Seminar: Telemonitoring in Hypertension Management

Abstract

This study contributes to the limited literature on applying machine learning (ML) to telemonitoring data for timely decision support systems (DSS). We propose an end-to-end DSS framework to enable clinicians to intervene proactively by generating actionable information from daily telemonitoring data.

The talk will have two parts. The goal in the first part was to predict the probability of patients having a higher-than-expected systolic blood pressure (SBP) level in the next 7 days using real-world data. We focus on the framework for effective development of a human-computer hybrid system that utilizes ML models to predict adverse hypertensive events and evaluate different ML models to determine optimal settings. Our goal is to create a system that enhances collaboration between ML and clinical expertise in hypertension management. We report on our findings where ML models (i.e., Logistic Regression, Random Forest (RF), XGBoost, Fusion Neural Network) were trained and evaluated with tenfold cross-validation, using AUCPR, AUCROC, and F1 score to study optimal configurations. In addition, different methods for calculating daily risk scores were compared. Finally, SHapley Additive exPlanations (SHAP) was used to investigate feature importance.

The goal in the second part is to understand barriers to adoption and engagement in telemonitoring for chronic disease management. We report findings from interviews with healthcare professionals and patients. The ultimate goal of this line of research is to better understand human behavior related to the use of the system and how best to design and implement the human parts of the human-computer hybrid system.

In this talk, we present the importance of incorporating human judgment and expert domain knowledge into the data science activities at all steps and the numerous design decisions required to obtain valid results and ultimately useful insights. In addition, any good AI system requires a human-computer hybrid system where appropriate tasks are performed by both humans and computers through an iterative spiral process. We share how research is conducted on understanding the necessary human behaviors for designing such systems. Real-world AI systems require multi-disciplinary collaboration, including data scientists and behavioral scientists.