Research on a System Combining Improved Wavelet Thresholding with LightGBM for Intelligent Pulse Recognition

DOI: https://doi.org/10.62517/jmhs.202605302

Author(s)

Junjie Jiang, Zihao Zhang*, Huimin Li, Ziheng Wei, Baixu Jiang, Simin Wang, Yi Liu

Affiliation(s)

Jiangsu Normal University KeWen College, Xuzhou, Jiangsu, China *Corresponding Author

Abstract

In response to the issues of subjective nature in traditional Chinese medicine’s pulse diagnosis, the lack of objective criteria, and the low level of intelligence and insufficient recognition accuracy of existing pulse diagnosis equipment, this paper proposes a method for intelligent pulse recognition that integrates improved wavelet threshold denoising with LightGBM. Firstly, an improved exponential wavelet threshold function is designed to address the shortcomings of traditional hard threshold functions, which are discontinuous, and soft threshold functions, which have constant biases. This function effectively removes baseline drift and high-frequency noise while preserving the key details of the pulse signal to the greatest extent. Secondly, multi-domain feature vectors are extracted from three dimensions: time domain, frequency domain, and time-frequency domain, to comprehensively characterize the morphological and dynamic characteristics of the pulse. Finally, a pulse classification model based on LightGBM is constructed, with bagging integration strategies enhancing the model’s generalizable discriminative ability. Experimental results show that the accuracy of pulse recognition using this method reaches 91.6%, with an average recall rate of 89.4%. Compared to traditional SVM and random forest methods, there is a significant improvement. The improved threshold function enhances the signal-to-noise ratio by approximately 4.7 dB, and the accuracy of distinguishing between “slippery pulse” and “sticky pulse” increases by 12.6% after incorporating multi-domain features. Based on this, a pulse intelligent recognition software system with a B/S architecture has been developed, validating the engineering feasibility of the proposed method.

Keywords

Pulse Pattern Recognition; Wavelet Threshold Denoising; LightGBM; Feature Extraction; Intelligent Diagnosis in Traditional Chinese Medicine

References

[1]Wang Yiqin. Progress in the Objectification of Traditional Chinese Medicine Pulse Diagnosis. Journal of Shanghai University of Traditional Chinese Medicine, 2020, 34 (3):1-8. [2]CHU Y W, CHEN L L, WANG H J, et al. Pulse diagnosis: A review of the sensor technologies and signal processing techniques. IEEE Reviews in Biomedical Engineering, 2022, 15:123-141. [3]Li Yonggang, Zhang Hao, Wang Zhiming, et al. Design of a System for Detecting Pulse Signals Based on PVDF Piezoelectric Sensors. Journal of Sensors Technology, 2021, 34 (8):1052-1059. [4]DONOHO D L, JOHNSTONE I M. Adapting to unknown smoothness via wavelet shrinkage. Journal of the American Statistical Association, 1995, 90 (432):1200-1224. [5]Chen Ming, Zhao Liang, Sun Tao. A Study on a Classification Method for Traditional Chinese Pulse Patterns Based on Random Forest. Journal of Biomedical Engineering, 2022, 39 (2):289-296. [6]ZHANG X Y, LIU Q, WANG R. A comparative study of machine learning algorithms for wrist pulse signal classification. Biomedical Signal Processing and Control, 2021, 68:102635. [7]Yang Jie, Liu Jia, Wang Ning, et al. Current Status and Key Technology Analysis of Pulse Diagnosis Devices. Chinese Journal of Medical Instruments, 2023, 47 (1):56-62. [8]Hou Chenyang. GB/T 47335.2-2026: Interpretation of Terminology Standards for Pulse Patterns and Objective Calibration Schemes for Pulse Diagnosis. Chinese Journal of Medical Devices, 2026, 50 (02):189-195. [9]Wang Yongjun, Chen Chen. A Quantification Method for Pulse Multi-Domain Characteristics Based on Elephant Array Theory. Journal of Shanghai University of Traditional Chinese Medicine, 2026, 40 (01):1-7. [10]KE G L, MENG Q, FINLEY T, et al. LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems (NeurIPS), 2017:3146-3154. [11]HUANG N E, SHEN Z, LONG S R, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proceedings of the Royal Society A, 1998, 454 (1971):903-995. [12]Zhou Zhihua. Machine Learning. Beijing: Tsinghua University Press, 2016. [13]Liu Muyang, Zhao Zhe. Improving the Wavelet-EMD Fusion LightGBM Pulse Classification Algorithm. Journal of Biomedical Engineering, 2025, 42 (03):321-328. [14]LIU M Y, ZHAO Z. Wrist pulse signal classification optimized by adaptive wavelet threshold and LightGBM. Biomedical Signal Processing and Control, 2024, 88:107623. [15]Chen Yutong. Study on the Noise Adaptive Suppression Algorithm for PVDF Pulse Sensors. Journal of Sensors Technology, 2023, 36 (09):1261-1268. [16]WANG Y Q. A review of intelligent pulse diagnosis based on machine learning 2023–2026. IEEE Journal of Biomedical and Health Informatics, 2023, 27 (11):5689-5702. [17]BANDT C, POMPE B. Permutation entropy: A natural complexity measure for time series. Physical Review Letters, 2002, 88 (17):174102.