Minsoo Kim

Background: In-hospital cardiac arrest (IHCA) remains a frequent and devastating event, and conventional early warning scores (EWSs) coupled with rapid response systems (RRSs) have shown inconsistent effects on patient-centered outcomes, largely owing to limited predictive accuracy and high false-alarm burden. AI-based EWSs (AI-EWSs) have demonstrated superior discrimination retrospectively, but high-quality prospective evidence in true external validation environments remains scarce. Existing AI-EWS trials have either been confined to internal-development settings or bundled the algorithm with specific implementation infrastructure, leaving the stand-alone effect of AI-based software as a medical device (AI-SaMD) on patient outcomes uncharacterized. Objective: This trial will evaluate whether implementation of AI-SaMD (VUNO Med -DeepCARS ) as a stand-alone clinical decision-support tool in routine general-ward care reduces IHCA compared with usual care based on conventional EWSs. Methods: The VITAL-MD trial (validation of an AI-based clinical deterioration prediction model on patient outcomes through a stepped-wedge cluster randomized trial [SW-CRT] multicenter evaluation of DeepCARS) is a pragmatic, multicenter, SW-CRT conducted at 4 tertiary academic hospitals in the Republic of Korea. Six tertiary academic hospitals were assessed for eligibility based on inpatient capacity, RRS operating hours aligned with the national RRS pilot program, and no prior AI-SaMD use. All adult patients aged 19 years or older admitted to general wards during the 12-month trial period are included; patients younger than 19 years, or with no documented vital signs throughout hospitalization, are excluded. The 4 sites are randomized to sequentially transition from preimplementation (usual care) to postimplementation (AI-SaMD–guided care) in 4 steps separated by 7-day washout periods. DeepCARS generates real-time risk scores from 4 vital signs and integrates into the existing electronic medical record interface without altering RRS workflows. The primary outcome is the incidence of general-ward IHCA, defined per the Utstein guidelines, among the target cohort (defined as patients exceeding the alert threshold at least once). Assuming a baseline IHCA incidence of 6.70%, a 40% relative risk reduction, an intracluster correlation coefficient of 0.01, and 350 patients per cluster-period, 7002 target-cohort patients (total enrollment approximately 72,695) provide 80% power. Analyses follow the intention-to-treat principle using generalized linear mixed models adjusting for prespecified covariates, with prespecified sensitivity, subgroup, and compliance analyses. Results: The trial was registered with the Clinical Research Information Service (KCT0010243) on February 28, 2025. Recruitment was conducted from April 1, 2025, to March 31, 2026. All 4 sites completed transition to the postimplementation period, and data cleaning is currently underway before the final database lock, with primary results expected in 2027. Conclusions: VITAL-MD will provide the first large-scale, multicenter randomized evidence evaluating whether an AI-EWS, deployed as a stand-alone device in external validation environments and benchmarked against real-world standard-of-care practice, improves patient-centered outcomes for general-ward patients, advancing the evaluation of AI-SaMD from retrospective diagnostic accuracy toward prospective, action-anchored clinical utility.

Background/Objectives: In-hospital cardiac arrest (IHCA) remains a devastating event associated with high morbidity and mortality among general ward patients. While Rapid Response Systems (RRS) can help identify deteriorating patients, maintaining these systems in secondary hospitals is frequently hindered by severe fiscal and personnel constraints. Consequently, evidence regarding the real-world clinical effectiveness of artificial intelligence software as a medical device (AI-SaMD) for predicting deterioration in such resource-constrained settings remains limited. Methods: We conducted a retrospective analysis on a multicenter, staggered-implementation study evaluating 164,761 eligible adult general ward admissions across three secondary hospitals in South Korea. The intervention involved deploying an AI-SaMD (DeepCARS), which utilizes four routine vital signs to predict ward IHCA within 24 h. The primary outcome was ward IHCA. Secondary outcomes included in-hospital mortality and length of stay (LOS). Exploratory analyses investigated the mechanisms of clinical outcomes by evaluating lead-times to interventions, outcomes in sepsis subgroups, changes in care directives, and post-arrest neurological outcomes. Results: AI-SaMD implementation was associated with a 21% reduction in ward IHCA incidence (adjusted rate ratio 0.79; 95% CI, 0.65–0.96; p = 0.016) and a 15% reduction in in-hospital mortality (aRR 0.85; 95% CI, 0.79–0.90; p < 0.001), alongside significantly shorter hospital and intensive care unit LOS. These associations were also observed in patients with sepsis (IHCA aRR 0.71; 95% CI, 0.54–0.93; p = 0.013). Lead-times to critical care intervention and to antibiotic escalation were numerically shorter in the AI-SaMD group by 16.3 h (p = 0.066) and 2.6 h (p = 0.523); poor neurological outcome at discharge among ward IHCA cases was 85/108 (78.7%) in the AI-SaMD group versus 63/102 (61.8%) in the standard-care group (aRR 1.13; 95% CI, 0.99–1.33; p = 0.058); and the full-code death rate did not differ between groups (aRR 0.94; 95% CI, 0.76–1.15)—none of these additional analyses reached statistical significance. Conclusions: In secondary hospitals unable to operate an RRS due to fiscal limitations, implementation of an AI-SaMD as an additional informational layer was associated with lower ward IHCA and in-hospital mortality. The AI-SaMD may serve as an actionable and scalable additional safety layer for general-ward patients in resource-constrained environments where RRS infrastructure is not feasible. Although this was a multicenter, large-scale study, the present analysis was retrospective and quasi-experimental in design; rigorous randomized studies are needed to confirm these associations.