Graph Convolutional networks for drug response prediction

Abstract

Improving techniques for assessing hypertension and blood pressure from clinical and physiological data is the main focus of this work. Methods and Supplies: Two groups use the PPG-BP dataset for non-invasive blood pressure prediction: K-Nearest Neighbors (N=10) and Novel Convolutional Neural Network (N=10). There are 500 patients' blood pressure readings in the dataset; 250 are male and 250 are female. The measurements are 905,400 by 875 labels. The findings reveal that when it comes to accuracy, Novel Convolutional Neural Network (73.3980%) beats K-Nearest Neighbors (61.5060%), with a significance value of 0.001 (independent sample t test p<0.05). This demonstrates, statistically speaking, that the two approaches are statistically separate. So, Novel Convolutional Neural Network is better than K-Nearest Neighbors in determining blood pressure from physiological data.