Abstract
Background: Survival from in-hospital cardiac arrest caused by ventricular fibrillation (VF) or pulseless ventricular tachycardia (pVT) decreases with every minute of delay before defibrillation. Conventional shock decisions depend on clinician recognition, guideline recall, team coordination, and standard shock-advisory algorithms that may require interruptions in chest compressions. Recent advances in artificial intelligence (AI) have demonstrated high accuracy in rhythm recognition and shock advisory during cardiopulmonary resuscitation (CPR), but their potential impact in Indian tertiary-care settings remains uncertain.
Objective: To estimate the potential effect of an AI-guided defibrillation and cardioversion decision-support system on time-to-first shock and resuscitation outcomes using a fully synthetic simulation model based on a tertiary-care hospital setting.
Methods: A virtual randomized trial comprising 2,000 synthetic adult episodes (1,000 per arm) was constructed. Cases included shockable cardiac arrest (VF/pVT) and unstable tachyarrhythmias requiring synchronized cardioversion. Standard care was compared with an AI-assisted model providing real-time rhythm classification, shock/no-shock recommendations, energy suggestions, and prompts to minimize CPR interruptions while preserving clinician authority. The primary outcome was time-to-first shock. Secondary outcomes included first-shock success, CPR hands-off time, return of spontaneous circulation (ROSC), survival to discharge, neurological outcome proxy, and number of shocks.
Results: Mean time-to-first shock decreased from 5.5±2.3 minutes to 3.7±1.6 minutes with AI assistance (mean reduction 1.8 minutes, p<0.001). First-shock success increased from 62.0% to 78.2%, CPR hands-off time decreased substantially, ROSC improved from 50.3% to 60.3%, and survival to discharge increased from 24.6% to 29.3%.
Conclusion: This simulation suggests that AI-assisted defibrillation and cardioversion may reduce treatment delays, improve CPR quality, and enhance resuscitation outcomes. Prospective clinical studies are required to validate these findings before implementation in practice.
Khurshid Jamadar, Shahin Faruk Shikalgar, Surekha Kisan Suryavanshi, Shrikant Tapare, Yashashri Pore, Shital Pimpalekar, Husain Nadaf, Mangesh Jabade
