Predict the Elderly Fall Using IoT and AI Technology

Falls pose a significant concern among the elderly, leading to severe injuries, diminished quality of life, and increased medical expenses. To mitigate the impact of falls and promote preventative measures, early detection and precise risk assessment are imperative. Consequently, accurately predicting fall incidents within the elderly population becomes essential. Researchers in this study have harnessed various data mining and analysis techniques, in conjunction with machine learning algorithms, to achieve this goal. They have conducted statistical analyses, including descriptive analysis, ANOVA, t-test, P-value, and Pearson coefficient calculations, utilizing data from a stick device. These statistical measures help predict falls based on an array of variables and assess the significance of each attribute in fall occurrence. Classification modeling algorithms were employed for this purpose. Furthermore, the accuracy range of predictions was evaluated, and the model’s generalizability and robustness were assessed through cross-validation techniques. The algorithms were evaluated based on accuracy, precision, recall, and F-1 score. Notably, among all the algorithms applied, XG-Boost demonstrated the highest accuracy at 94.5%, effectively identifying nearly all instances of falls. In contrast, random forest exhibited the lowest accuracy at 86.9%. Remarkably, this study achieved consistently high accuracy, with all models exceeding a 90% accuracy rate. This accuracy serves as a measure of fall probability, categorized into three groups: no fall (0) and definite fall (1), assigning a numerical value to each outcome category.