A Gradient Boosting Regression Model for the Prediction of an Individual's Short Term Blood Pressure

Abstract

Hypertension is a serious problem across the globe because of its mortality rate per year.  High blood pressure (BP) has no warning signs nor symptoms, and measuring the BP level is the only way to know about a person's BP status.  Though there is treatment to help those with hypertension manage the condition, there is a lack of a suitable solution to predict a person's BP based on previous readings and an application which the individuals put in their planned activities to aid in prediction.  This study took a different approach to this problem through the use of artificial intelligence (AI), machine learning (ML) in particular.  An ML model was used to predict future fluctuations of an individual’s BP using their future calendar events.  The study was done in Uasin-Gishu County.  The researcher employed design science and experimental methods for the study.  Rapid Application Development was used in order to design the smartphone application that captured the data from the individuals.  The data for the study was collected using a smartwatch, which collected the BP and heartrate and a smartphone application which collected the mood, activities and calendar events of the individuals.  The Gradient Boosting Regression predictive model was implemented using the Iterative and Incremental Development Model.  The Holdout method’s, test dataset was used along with R-Squared (R2) and Mean Absolute Error (MAE) to evaluate the prototype.  The ML model gave an accuracy score of 99% in predicting an individual’s BP using the future planned activities. From the findings of the study, it is recommended that further studies apply these findings to create custom informative notifications through a phone application, email or Short Message Service (SMS) for each individual in order to prevent BP or even lower BP in case of a hypertensive patient.