November 12, 2020

AI-Powered Approach to Maintain Stable Vital Signs During Cesarean Delivery

11 – 11:45am

Optimal maternal blood pressure control is important for the well-being of the mother and baby. Using artificial intelligence (AI) methods, we created an algorithm able to predict the blood pressure and determine the medication (phenylephrine) doses needed to maintain stability of the vitals signs in healthy pregnant women presenting for C-section. A cohort of 1400 patients’ records was used to train the model. Additional 300 retrospective records demonstrated that the model was able to predict SBP<90 mmHg with sensitivity 0.997 and specificity 0.965. We developed a software application for clinical use that generates predictions for the blood pressure and phenylephrine dose every minute. Due to the inherent transparent nature of our methods, we can demonstrate the decision making of the algorithm and adjust a range of parameters to empower the physician to take care of patients even with conditions that were not present in the training cohort. We envision that our model can help personalize and automate the titration of drug infusions. This will increase safety, decrease cost of care and, ultimately, transform the way we perform anesthesia.

Presenters: Vesela Kovacheva, MD, PhD and Raphael Cohen