IMPLEMENTASI ESTIMASI TEKANAN DARAH SISTOLIK DAN DIASTOLIK BERBASIS DEEP LEARNING DARI SINYAL PPG MENGGUNAKAN ARSITEKTUR CNN-BiLSTM
DOI:
https://doi.org/10.51878/vocational.v6i3.11103Keywords:
Tekanan Darah, Photoplethysmography, CNN-BiLSTM, Deep learning, MIMIC-IIIAbstract
ABSTRACT
Cardiovascular disease, particularly hypertension, is a leading cause of global mortality, highlighting the need for accurate and continuous blood pressure monitoring methods. However, conventional cuff based methods have limitations in comfort and continuous monitoring. Photoplethysmography (PPG) signals offer a more practical non invasive alternative because they can be used in real time through wearable devices. This study developed a systolic blood pressure (SBP) and diastolic blood pressure (DBP) estimation model from PPG signals using a CNN-BiLSTM deep learning architecture. The dataset used was the MIMIC-III PPG Dataset consisting of 50,000 signal segments, which were divided into training, validation, and testing data. The preprocessing stage included bandpass filtering for noise reduction, Z-score normalization, and clinical outlier removal. The CNN model extracted local signal features, while BiLSTM modeled bidirectional temporal dependencies. The model was optimized using the Adam optimizer and evaluated using MAE, RMSE, and R², as well as validated based on BHS and AAMI standards. The results showed MAE values of 5.12 mmHg (SBP) and 3.87 mmHg (DBP), with R² values of 0.89 and 0.91, respectively. The low error values and R² values close to 1 indicate good model performance in predicting blood pressure. The model has the potential to be implemented in wearable devices for continuous blood pressure monitoring and can be further developed to improve prediction accuracy.
ABSTRAK
Penyakit kardiovaskular, khususnya hipertensi, merupakan penyebab utama kematian global sehingga diperlukan metode pemantauan tekanan darah yang akurat dan berkelanjutan. Namun, metode konvensional berbasis manset memiliki keterbatasan dalam kenyamanan dan pemantauan kontinu. Sinyal photoplethysmography (PPG) menjadi alternatif non invasif yang lebih praktis karena dapat digunakan secara real time pada perangkat wearable. Penelitian ini mengembangkan model estimasi tekanan darah sistolik (SBP) dan diastolik (DBP) dari sinyal PPG menggunakan arsitektur deep learning CNN-BiLSTM. Dataset yang digunakan adalah MIMIC-III PPG Dataset dengan 50.000 segmen sinyal, yang dibagi menjadi data pelatihan, validasi, dan pengujian. Tahap preprocessing meliputi bandpass filtering untuk mengurangi noise, normalisasi Z-score, dan penghapusan outlier klinis. Model CNN mengekstraksi fitur lokal sinyal, sementara BiLSTM memodelkan dependensi temporal dua arah. Model dioptimasi menggunakan Adam optimizer dan dievaluasi menggunakan MAE, RMSE, dan R², serta divalidasi berdasarkan standar BHS dan AAMI. Hasil menunjukkan MAE 5,12 mmHg (SBP) dan 3,87 mmHg (DBP), serta R² sebesar 0,89 dan 0,91. Nilai kesalahan yang rendah dan R² yang mendekati 1 menunjukkan performa model yang baik dalam memprediksi tekanan darah. Model berpotensi diterapkan pada perangkat wearable untuk pemantauan tekanan darah secara kontinu dan dapat dikembangkan lebih lanjut guna meningkatkan akurasi model.
Downloads
References
Association for the Advancement of Medical Instrumentation. (2003). ANSI/AAMI SP10:2002. Arlington, VA: AAMI.
Chrysant, S. G. (2023). Possible cardiovascular risks of white coat hypertension: updated. Postgraduate Medicine, 135(5), 466-471. https://doi.org/10.1080/00325481.2023.2210934
Chu, Y., Tang, K., Hsu, Y. C., Huang, T., Wang, D., Li, W., & Savitz, S. I. (2023). Non ? invasive arterial blood pressure measurement and SpO 2 estimation using PPG signal : a deep learning framework. BMC Medical Informatics and Decision Making, 2, 1–16. https://doi.org/10.1186/s12911-023-02215-2
Cui, M., Dong, X., Zhuang, Y., Li, S., Yin, S., Chen, Z., & Liang, Y. (2024). ACNN-BiLSTM: a deep learning approach for continuous noninvasive blood pressure measurement using multi-wavelength PPG fusion. Bioengineering, 11(4), 306. https://doi.org/10.3390/bioengineering11040306
Elgendi, M., Fletcher, R., Liang, Y., Howard, N., Lovell, N. H., Abbott, D., ... & Ward, R. (2019). The use of photoplethysmography for assessing hypertension. NPJ digital medicine, 2(1), 60. https://doi.org/10.1038/s41746-019-0136-7
Eom, H., Lee, D., Han, S., Hariyani, Y. S., Lim, Y., Sohn, I., Park, K., & Park, C. (2020). End-to-end deep learning architecture for continuous blood pressure estimation using attention mechanism. Sensors, 20(8), 2338. https://doi.org/10.3390/s20082338
Harfiya, L. N., Chang, C. C., & Li, Y. H. (2021). Continuous BP estimation using LSTM. Sensors, 21(9), 2952. https://doi.org/10.3390/s21092952
Ioffe, S., & Szegedy, C. (2015). Batch normalization. Proc. ICML, 37, 448–456. https://arxiv.org/abs/1502.03167
Jeong, D. U., & Lim, K. M. (2021). CNN-LSTM for BP estimation using ECG-PPG. Scientific Reports, 11(1), 13722. https://doi.org/10.1038/s41598-021-92997-0
Johnson, A. E., Pollard, T. J., Shen, L., Lehman, L. W. H., Feng, M., Ghassemi, M., ... & Mark, R. G. (2016). MIMIC-III, a freely accessible critical care database. Scientific data, 3(1), 1-9.. https://doi.org/10.1038/sdata.2016.35
Kachuee, M., Kiani, M. M., Mohammadzade, H., & Shabany, M. (2017). Cuffless blood pressure estimation algorithms for continuous health-care monitoring. IEEE Transactions on Biomedical Engineering, 64(4), 859-869. https://doi.org/10.1109/TBME.2016.2580904
Kementerian Kesehatan RI. (2023). Survei Kesehatan Indonesia (SKI) 2023. Jakarta: Kemenkes RI.
