Scholarlink

The EMG signals have become more advanced in healthcare, especially in predicting paralysis based on signals from lower limb EMGs. Paralysis, caused by different things like injuries or medical issues, has a big impact on people’s lives. This new method tries to predict paralysis by using a special model that combines different machine-learning methods. It starts by collecting EMG data from lower limbs in different channels, both normal and abnormal in paralysis patients. Then, it cleans up the data to make it better by reducing noise and picking out important features. It uses 13 different machine learning models as baseline classifiers to learn EMG data, like LR_L2, LDA, KNN9, DTC, NB, AB, GBM, RF_Ent100, RF_Gini100, ET100, MLP, SGD3000, and XGB_100. In this study, the top classifiers from the baseline model were validated using k-fold validation and then combined these models together to make a proposed stack ensemble model for better predictions. This stack ensemble model aims to make predicting paralysis more accurate and reliable, helping doctors make better choices for patients. The results show a 71.68% accuracy rate which is better than individual classifiers. Finally, the proposed stacking ensemble model is compared to previous literature work.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Stack Ensemble Model for Paralysis Prediction from Lower Limb EMG Signals Using Machine Learning Approaches

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Harshita S. Awachat,
  • Rajendra Rewatkar,
  • K. T. V. Reddy

摘要

The EMG signals have become more advanced in healthcare, especially in predicting paralysis based on signals from lower limb EMGs. Paralysis, caused by different things like injuries or medical issues, has a big impact on people’s lives. This new method tries to predict paralysis by using a special model that combines different machine-learning methods. It starts by collecting EMG data from lower limbs in different channels, both normal and abnormal in paralysis patients. Then, it cleans up the data to make it better by reducing noise and picking out important features. It uses 13 different machine learning models as baseline classifiers to learn EMG data, like LR_L2, LDA, KNN9, DTC, NB, AB, GBM, RF_Ent100, RF_Gini100, ET100, MLP, SGD3000, and XGB_100. In this study, the top classifiers from the baseline model were validated using k-fold validation and then combined these models together to make a proposed stack ensemble model for better predictions. This stack ensemble model aims to make predicting paralysis more accurate and reliable, helping doctors make better choices for patients. The results show a 71.68% accuracy rate which is better than individual classifiers. Finally, the proposed stacking ensemble model is compared to previous literature work.