Autoregressive With Exogenous Input (ARX) Decision Support for Blood Pressure Maintenance During Cesarean Delivery Under Spinal Anesthesia: A Prospective Pilot Study With Matched Nonconcurrent Controls

BackgroundSpinal anesthesia for cesarean delivery commonly causes maternal hypotension, which may compromise uteroplacental perfusion and maternal comfort. Guidelines recommend maintaining maternal blood pressure near baseline with prophylactic vasopressor strategies, yet titration remains reactive. We evaluated an autoregressive with exogenous input (ARX) decision-support algorithm that provides real-time forecasts of maternal mean arterial pressure (MAP) to support vasopressor management during cesarean delivery under spinal anesthesia. MethodsIn this single-center, open-label, prospective pilot study, 20 pregnant patients at term undergoing elective cesarean delivery under spinal anesthesia received standard care supplemented by ARX-generated MAP predictions at 1-, 2- and 3-minute horizons. Clinicians titrated phenylephrine per institutional protocol while reviewing ARX predictions, retaining full autonomy for dosing decisions. Predictive performance was quantified using root mean square error (RMSE), mean absolute error (MAE), coefficient of determination (R{superscript 2}), and fraction of improvement in total error (FIT). ARX-guided patients were matched 1:2 to nonconcurrent controls (n = 40) on attending anesthesiologist and intrathecal bupivacaine dose, with nearest-neighbor matching on age and body mass index. Exploratory outcomes included hypotension (MAP <80% of baseline), phenylephrine dose, maternal nausea, and neonatal outcomes. For minute-level hypotension classification performance, sensitivity/specificity (and related metrics) were estimated using generalized estimating equations (GEE) to account for within-patient clustering of repeated observations. ResultsOne-minute-ahead ARX predictions achieved a mean ({+/-}SD) RMSE of 3.71 {+/-} 3.26 mmHg and MAE of 2.75 {+/-} 2.52 mmHg, with R{superscript 2} 0.34 {+/-} 0.63 and FIT 21.1% {+/-} 18.7%. Predictive performance decreased at longer horizons. For hypotension prediction, one-minute-ahead GEE-estimated population-average sensitivity and specificity were 57.39% and 99.74%, respectively. During the observation window, in exploratory comparisons with matched nonconcurrent controls, ARX-guided patients had a shorter duration of hypotension (0.8 {+/-} 1.9 vs 3.0 {+/-} 3.8 minutes; P = .005) and a lower incidence of hypotension (25.0% vs 52.5%; P = .033), but a higher phenylephrine dose (1823 {+/-} 659 vs 974 {+/-} 328 {micro}g; P = .001). Maternal nausea incidence was lower in the ARX group compared with matched nonconcurrent controls (5% vs 35%; P = .014), with similar neonatal outcomes. ConclusionsIn this prospective pilot study, an ARX decision-support algorithm provided accurate 1-minute-ahead MAP forecasts and was associated with higher phenylephrine dosing and shorter maternal hypotension duration compared with matched nonconcurrent controls. These findings support further evaluation in larger, randomized trials. Summary statementIn this prospective pilot study of 20 patients undergoing cesarean delivery under spinal anesthesia, an autoregressive with exogenous input (ARX) decision-support algorithm provided real-time blood pressure forecasts and was associated with a shorter hypotension duration but higher phenylephrine dose compared with matched nonconcurrent controls. These preliminary data support further evaluation of ARX-guided, algorithmic vasopressor management in larger, multicenter trials. Key PointsO_LIQuestion: In pregnant patients at term undergoing elective cesarean delivery under spinal anesthesia, can a real-time ARX algorithm accurately forecast MAP and support vasopressor management? C_LIO_LIFindings: One-minute-ahead forecasts were accurate (RMSE 3.71 mmHg), and ARX-guided care was associated with a shorter duration of hypotension and a higher phenylephrine dose versus matched nonconcurrent controls C_LIO_LIMeaning: Real-time MAP forecasting is feasible and warrants randomized evaluation to confirm clinical benefit and characterize trade-offs. C_LI