Reduction of Intraoperative Hypotension during Electrophysiological Procedures: Results based on AI-Guided Non-Invasive Blood Pressure Monitoring

Background:
Intraoperative hypotension (IOH) is common during electrophysiological (EP) procedures. Its etiology is multifactorial and can be related to anthropometric characteristics, amount of sedation, and duration of the procedure. It reflects a deregulation of hemodynamics and may also imply perioperative complications. Several studies have demonstrated an association between IOH with worse clinical outcomes. Traditionally, the invasive arterial line is the gold standard for continuous blood pressure (BP) measurement. However, new, innovative artificial intelligence (AI) - guided devices supply continuous non-invasive BP monitoring and information concerning upcoming IOH.

Aim:
This study investigates the implementation of AI-guided devices for non-invasive BP monitoring during EP procedures, both as bystanders and as assistants, for the prevention and management of IOH.

Methods:
Patients receiving EP were included, and the BP was measured via a non-invasive cuff sensor system. The BP measurements from the AI-guided device were initially observed (group 1) and then taken into account for active medical interventions (group 2), aiming to reduce IOH. A reaction threshold of 60 mmHg for mean arterial pressure (MAP) was set. The medical intervention included volume substitution and the administration of catecholamines. The AI-guided device supplied data regarding MAP every 20 seconds. Information regarding the upcoming IOH was obtained through evaluation of changes in hemodynamics by the integrated AI algorithm of the device. The burden of IOH was calculated using the time-weighted average (TWA) and the area under the curve (AUC) for MAP thresholds of 50 mmHg and 60 mmHg. Through TWA, the average BP during the monitoring period was estimated. The AUC estimated the IOH cumulatively.

Results:
A total of 212 EP procedures have been performed using the AI-guided non-invasive BP monitoring. The device was used as a bystander for 105 EP procedures (group 1) and as a decision-maker guiding tool for 107 (group 2). The median duration of the EP procedures was 115 minutes [IQR 92, 154] for group 1 and 109 minutes [IQR 86, 145] for group 2 (p-value = 0.2). Fewer patients experienced IOH in the intervention group, a result that remains consistent across different hypotension thresholds (MAP < 60 mmHg: 56,19% (59/105) patients in group 1 versus 26,16% (28/107) patients in group 2, p-value < 0.05 and MAP < 50 mmHg: 12,38% (13/105) patients in group 1 versus 0,93% (1/107) patients in group 2, p-value < 0.05). Significantly lower TWA was found in the intervention group for a MAP threshold of 60 mmHg (TWA for MAP < 60 mmHg: median 0.05 [IQR 0, 0.52] for group 1 versus median 0 [IQR 0, 0.01] for group 2, p-value < 0.05). Additionally, lower AUC was also documented in group 2 (AUC for MAP < 60 mmHg: median 6.24 [IQR 0, 62.35] for group 1 versus median 0 [IQR 0, 0.89] for group 2, p-value < 0.05).

Conclusion:

This study investigated the burden of hypotension during EP procedures. The use of AI-guided devices supports real-time, non-invasive decision-making and is associated with a significant reduction in the burden of IOH.