Issue |
E3S Web Conf.
Volume 605, 2025
The 9th International Conference on Energy, Environment, Epidemiology and Information System (ICENIS 2024)
|
|
---|---|---|
Article Number | 03051 | |
Number of page(s) | 9 | |
Section | Environment | |
DOI | https://doi.org/10.1051/e3sconf/202560503051 | |
Published online | 17 January 2025 |
A comparative study of machine learning algorithms for fall detection in technology-based healthcare system: Analyzing SVM, KNN, decision tree, random forest, LSTM, and CNN
1 Department of Informatic, Engineering Faculty, Universitas Jenderal Soedirman, Purwokerto, Indonesia.
2 Department of Computer Engineering, Diponegoro University, Semarang, Indonesia.
* Corresponding author : lasmedi.afuan@unsoed.ac.id
Fall detection is a major challenge in the development of technology-based healthcare systems, particularly in elderly care. This study aims to compare the performance of six classification algorithms: Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree, Random Forest, Long Short-Term Memory (LSTM), and Convolutional Neural Networks (CNN) in detecting fall incidents using wearable sensor data such as accelerometers and gyroscopes. The research utilizes a dataset consisting of 1,428 training samples and 573 testing samples, evaluated using a 10-fold cross-validation technique. The results show that CNN and LSTM deliver the best performance with accuracies of 94% and 92%, while Random Forest offers a good balance between accuracy and processing time. SVM and KNN exhibit faster processing times but slightly lower accuracies, at 87% and 84%, respectively. The superiority of CNN and LSTM in detecting more complex fall patterns aligns with previous studies emphasizing the capabilities of deep learning models in sensor data classification. The implications of these findings that the selection of fall detection algorithms must consider system priorities, whether focused on high accuracy or processing efficiency. Additionally, this research opens avenues for optimizing deep learning models and leveraging edge computing technologies to reduce response times in wearable device applications.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.