Ensemble Learning Approach for Advanced Predictive Modeling of Biometric Data and Action States With Smart Sensing

The advancement in wearable technology with integrated biometric sensors allows real-time continuous monitoring of vital data such as heart rate and breathing rate. The utilization of this real-time data can yield significant insights into a person's health status by facilitating the early identification of anomalies or patterns that may point to possible health problems. Moreover, it illustrates how the vitals alter or respond as the individual carries out a certain task. For a physically or mentally demanding task, understanding this response could ensure better organization of that task and minimal strain on human health by reducing stress and cognitive load. Unfortunately, it is not always feasible to continuously monitor biometric data. In this paper, we propose an alternative solution to the problem by using predictive models to forecast data points in the future. This paper presents a method with an ensemble learning approach capable of forecasting vitals from small chunks of data. The intelligent system proposed in the paper is connected to a Dielectric Elastomer to provide smart sensing to the user. The paper presents a new dataset containing vitals and contributes to classifying a particular task or action state based on recorded human vitals. While existing research focuses on the prediction of biometrics, our model is capable of forecasting the biometrics and outperforms standard machine learning predictive models by a margin across multiple evaluation metrics and the forecasted graph shows the effectiveness of our approach. This technique offers a viable alternative for data monitoring by data simulation, improving task management, and minimizing health strain.