Non-contact blood pressure detection based on weighted ensemble learning model

This paper proposes a non-contact method for measuring blood pressure using imaging photoplethysmography (IPPG) based on weighted ensemble learning model. The method involves recording video of the face and hand using a webcam in ambient light conditions and extracting blood pressure-related features from the IPPG signal. Machine learning methods are employed to build a blood pressure prediction model. Six machine learning algorithms were used to construct blood pressure prediction models, respectively, and the performance of the models was evaluated by Pearson’s correlation coefficient. The three machine learning algorithms with the highest correlation with the actual values were selected as base learners and input into the selected meta-learner through weight allocation. Finally, a blood pressure prediction model based on a weighted ensemble learning model was constructed. Laboratory and hospital scenarios were used to evaluate the model’s performance. In addition, this paper proposes an algorithm based on EEMD (ensemble empirical mode decomposition)–WT (wavelet transform) joint filtering to extract the IPPG signal. Finally, a non-contact blood pressure detection system was built, which is of great significance for the development of future non-contact blood pressure detection equipment.