Non-Invasive Blood Pressure Detection Method Based on CNN-LSTM-LightGBM Combination Model

To improve the accuracy of non-invasive blood pressure prediction models and reduce the influence of individual body differences and the shortcomings of single prediction models on the accuracy of non-invasive blood pressure prediction, this paper proposes a non-invasive blood pressure detection method based on a combination model of convolutional neural networks (CNN), long short-term memory (LSTM), and light gradient boosting machine (lightGBM). First, non-invasive blood pressure prediction models were separately established for CNN-LSTM and lightGBM. In the CNN-LSTM model, CNN was used for feature processing of preprocessed photoplethysmography (PPG) signal features and human feature datasets, and the processed features were used as inputs for the LSTM model, thereby constructing a non-invasive blood pressure prediction model for CNN-LSTM. Finally, the optimal weighted combination method was used to combine the predicted results of the CNN-LSTM model and the lightGBM model to construct a combined non-invasive blood pressure prediction model. Experimental results show that the combined model has higher prediction accuracy than other single models, and the systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean absolute error (MAE) obtained by the CNN-LSTM-lightGBM model meet the standards set by the Association for the Advancement of Medical Instrumentation (AAMI) (±5mmHg), showing high consistency with the traditional mercury sphygmomanometer blood pressure measurement.