Minerva Anestesiol. 2026 Mar 23. doi: 10.23736/S0375-9393.25.19432-7. Online ahead of print.
ABSTRACT
BACKGROUND: Intraoperative hypotension (IOH) is associated with increased postoperative morbidity and mortality. This prospective study evaluated the predictive accuracy of the Hypotension Prediction Index (HPI), an artificial intelligence-based algorithm, for predicting IOH during liver transplantation (LT).
METHODS: Twenty-one adult patients undergoing LT for liver cirrhosis and/or hepatocellular carcinoma were enrolled in this study. Under general anesthesia, arterial pressure waveform data were collected using Acumen IQ transducers and transmitted to the HPI monitor for continuous hemodynamic monitoring. Hemodynamic data were recorded at 20-second intervals, resulting in 34,592 data points. Analyses were conducted for the overall cohort and across LT phases (preanhepatic, anhepatic, neohepatic) and stratified using a Model for End-Stage Liver Disease (MELD) score threshold of 15. Predictive performance was assessed using the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity.
RESULTS: HPI demonstrated significant predictive capability, with an overall AUC of 94.2% (95% CI: 94.0-94.5) for IOH prediction within five minutes before hypotensive events. Sensitivity and specificity were 91.7% (95% CI: 91.2-92.3) and 77.5% (95% CI: 76.9-78.0), respectively. Subgroup analyses showed AUCs ranging from 86.0% to 96.3% across different LT phases. In patients with MELD score ≥15, the HPI maintained high predictive accuracy, with an AUC of 94.2% (95% CI: 93.8-94.5).
CONCLUSIONS: The HPI reliably predicts IOH during LT and may support early intervention strategies to mitigate perioperative haemodynamic instability.
PMID:41870962 | DOI:10.23736/S0375-9393.25.19432-7