Background: Identification of ICU patients at increased risk of death may support clinical assessment and prioritization of care. However, many mortality prediction approaches incorporate extensive laboratory information, conventional severity scores, or computationally complex models that may limit straightforward interpretation at the bedside. This study investigated whether routinely recorded physiological measurements collected during the first 24 h of ICU care could support an interpretable machine learning approach to subsequent in-hospital mortality prediction. Methods: We performed a retrospective analysis of 51,598 adult first ICU admissions from MIMIC-IV using a strict 24 h landmark design. Patients with a missing ICU duration, a documented ICU stay shorter than 24 h, or a documented death occurring on or before the 24 h landmark were excluded. Physiological measurements recorded during the first 24 h were summarized to construct the predictor set. Five classification algorithms—Logistic Regression, Random Forest, XGBoost, LightGBM, and CatBoost—were developed using a stratified 80:20 train–test split. Performance was evaluated using discrimination, precision–recall, calibration, threshold-dependent classification measures, and decision-curve analysis. Bootstrap resampling was used to estimate 95% confidence intervals and pairwise performance differences. Shapley Additive Explanations (SHAPs) were used to interpret the XGBoost model. Results: The revised cohort included 46,259 survivors and 5339 patients with an in-hospital mortality outcome. XGBoost achieved the numerically highest AUROC of 0.803 (95% CI: 0.790–0.815) and AUPRC of 0.341 (95% CI: 0.313–0.369), with a Brier score of 0.080 (95% CI: 0.076–0.084). Paired bootstrap comparisons showed that XGBoost had a significantly higher AUROC than Random Forest and LightGBM, but not CatBoost. Its AUPRC was significantly higher than that of Random Forest but did not differ significantly from LightGBM or CatBoost, while its Brier score was significantly lower than that of Random Forest but did not differ significantly from LightGBM or CatBoost. At the default probability threshold of 0.50, XGBoost achieved high specificity (0.990) but low recall (0.091), whereas lowering the threshold increased mortality detection at the cost of additional false-positive classifications. Conclusions: Routinely recorded physiological information accumulated during the first 24 h of ICU care supported an interpretable machine learning framework with useful discrimination and probability estimation for subsequent in-hospital mortality. XGBoost achieved the highest numerical performance, although differences among the gradient-boosting models were generally small and not significant for all evaluated metrics. Threshold-sensitivity analysis demonstrated that classification behavior depended strongly on the selected operating threshold. Independent multicenter validation and prospective clinical evaluation are required before routine deployment.

Lee S, et al. Digit Health Res. 2025 Mar;3(1):e1. https://doi.org/10.61499/dhr.2025.3.e1

Falls in hospital wards remain a significant safety concern, and conventional monitoring relying on intermittent observations may delay escalation. We developed a ceiling-mounted, top-view fisheye framework integrating a Panoptic-DeepLab-inspired, multi-task model within real-time ward operations. The network employs an EfficientNetV2-S encoder and bidirectional feature pyramid network decoders with parallel heads for patient detection, eight-class posture recognition, and seven-class human–object interaction classification, supported by auxiliary depth and normal supervision during training. We analyzed a de-identified, non-public actor-staged dataset collected at hospital and hospital-like sites (nine participants, 40 scripted actions) with fine-grained annotations, including skeletons, postures, interactions, and object polygons. Evaluation used leave-one-site-out cross-validation across the recording sites. The proposed system achieved a mean fall-positive precision of 0.9829 ± 0.0297, recall of 0.9219 ± 0.2040, and F1-score of 0.9343 ± 0.1766. During real-world deployment across seven wards over 1,450 monitoring days, the system recorded 325 true falls and 695 false alarms, corresponding to an average false-alarm frequency of 0.48 per monitoring day. Seated falls occurred as frequently as recumbent falls, and over 20,000 non-alarming “fall cautions” reflected the prevalence of bed-edge dynamics in routine care. These findings support the feasibility of context-aware, ceiling-mounted fisheye monitoring for room-scale fall surveillance; however, broader prospective validation remains necessary to establish generalizability across routine clinical environments.