Gembillo, Guido

Introduction: Large language models (LLMs) match or exceed physician accuracy on benchmark diagnostic tasks. Whether higher accuracy reflects human-like diagnostic behavior, which bears on how safely clinicians can supervise these systems, is rarely assessed. Objectives: To determine whether diagnostic accuracy and alignment with human reader responses move together across widely accessible LLMs under default, point-of-care conditions. Methods: This cross-sectional evaluation submitted the 29 most recent New England Journal of Medicine (NEJM) case challenges to 5 publicly accessible LLMs (GPT-5, Claude, Google Gemini, DeepSeek, and Grok) through each model’s web interface (November 5-8, 2025) at default settings, one query per case, no system prompt. Each case offered a clinical vignette and 6 candidate diagnoses; published human reader response distributions (4386-17 714 per case; median, 7732) were the comparator. Diagnostic accuracy against the published gold-standard diagnosis, compared with the human reader majority by the exact McNemar test; agreement by Cohen κ; alignment by a human-error similarity index (the share of model errors on cases the reader majority also missed) and by the Jensen-Shannon divergence (JSD) between each model’s response and the reader distribution; and ordinary least squares regression of per-case JSD on normalized reader-disagreement entropy. Results: Reader majority accuracy was 48.3% (95% CI, 31.4-65.6). All 5 reached or exceeded this value, from 51.7% for DeepSeek (95% CI, 34.4-68.6) to 82.8% for Grok (95% CI, 65.5-92.4); Grok (P = .02), GPT-5 (P = .004), and Gemini (P = .04) significantly exceeded the reader majority, whereas Claude and DeepSeek did not. Grok, the most accurate model, showed the weakest alignment (similarity index, 0.40; mean JSD, 0.50 bits), with 3 of 5 errors on cases readers answered correctly. GPT-5 showed the strongest alignment (index, 1.000), failing only on cases that also challenged readers. For 4 of 5 models, divergence rose with reader-disagreement entropy; DeepSeek was the exception. Conclusions and Relevance: In this evaluation, diagnostic accuracy and alignment with human reader responses diverged across LLMs. Because a model can outperform clinicians in aggregate yet fail where oversight is least likely to detect it, evaluation of clinical AI should weigh error structure alongside accuracy.

Background: Lactic acidosis is a heterogeneous condition with multiple underlying causes and associated outcomes. The use of multi-dimensional patient data to subtype lactic acidosis can personalize patient care. Machine learning consensus clustering may identify lactic acidosis subgroups with unique clinical profiles and outcomes. Methods: We used the Medical Information Mart for Intensive Care III database to abstract electronic medical record data from patients admitted to intensive care units (ICU) in a tertiary care hospital in the United States. We included patients who developed lactic acidosis (defined as serum lactate ≥ 4 mmol/L) within 48 h of ICU admission. We performed consensus clustering analysis based on patient characteristics, comorbidities, vital signs, organ supports, and laboratory data to identify clinically distinct lactic acidosis subgroups. We calculated standardized mean differences to show key subgroup features. We compared outcomes among subgroups. Results: We identified 1919 patients with lactic acidosis. The algorithm revealed three best unique lactic acidosis subgroups based on patient variables. Cluster 1 (n = 554) was characterized by old age, elective admission to cardiac surgery ICU, vasopressor use, mechanical ventilation use, and higher pH and serum bicarbonate. Cluster 2 (n = 815) was characterized by young age, admission to trauma/surgical ICU with higher blood pressure, lower comorbidity burden, lower severity index, and less vasopressor use. Cluster 3 (n = 550) was characterized by admission to medical ICU, history of liver disease and coagulopathy, acute kidney injury, lower blood pressure, higher comorbidity burden, higher severity index, higher serum lactate, and lower pH and serum bicarbonate. Cluster 3 had the worst outcomes, while cluster 1 had the most favorable outcomes in terms of persistent lactic acidosis and mortality. Conclusions: Consensus clustering analysis synthesized the pattern of clinical and laboratory data to reveal clinically distinct lactic acidosis subgroups with different outcomes.

Background: Lactic acidosis is the most common cause of anion gap metabolic acidosis in the intensive care unit (ICU), associated with poor outcomes including mortality. We sought to compare machine learning (ML) approaches versus logistic regression analysis for prediction of mortality in lactic acidosis patients admitted to the ICU. Methods: We used the Medical Information Mart for Intensive Care (MIMIC-III) database to identify ICU adult patients with lactic acidosis (serum lactate ≥4 mmol/L). The outcome of interest was hospital mortality. We developed prediction models using four ML approaches consisting of random forest (RF), decision tree (DT), extreme gradient boosting (XGBoost), artificial neural network (ANN), and statistical modeling with forward stepwise logistic regression using the testing dataset. We then assessed model performance using area under the receiver operating characteristic curve (AUROC), accuracy, precision, error rate, Matthews correlation coefficient (MCC), F1 score, and assessed model calibration using the Brier score, in the independent testing dataset. Results: Of 1919 lactic acidosis ICU patients, 1535 and 384 were included in the training and testing dataset, respectively. Hospital mortality was 30%. RF had the highest AUROC at 0.83, followed by logistic regression 0.81, XGBoost 0.81, ANN 0.79, and DT 0.71. In addition, RF also had the highest accuracy (0.79), MCC (0.45), F1 score (0.56), and lowest error rate (21.4%). The RF model was the most well-calibrated. The Brier score for RF, DT, XGBoost, ANN, and multivariable logistic regression was 0.15, 0.19, 0.18, 0.19, and 0.16, respectively. The RF model outperformed multivariable logistic regression model, SOFA score (AUROC 0.74), SAP II score (AUROC 0.77), and Charlson score (AUROC 0.69). Conclusion: The ML prediction model using RF algorithm provided the highest predictive performance for hospital mortality among ICU patient with lactic acidosis.