Machine Learning-Guided Hypotension Prediction and Goal-Directed Therapy During Transcatheter Valve Repair: A Prospective Observational Study

Objectives: Percutaneous transcatheter valve repair (PTVR) can result in hemodynamic instability, elevated burden of intraoperative hypotension (IOH), and related postoperative complications. The aim of this study was to describe the burden of IOH in patients treated with hemodynamic guidance based on an algorithm that predicts the likelihood of IOH and optimizes cardiac output (Hypotension Prediction Index [HPI]). Design and Setting: Prospective observational study conducted at a single hospital. Participants: Sixty-nine patients undergoing PTVR were included. Interventions: When the HPI exceeded 85, a proactive treatment protocol to prevent hypotension was provided. Measurements and Main Results: The primary outcome was time-weighted average (TWA) mean arterial pressure (MAP) below a threshold of 65 mmHg. Secondary outcomes were number of patients with at least 1 hypotensive event, number of events per patient, depth and duration of hypotensive events, and time spent in hypotension. During an average monitoring time per patient of 191.61 ± 52 minutes, the median TWA-MAP < 65 mmHg was 0.17 mmHg (interquartile range [IQR], 0.01-0.7 mmHg). Two-thirds of the patients experienced IOH, with an average number of hypotensive events of 2.49 (IQR, 0-4) per patient. The median percentage of procedural time spent in hypotension was 4.8%. The majority of hypotensive events occurred after induction of general anesthesia. Conclusions: The main result of this study consisted of safe hemodynamic management during PTVR in terms of limiting the incidence and severity of IOH. This is the first detailed description of characteristics of IOH in patients undergoing PTVR under general anesthesia following the application of HPI-based hemodynamic guidance.