We investigated whether the lymphocyte-to-monocyte ratio (LMR) 48 hours after admission is related to 30-day and long-term mortality in patients with ST-elevation myocardial infarction (STEMI) who were treated with primary percutaneous coronary intervention (PCI). We evaluated 318 consecutive patients with STEMI who were undergoing primary PCI. The relationship between the LMR _48h and all-cause mortality (30-day and long-term) was analyzed by categorizing the patients into tertiles (T) according to LMR _48h —T1 (>2.46), T2 (1.67-2.46), and T3 (<1.67). The T3 group exhibited the highest risk of 30-day all-cause mortality (hazard ratio [HR]: 8.093 [1.006-65.074]; P = .049). For long-term mortality, a significantly higher mortality risk was observed in both T2 (HR: 2.005 [1.021-3.939]; P = .043) and T3 groups (HR: 2.374 [1.160-4.857]; P < .001) compared to the T1 group (reference group). In multivariate analysis, these associations remained unaltered even after adjusting for confounders. A low LMR at 48 hours after admission may be independently associated with both 30-day and long-term mortality in patients with STEMI who were treated with primary PCI. This marker may be used for identifying patients with STEMI at high risk.

The purpose of the study was to investigate whether the addition of left ventricular ejection fraction (LVEF) to the MELD score enhances the prediction of mortality in patients with acute coronary syndrome (ACS) undergoing percutaneous coronary intervention (PCI). This retrospective study analyzed 846 consecutive patients with ACS undergoing PCI who were not receiving previous anticoagulant therapy. The patients were grouped as survivors or non-survivors. The MELD score and LVEF were calculated in all patients. The primary end point was all-cause death during the median follow-up of 28 months. During the follow-up, there were 183 deaths (21.6%). MELD score was significantly higher in non-survivors than survivors (10.1 ± 4.4 vs 7.8 ± 2.4, p < 0.001). LVEF was lower in non-survivors compared with survivors (41.3 ± 11.8% vs. 47.5 ± 10.0%, p < 0.001). In multivariate analysis, both MELD score and LVEF were independent predictors of total mortality. (HR: 1.116, 95%CI: 1.069–1.164, p < 0.001; HR: 0.972, 95%CI: 0.958–0.986, p < 0.001, respectively). The addition of LVEF to MELD score was associated with significant improvement in predicting mortality compared with the MELD score alone (AUC:0.733 vs 0.690, p < 0.05). Also, the combining LVEF with MELD score improved the reclassification (NRI:24.6%, p < 0.001) and integrated discrimination (IDI:0.045, p < 0.001) of patients compared with MELD score alone. Our study demonstrated that the combining LVEF with MELD score may be useful to predict long-term survival in patients with ACS who were undergoing PCI.

The prognostic significance of changes in mean platelet volume (MPV) during hospitalization in ST segment elevation myocardial infarction (STEMI) patients underwent primary percutaneous coronary intervention (pPCI) has not been previously evaluated. The aim of this study was to determine the association of in-hospital changes in MPV and mortality in these patients. Four hundred eighty consecutive STEMI patients were enrolled in this retrospective study. The patients were grouped as survivors (n = 370) or non-survivors (n = 110). MPV at admission, and at 48–72 h was evaluated. Change in MPV (MPV at 48–72 h minus MPV on admission) was defined as ΔMPV. At follow-up, long-term mortality was 23%. The non-survivors had a high ΔMPV than survivors (0.37 (− 0.1–0.89) vs 0.79 (0.30–1.40) fL, p < 0.001). A high ΔMPV was an independent predictor of all cause mortality ((HR: 1.301 [1.070–1.582], p = 0.008). Morever, for long-term mortality, the AUC of a multivariable model that included age, LVEF, Killip class, and history of stroke/TIA was 0.781 (95% CI:0.731–0.832, p < 0.001). When ΔMPV was added to a multivariable model, the AUC was 0.800 (95% CI: 0.750–0.848, z = 2.256, difference p = 0.0241, Fig. 1). Also, the addition of ΔMPV to a multivariable model was associated with a significant net reclassification improvement estimated at 24.5% (p = 0.027) and an integrated discrimination improvement of 0.014 (p = 0.0198). Rising MPV during hospitalization in STEMI patients treated with pPCI was associated with long-term mortality.

Large language models (LLMs) are increasingly evaluated using medical examination datasets, yet most studies emphasize overall accuracy rather than the psychometric structure of test items. We evaluated five LLMs on 199 text-only cardiology residency in-service examination items previously characterized using resident-derived psychometric metrics. Three frontier models were compared with two open-source comparators using a standardized zero-shot, repeated-query protocol and strict-majority scoring. Frontier models achieved substantially higher accuracy than open-source comparators, with Claude Opus 4.6, Gemini 3.1 Flash-Lite, and GPT-5.4 reaching 86.4%, 82.9%, and 81.9%, respectively, compared with 53.3% for MedQwen and 18.6% for Qwen-3.5-35B. Across frontier models, performance increased progressively from hard to easy resident-derived item strata. In multivariable analyses, item difficulty was the only classical psychometric factor consistently associated with AI correctness. IRT-based re-analysis confirmed that higher latent item difficulty was independently associated with lower frontier-model accuracy. Human–AI item-level correlations were modest but exceeded permutation-based null expectations, and frontier-model errors were concentrated among highly ranked human distractors. These findings show that item-level psychometric analysis helps explain variation in frontier LLM performance beyond overall accuracy alone. However, expert-rated rationale assessment and none-of-the-above perturbation testing revealed that strong examination accuracy did not guarantee high-quality explanatory support or reliable recognition of answer absence, indicating that examination performance and answer-selection robustness represent related but distinct dimensions of model behavior.