the variability in content quality, healthcare providers should guide patients toward credible, evidence-based resources to optimize digital health literacy.

This study aims to evaluate the diagnostic accuracy of ChatGPT-4V, an AI model with visual capabilities, in detecting sacroiliitis on MRI and compares its performance to expert radiologists. This retrospective study included 125 patients (250 sacroiliac joint images) from a tertiary hospital's Picture Archiving and Communication System. MRI scans, including coronal T1-weighted and semicoronal STIR sequences, were assessed by two experienced radiologists. ChatGPT-4V was prompted with standardized queries to analyze the images for signs of active or chronic sacroiliitis. Its diagnostic outputs were compared to the radiologists' assessments. Performance metrics, including sensitivity, specificity, precision, and area under the curve (AUC), were calculated. ChatGPT-4V demonstrated high sensitivity for detecting bone marrow edema (0.955; AUC, 0.84) but lower sensitivity for sclerosis (0.211; AUC, 0.55), joint space narrowing (0.298; AUC, 0.59), and joint surface irregularities (0.433; AUC, 0.59). The overall accuracy of the model was 0.624, with a weighted-average AUC of 0.62. ChatGPT-4V excelled in identifying active inflammatory changes but underperformed in detecting chronic structural abnormalities. ChatGPT-4V shows promise in detecting active inflammatory sacroiliitis, particularly bone marrow edema, but its current inability to reliably identify chronic structural abnormalities limits its standalone clinical utility. To achieve enhanced diagnostic capability and enable clinical integration, future efforts must focus on model fine-tuning using specialist-labeled radiological datasets.