The "TriAxialAccelerometerForecasting" project was initiated following the publication of the article [BTPBea22] (https://physionet.org/content/wearable-exercise-frailty/1.0.0/), which made available a dataset containing data from both ECG (electrocardiogram) and a triaxial accelerometer for 80 frail elderly patients undergoing post-operative cardiac rehabilitation. These data were collected during physical tests using a chest-worn heart rate monitor called the Polar H10.
The primary objective of this project is to develop a deep learning model capable of making predictions based on the three signals recorded by the accelerometer along the x, y, and z axes. This model aims to identify the optimal time to predict potential falls among frail elderly individuals, with the goal of preventing harmful incidents. This innovative approach to managing elderly frailty seeks to anticipate and prevent falls rather than dealing with their consequences.
The report will provide details about the dataset used, the methodology employed to train and optimize the deep learning models, and the practical results obtained. This research represents a significant step forward in fall prevention among frail elderly individuals, paving the way for further developments in the field.