Low Cost Design to Support Health Care Needs of Underserved Populations: Using SMS and AI To Engage At-Risk Populations

Abstract

This thesis focuses on building a system for underserved populations to manage their health by interacting through the short messaging system (SMS) of mobile devices. The proposed system is novel in that it does not rely on telecommunication data plans, which can be expensive and not affordable by underserved populations. Additionally, the system will afford novel mechanisms to help better track and measure healthy behaviors within this user population. In this study, I present the design, implementation and simulation of such a system that helps manage health information for these types of users (underserved population). The system utilizes SMS technology to aid in personal information storage and retrieval. SMS was chosen because it is an effective and efficient method to connect with these populations, many of whom currently have cellular devices. SMS also provides a low barrier to entry for those who do not have a cellular device. This management of personal health information is of concern for underserved populations facing chronic illnesses The proposed system focuses on three steps; User Input, Data Processing and Information Output. First, data is collected from the user through questionnaire via SMS, which is curated to target patterns of healthy behaviors and irregularities in health. Next, a decision tree classification technique is used to create a machine-learning model by training on sample dataset. Finally, the system predicts if the user behavior is healthy or not by using the trained model and provide suggestions to manage their health properly.