A One-Dimensional Convolutional Neural Network and Long Short-Term Memory Model for Limb Movement Detection

Abstract

Based on the information from an electrocardiogram (ECG), this research demonstrates that a deep learning model known as deepPLM may automatically diagnose periodic limb movement syndrome (PLMS). The deepPLM model that was built has a total of five layers: a completely connected layer, two long-term memory units, four 1D convolutional layers, and one FCL. The dataset from the MrOS project was utilized in the process of developing the model, in which the model was trained, validated, and tested. Each of the 52 people who participated in the MrOS dataset had a single-lead electrocardiogram (ECG) signal based on the polysomnographic tape. After being normalized and segmented, the electrocardiogram signal was then split into three different sets: the training set, the validation set, and the test set. The deepPLM model's effectiveness was evaluated using the following metrics: Fl-score (93%), precision (91%), and recall (94.2%) for the controlling set; Fl-score (93%), precision (91%), and recall (94.2%) for the treatment set. The results show that autonomous PLMS categorization may be performed on sufferers by utilizing the deepPLM model, that is based on a single-lead ECG. This has the potential to be an effective tool for delivering treatment to seniors in the comfort of their own homes and a different approach to testing for PLMS.