Deep learning has achieved widespread adoption for medical image diagnosis, with extensive research dedicated to mammographic image analysis for breast cancer screening. This study investigates the hypothesis that incorporating region-of-interest (ROI) mask information for individual mammographic images during deep learning can improve the accuracy of benign/malignant diagnoses. Swin Transformer and ConvNeXtV2 deep learning models were used to evaluate their performance on the public VinDr and CDD-CESM datasets. Our approach involved stratifying mammographic images based on the presence or absence of ROI masks, performing independent training and prediction for each subgroup, and subsequently merging the results. Baseline prediction metrics (sensitivity, specificity, F-score, and accuracy) without ROI-stratified separation were the following: VinDr/Swin Transformer (0.00, 1.00, 0.00, 0.85), VinDr/ConvNeXtV2 (0.00, 1.00, 0.00, 0.85), CDD-CESM/Swin Transformer (0.29, 0.68, 0.41, 0.48), and CDD-CESM/ConvNeXtV2 (0.65, 0.65, 0.65, 0.65). Subsequent analysis with ROI-stratified separation demonstrated marked improvements in these metrics: VinDr/Swin Transformer (0.93, 0.87, 0.90, 0.87), VinDr/ConvNeXtV2 (0.90, 0.86, 0.88, 0.87), CDD-CESM/Swin Transformer (0.65, 0.65, 0.65, 0.65), and CDD-CESM/ConvNeXtV2 (0.74, 0.61, 0.67, 0.68). These findings provide compelling evidence that validate our hypothesis and affirm the utility of considering ROI mask information for enhanced diagnostic accuracy in mammography.

Deep Learning (DL) has undergone widespread adoption for medical image analysis and diagnosis. Numerous studies have explored mammographic image analysis for breast cancer screening. For this study, we assessed the hypothesis that stratifying mammography images based on the presence or absence of a corresponding region of interest (ROI) improves classification accuracy for both normal–abnormal and benign–malignant classifications. Our methodology involves independently training models and performing predictions on each subgroup with subsequent integration of the results. We used several DL models, including ResNet, EfficientNet, SwinTransformer, ConvNeXt, and MobileNet. For experimentation, we used the publicly available VinDr., CDD-CESM, and DMID datasets. Our comparison with prediction results obtained without ROI-based stratification demonstrated that the utility of considering ROI presence to enhance diagnostic accuracy in mammography increases along with the data volume. These findings support the usefulness of our stratification approach, particularly as a dataset size grows.

Deep Learning (DL) has undergone widespread adoption for medical image analysis and diagnosis. Numerous studies have explored mammographic image analysis for breast cancer screening. For this study, we assessed the hypothesis that stratifying mammography images based on the presence or absence of a corresponding region of interest (ROI) improves classification accuracy for both normal–abnormal and benign–malignant classifications. Our methodology involves independently training models and performing predictions on each subgroup with subsequent integration of the results. We used several DL models, including ResNet, EfficientNet, SwinTransformer, ConvNeXt, and MobileNet. For experimentation, we used the publicly available VinDr., CDD-CESM, and DMID datasets. Our comparison with prediction results obtained without ROI-based stratification demonstrated that the utility of considering ROI presence to enhance diagnostic accuracy in mammography increases along with the data volume. These findings support the usefulness of our stratification approach, particularly as a dataset’s size grows.

Patient perception involves a patient’s thoughts and beliefs regarding their health status. It is also associated with medical compliance and outcomes. However, discrepancies often arise between patient perception and physicians’ documentation within the medical records, resulting in misunderstanding and suboptimal doctor-patient communication. In this study, we assessed the efficacy of generative artificial intelligence (AI) in comparing the content of patient perception as recorded in patient questionnaires and physicians’ records of the Department of Breast Surgery. We evaluated the precision and recall of the generative AI by comparison with human-created ground truth. Our results demonstrated the high performance of the generative AI in comprehending and contrasting symptoms and the entire content recorded differently by patients and physicians, with F1 scores ranging from 0.77 to 0.97. These results highlight the potential contribution of a generative AI to deeper mutual comprehension in healthcare scenarios.

Age-related macular degeneration, one of the primary causes of blindness, requires effective prediction of mid-term treatment outcomes to support the ongoing administration of the standard therapy. Our previous work relied on specialist interpretation and manual extraction of images, which limits their applicability in broader practical application. To address this limitation, this study aims to develop a deep learning-based framework capable of automatically identifying an optimal “champion” image from a set of candidate images.

Background: A persistent discrepancy exists between patient-reported information and physician documentation. While conversational agents have been developed to collect medical histories prior to consultations, existing evaluations have largely focused on diagnostic accuracy or user satisfaction rather than on the completeness and clinical relevance of the information collected. There remains a need to assess the extent to which clinically relevant information is captured through chatbot-based interviews, and to understand how model configurations and instructional strategies influence this coverage. Objective: This study aimed to evaluate the extent to which a chatbot can obtain clinically useful patient history information, and to examine how prompt detail and internal reasoning influence information coverage during chatbot-based medical interviews. Methods: We developed a medical history-taking chatbot using the Qwen3-14B-Instruct model and evaluated 4 configurations in a 2×2 factorial design: detailed and thinking mode, detailed and nonthinking mode, minimal and thinking mode, and minimal and nonthinking mode. These configurations were compared against a rule-based system baseline (choice mode) using 66 standardized primary care clinical cases, with simulated patients interacting with the chatbot according to predefined case scripts. Information coverage (%) was assessed using a checklist inspired by Objective Structured Clinical Examination (OSCE) frameworks. Three physicians independently evaluated transcript coverage, with interrater agreement assessed using full agreement rates and Fleiss κ. For the 4 LLM configurations, coverage was analyzed using 2-way repeated-measures ANOVA to examine the effects of prompt detail, internal reasoning, and their interaction. All 5 configurations, including the rule-based baseline, were additionally compared using 1-way repeated-measures ANOVA with post hoc 2-tailed paired t tests. Results: Interrater agreement was substantial (Fleiss κ=0.75). Across all 66 simulated cases, information coverage differed significantly among configurations ( P <.001), with the detailed prompt with thinking (detailed and thinking) mode achieving the highest mean coverage (72.3%, SD 14.3%), compared with moderate coverage in configurations using either thinking or detailed prompts alone (approximately 60%) and lower coverage in minimal nonthinking and rule-based configurations (approximately 51%‐54%). In the factorial analysis of the 4 LLM configurations, both prompt detail and internal reasoning were significantly associated with improved information coverage, with a significant interaction between prompt detail and internal reasoning interaction. Mode-related differences were most pronounced for past medical and family history domains. Symptom-level analyses revealed substantial variability, with higher coverage for symptoms associated with well-defined diagnostic frameworks and lower coverage for multisystem presentations. Conclusions: In this controlled, simulated setting, the detailed prompt with a thinking mode achieved the highest overall checklist-based information coverage. The findings suggest that combining structured clinical prompts with internal reasoning may improve the completeness of chatbot-collected patient histories. Further research is needed to evaluate its impact on clinical documentation, workflow integration, and real-world usefulness.