Point-of-Care AI Assistance and Critical Care Outcomes: A Randomized Trial

Acute Kidney Injury, Acute Respiratory Failure (ARF), Critical Illness, Delirium Confusional State, Multi-organ Failure, Sepsis, Shock

Critical Care, Intensive Care Unit, Large Language Model, Artificial Intelligence, Diagnostic Accuracy, Clinical Decision Support, Critical Care Outcomes, Sepsis, Shock, Acute Respiratory Failure, Multiorgan Failure

This is a prospective, unmasked, randomized, multicenter clinical trial evaluating the impact of point-of-care large language model (LLM)-based decision support on diagnostic accuracy and clinical outcomes in adult medical intensive care unit (MICU) patients. Consecutive adult ICU admissions at participating community hospitals (initially MetroWest Medical Center and St. Vincent Hospital) will be screened for eligibility. Eligible patients will be randomized 1:1 to standard care or an AI-assisted group. In both arms, initial evaluation and management will follow usual practice. For patients randomized to AI assistance, de-identified admission data (history and physical, labs, imaging reports, and other relevant documentation) will be formatted and submitted to a state-of-the-art LLM (ChatGPT-5) at the time of admission. The AI-generated differential diagnosis and therapeutic recommendations will be provided to the admitting team for consideration. For the standard care arm, LLM output will be generated but not shared with clinicians. After discharge, a masked chart review will determine the ground truth primary diagnosis and extract outcomes including: Primary Outcome - a composite of medical errors (from time of ICU admission through day 7 of ICU stay, or ICU discharge, whichever comes first); Secondary Outcomes - 90-day mortality, ICU and hospital length of stay, and ventilator-free days.

The rapid development of large language models (LLMs) such as ChatGPT has created new opportunities and risks for their use in medicine. Although early studies suggest high diagnostic accuracy in complex clinical scenarios and ICU admissions, the impact of LLMs on real-world clinical outcomes and the optimal mode of physician-AI interaction remain uncertain. Published work from our group showed that ChatGPT-4 achieved diagnostic accuracy comparable to board-certified intensivists for ICU admissions in a retrospective study. However, prospective, randomized data on clinical outcomes are lacking. This trial will evaluate a pragmatic paradigm for integrating LLMs at the time of ICU admission (point-of-care AI). All eligible adult MICU admissions at participating sites will be prospectively randomized to: (1) standard care, or (2) AI-assisted care in which an LLM receives standardized, de-identified admission data and returns a proposed primary diagnosis, ranked differential diagnosis (up to five conditions), suggested additional information, and prioritized therapeutic interventions. Admitting clinicians in the AI-assisted arm will be asked to review and optionally incorporate the AI recommendations and will complete a brief questionnaire regarding perceived utility and any changes in diagnosis or management. A masked clinical adjudication panel will perform longitudinal chart review to define the ground truth primary diagnosis and assess error rates and outcomes. The primary endpoint is a composite of medical errors. The specific time frame will be from the time of ICU admission through day 7 of ICU stay, or ICU discharge, whichever comes first. Secondary endpoints will include 90-day mortality, ICU and hospital length of stay, and ventilator-free days. Other exploratory secondary endpoints will be considered. The trial is designed to enroll approximately 1000 patients across multiple MICUs, with interim analysis at 12 months to assess feasibility, integrity, and futility. The study is minimal risk, uses de-identified data for AI queries, and does not alter standard diagnostic testing or therapeutic options.

No related papers found yet.

United States