齋藤 大之

OBJECTIVES: To operationalize and temporally validate an electronic medical record (EMR)-integrated machine learning system (Big data-driven Evaluation of Survival and Treatment in Acute Illness [BEST-AI]) that generates hourly predictions for multiple ICU outcomes, with emphasis on discrimination, calibration, and workflow integration. DESIGN: Single-center hybrid study with stepwise clinical deployment and forward-in-time temporal validation. SETTING: Thirty-bed tertiary mixed medical-surgical ICU in Japan. PATIENTS: All ICU admissions from August 2017 to March 2025. Exclusions: age younger than 16 years or ICU stay less than 4 hours. Development cohort (n = 11,176; from August 2017 to July 2024) and temporal validation cohort (n = 1,127; from August 2024 to March 2025). INTERVENTIONS: EMR-integrated deployment of BEST-AI providing hourly probabilistic predictions to clinicians within the EMR; no protocolized clinical interventions were mandated. MEASUREMENTS AND MAIN RESULTS: Six prediction tasks (in-hospital mortality, ICU mortality, ICU discharge ≤ 72 hr, intubation ≤ 72 hr, extubation ≤ 72 hr, tracheostomy at ICU discharge) were evaluated. In temporal validation, the area under the receiver operating characteristic curves ranged from 0.856 to 0.960, and the area under the precision-recall curves from 0.302 to 0.786. Decile-based calibration showed overall good agreement; hospital mortality was slightly overestimated at higher predicted probabilities, whereas ICU mortality remained well aligned. The intubation task had comparatively lower discrimination and greater deviation from perfect calibration, consistent with low event counts and heterogeneous timing. A 24-hour landmark sensitivity analysis (one prediction per patient at 24 hr after ICU admission) preserved discrimination and calibration relative to the main analysis, supporting robustness beyond repeated-measures evaluation. The system was successfully maintained with automated hourly updates and EMR-embedded patient- and unit-level visualizations, without prescriptive alerts. CONCLUSIONS: A continuously deployed, EMR-integrated ICU prediction system achieved strong temporal discrimination and generally good calibration. Embedding real-time predictions into routine workflow was feasible, and the system was maintained with automated hourly updates. Prospective multicenter studies are warranted to assess transportability and clinical impact.

Background: Intensive care unit (ICU) admission rates after rapid response system (RRS) activation vary widely across institutions. This study examined institutional differences in ICU admission rates and their association with outcomes. Methods: A multicenter retrospective observational study was conducted using a Japanese in-hospital emergency registry, including patients with RRS activation between 2018 and 2022. The ICU admission rate (ICU admissions/RRS activations) and the standardized ICU admission ratio (SIAR; actual/predicted ICU admissions) for each of 35 participating institutions were calculated. The association between SIAR and outcomes was assessed using generalized estimating equation logistic regression with hospital-level clustering. The primary outcome was death within 30 days, and the secondary outcome was a composite of Cerebral Performance Category (CPC) of 3 or higher or death within 30 days. Outcomes were defined as events occurring during hospitalization, within a maximum of 30 days after RRS activation. Results: The study included 8,794 patients. The median ICU admission rate was 0.33 (interquartile range, 0.21 to 0.47), and the median SIAR was 0.98 (interquartile range, 0.75 to 1.17). In univariable analysis, SIAR showed a nonsignificant association with the incidence of death within 30 days (β = –0.05; 95% CI, –0.12 to 0.01; P = 0.108) and a significant negative association with the incidence of CPC of 3 or higher or death within 30 days (β = –0.15; 95% CI, –0.27 to –0.03; P = 0.015). In multivariable analysis, a 0.1-unit increase in SIAR was associated with an odds ratio of 0.98 (95% CI, 0.97 to 0.99; P = 0.104) for death within 30 days and 0.94 (95% CI, 0.92 to 0.96; P < 0.001) for CPC of 3 or higher or death within 30 days. Conclusions: Higher SIAR values were significantly associated with a lower incidence of CPC of 3 or higher or death within 30 days. Greater ICU utilization after RRS activation may improve outcomes, although underlying mechanisms require further study.

To improve severe ketoacidosis with COVID-19, insulin treatment, invasive mechanical ventilation therapy, and continuous hemodiafiltration with sodium bicarbonate infusion were effective.