Data Collection Methods and Predictive Analysis for Fall Prevention in Elderly Populations

One of the most severe health risks to elderly populations is the risk of falling. Falls can lead to acute injuries, correlating with a decline in physical health. Recognizing the significance of this issue, the Fall Prevention Project aims to develop an algorithm that can accurately predict and prevent falls in elderly demographics. Data collection is essential to understanding and predicting falls. This article addresses the data collection methods used in the Fall Prevention Project. By collecting data to train an algorithm, injuries, and fatalities can be modeled to improve the lives of at-risk patients. For data collection, MediaPipe, a computer vision framework, is combined with an Intel RealSense D435 Camera. After skeleton tracking is completed, a Python code navigates through this data, which is then exported to an Excel file containing the coordinates and velocities for markers on the human body. A machine learning algorithm is used to classify patient movement as either “Risky” or “Safe”. From this algorithm, future patient behavior can be analyzed to identify and prevent when a fall is occurring.