Perioperative goal-directed therapy with artificial intelligence to reduce the incidence of intraoperative hypotension and renal failure in patients undergoing lung surgery: A pilot study

Abstract

Study objective: The aim of this study was to investigate whether goal-directed treatment using artificial intelligence, compared to standard care, can reduce the frequency, duration, and severity of intraoperative hypotension in patients undergoing single lung ventilation, with a potential reduction of postoperative acute kidney injury (AKI).

Design: single center, single-blinded randomized controlled trial.

Setting: University hospital operating room.

Patients: 150 patients undergoing lung surgery with single lung ventilation were included.

Interventions: Patients were randomly assigned to two groups: the Intervention group, where a goal-directed therapy based on the Hypotension Prediction Index (HPI) was implemented; the Control group, without a specific hemodynamic protocol.

Measurements: The primary outcome measures include the frequency, duration of intraoperative hypotension, furthermore the Area under MAP 65 and the time-weighted average (TWA) of MAP of 65. Other outcome parameters are the incidence of AKI and myocardial injury after non-cardiac surgery (MINS).

Main results: The number of hypotensive episodes was lower in the intervention group compared to the control group (0 [0-1] vs. 1 [0-2]; p = 0.01), the duration of hypotension was shorter in the intervention group (0 min [0-3.17] vs. 2.33 min [0-7.42]; p = 0.01). The area under the MAP of 65 (0 mmHg * min [0-12] vs. 10.67 mmHg * min [0-44.16]; p < 0.01) and the TWA of MAP of 65 (0 mmHg [0-0.08] vs. 0.07 mmHg [0-0.25]; p < 0.01) were lower in the intervention group. The incidence of postoperative AKI showed no differences between the groups (6.7 % vs.4.2 %; p = 0.72). There was a trend to lower incidence of MINS in the intervention group (17.1 % vs. 31.8 %; p = 0.07). A tendency towards reduced postoperative infection was seen in the intervention group (16.0 % vs. 26.8 %; p = 0.16).

Conclusions: The implementation of a treatment algorithm based on HPI allowed us to decrease the duration and severity of hypotension in patients undergoing lung surgery. It did not result in a significant reduction in the incidence of AKI, however we observed a tendency towards lower incidence of MINS in the intervention group, along with a slight reduction in postoperative infections.

Keywords: Acute kidney injury; Artificial intelligence; Hypotension prediction index; Intraoperative hypotension; Thoracic surgery.