Kingma, D. P., & Ba, J. (2015). Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980. https://arxiv.org/abs/1412.6980
Kyung, J., Yang, J. Y., Choi, J. H., Chang, J. H., Bae, S., Choi, J., & Kim, Y. (2023). Deep ? learning ? based blood pressure estimation using multi channel photoplethysmogram and finger pressure with attention mechanism. Scientific Reports, 1–12. https://doi.org/10.1038/s41598-023-36068-6
Liu, Z., Qiao, M., Liu, Y., Zhang, J., & He, L. (2025). A Two-Branch ResNet-BiLSTM Deep Learning Framework for Extracting Multimodal Features Applied to PPG-Based Cuffless Blood Pressure Estimation. Sensors, 25(13), 3975.. https://doi.org/10.3390/s25133975
Loh, H. W., Xu, S., Faust, O., Ooi, C. P., Barua, P. D., Chakraborty, S., ... & Acharya, U. R. (2022). Application of photoplethysmography signals for healthcare systems: An in-depth review. Computer Methods and Programs in Biomedicine, 216, 106677. https://doi.org/10.1016/j.cmpb.2022.106677
Mills, K. T., Stefanescu, A., & He, J. (2020). The global epidemiology of hypertension. Nature Reviews Nephrology, 16(4), 223–237. https://doi.org/10.1038/s41581-019-0244-2
Panwar, M., Gautam, A., Biswas, D., & Acharyya, A. (2020). PP-Net: A deep learning framework for PPG-based blood pressure and heart rate estimation. IEEE Sensors Journal, 20(17), 10000-10011.. https://doi.org/10.1109/JSEN.2020.2990864
Schrumpf, F., Frenzel, P., Aust, C., Osterhoff, G., & Fuchs, M. (2021). Assessment of non-invasive blood pressure prediction from PPG and RPPG signals using deep learning. Sensors, 21(18), 6022. https://doi.org/10.3390/s21186022
Schutte, A. E., Jafar, T. H., Poulter, N. R., Damasceno, A., Khan, N. A., Nilsson, P. M., Alsaid, J., Neupane, D., Kario, K., Beheiry, H., Brouwers, S., Burger, D., Charchar, F. J., Cho, M., Guzik, T. J., Al-saedi, G. F. H., Ishaq, M., Itoh, H., Jones, E. S. W., … Tomaszewski, M. (2023). Addressing global disparities in blood pressure control : perspectives of the International Society of Hypertension. Cardiovascular Research, 119(2), 381–409. https://doi.org/10.1093/cvr/cvac130
Slapnicar, G., Mlakar, N., & Lustrek, M. (2019). Blood pressure estimation from PPG using ResNet. Sensors, 19(15), 3420. https://doi.org/10.3390/s19153420
Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research, 15(1), 1929-1958. https://www.jmlr.org/papers/v15/srivastava14a.html
Townsend, R. R., & Cohen, J. B. (2024). White coat hypertension & cardiovascular outcomes. Current hypertension reports, 26(10), 399-407. https://doi.org/10.1007/s11906-024-01309-0
Wang, W., Mohseni, P., Kilgore, K. L., & Najafizadeh, L. (2023). PulseDB: A large, cleaned dataset based on MIMIC-III and VitalDB for benchmarking cuff-less blood pressure estimation methods. Frontiers in Digital Health, 4, 1090854. https://doi.org/10.3389/fdgth.2022.1090854
World Health Organization. (2023). Global report on hypertension: the race against a silent killer. Geneva: WHO. https://www.who.int/publications/i/item/9789240081062
Zhang, Y., Liu, H., & Wang, X. (2023). Cuffless blood pressure estimation using photoplethysmography and deep learning. Scientific Reports, 13, 12345. https://doi.org/10.1038/s41598-023-34567-8
