Abstract
Healthcare time series classification is to classify the collected human physiological information based on artificial intelligence technologies. The main purpose is to use pattern recognition technology to enable machines to analyze characteristics of human physiological signals based on deep learning in electronic health (E-health) industry 4.0. Healthcare time series classification can analyze various physiological information of the human body, make correct disease treatments, and reduce medical costs. In this paper, we propose a multiple-head convolutional LSTM (MCL) model for healthcare time series classification. MCL is a convolutional LSTM (ConvLSTM) model with multiple heads. It can extract both time and spatial features of healthcare data and increase the number of features to achieve more accurate classification results.
Highlights
In recent years, the application of pattern recognition technology in healthcare data analysis and processing has become more and more extensive
Much work has been done to analyze the healthcare time series based on artificial intelligence (AI) technology
(3) Conduct several experiments based on human activity recognition (HAR), MITBIH, and Sleep-EDF to verify the effectiveness of multiple-head convolutional LSTM (MCL)
Summary
The application of pattern recognition technology in healthcare data analysis and processing has become more and more extensive. This work promotes the application of AI deep learning in the field of medicine and effectively promotes the development of modern medicine. Through these researches, we find that the improvement of the accuracy of Healthcare Timing Classification has always been a focus and difficulty in academic research. One of the main problems that makes Healthcare Timing Classification accuracy difficult to improve is category imbalance. (1) Propose a method of normalization, pruning, and adjusting the weight of sample labels to solve the problem of category imbalance (2) Design a MCL model for high-accuracy healthcare timing classification (3) Conduct several experiments based on HAR, MITBIH, and Sleep-EDF to verify the effectiveness of MCL.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
