Validation of GPT-4 for clinical event classification: A comparative analysis with ICD codes and human reviewers.

BACKGROUND AND AIM: Effective clinical event classification is essential for clinical research and quality improvement. The validation of artificial intelligence (AI) models like Generative Pre-trained Transformer 4 (GPT-4) for this task and comparison with conventional methods remains unexplored. METHODS: We evaluated the performance of the GPT-4 model for classifying gastrointestinal (GI) bleeding episodes from 200 medical discharge summaries and compared the results with human review and an International Classification of Diseases (ICD) code-based system. The analysis included accuracy, sensitivity, and specificity evaluation, using ground truth determined by physician reviewers. RESULTS: GPT-4 exhibited an accuracy of 94.4% in identifying GI bleeding occurrences, outperforming ICD codes (accuracy 63.5%, P < 0.001). GPT-4's accuracy was either slightly lower or statistically similar to individual human reviewers (Reviewer 1: 98.5%, P < 0.001; Reviewer 2: 90.8%, P = 0.170). For location classification, GPT-4 achieved accuracies of 81.7% and 83.5% for confirmed and probable GI bleeding locations, respectively, with figures that were either slightly lower or comparable with those of human reviewers. GPT-4 was highly efficient, analyzing the dataset in 12.7 min at a cost of 21.2 USD, whereas human reviewers required 8-9 h each. CONCLUSION: Our study indicates GPT-4 offers a reliable, cost-efficient, and faster alternative to current clinical event classification methods, outperforming the conventional ICD coding system and performing comparably to individual expert human reviewers. Its implementation could facilitate more accurate and granular clinical research and quality audits. Future research should explore scalability, prompt and model tuning, and ethical implications of high-performance AI models in clinical data processing.

Acute esophageal stricture after bone marrow transplant.

Esophageal stricture after bone marrow transplantation (BMT) is exceptionally rare, with only a few cases reported in the literature. We present an interesting case of a 58-year-old male with refractory multiple myeloma who developed dysphagia five days following his second bone marrow transplantation. He was found to have a severe esophageal stricture. The patient was treated with multiple esophageal dilations and triamcinolone injections in the following weeks to months, resulting in an improvement in symptoms. Although the exact underlying mechanism remains unknown, high-dose chemotherapy conditioning with melphalan prior to BMT likely contributed to the stricture. Our case highlights the importance of heightened post-bone marrow transplantation management for rare complications, such as an esophageal stricture.

Kindchenschema and cuteness elicit interest in caring for and playing with young children, but less so when children are masked.

Cuteness in the young has long been theorized to elicit care and protection. Most research on this topic has focused on human infants, despite theories suggesting that cuteness may elicit broader social interest that could support learning and development beyond infancy. In four experiments (N = 531 adults, 98 children), we tested whether 'kindchenschema'-facial features associated with cuteness-and perceived cuteness elicit interest in playing with and caring for children, and whether masks disrupt these processes. Participants viewed images of children's faces, masked or unmasked. Kindchenschema correlated with perceived cuteness and age, and these variables predicted adults' interest in playing with and caring for children. Masks did not reduce cuteness ratings or interest in children, although they weakened relations between perceived cuteness and interest, and between perceived age and interest. Cuteness and related signals may guide adults' interactions with children, fostering learning and development.