Pengembangan Wearable Health Vest Berbasis ECG Dan PPG Untuk Pemantauan Tanda Vital Dengan Fitur Deteksi Infark Miokard

Nadhif, Haykal (2026) Pengembangan Wearable Health Vest Berbasis ECG Dan PPG Untuk Pemantauan Tanda Vital Dengan Fitur Deteksi Infark Miokard. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5023221001-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Penyakit kardiovaskular merupakan salah satu penyebab kematian utama di dunia, sehingga pemantauan tanda vital secara kontinu menjadi kebutuhan yang penting. Perangkat wearable yang ada saat ini umumnya masih terbatas pada pemantauan parameter dasar, sementara perangkat health vest terdahulu masih memiliki keterbatasan pada rentang estimasi laju pernapasan, integrasi sensor PPG, serta belum adanya deteksi dini infark miokard. Penelitian ini mengembangkan wearable health vest berbasis ECG dan PPG yang mampu memantau parameter vital secara komprehensif sekaligus memberikan deteksi dini infark miokard (MI), dengan estimasi laju detak jantung, laju pernapasan, variabilitas detak jantung, SpO₂, tekanan darah, dan kadar glukosa darah, serta dilengkapi sistem transmisi berbasis IoT. Pengujian dilakukan menggunakan dataset publik (MIT-BIH, BIDMC, PTB-XL) dan data primer dari 27 subjek. Ekstraksi R-peak mencapai sensitivitas 99,322% dan positive predictivity 99,080%, melampaui standar AAMI EC57. Estimasi laju detak jantung memperoleh MAE sebesar 1,127 bpm, laju pernapasan MAE sebesar 1,288 brpm, dan SpO₂ MAE sebesar 0,847%. Estimasi tekanan darah menggunakan model CNN+SE+BiLSTM+Attention mencapai MAE sebesar 8,62 mmHg untuk SBP dan 6,40 mmHg untuk DBP, sedangkan estimasi glukosa darah menggunakan model CNN+SE mencapai MAE sebesar 14,52 mg/dL dan RMSE sebesar 18,75 mg/dL, dengan 76,5% data berada pada Zona A dan 100% pada Zona A+B Clarke Error Grid; keduanya masih bersifat pendahuluan mengingat keterbatasan jumlah subjek. Deteksi MI menggunakan 1D-CNN dengan konvolusi multi-skala, Squeeze-Excite attention, dan konvolusi dilatasi mencapai akurasi 85,957%, sensitivitas 86,543%, dan AUC-ROC sebesar 0,930 pada PTB-XL. Analisis lanjutan berdasarkan manifestasi morfologi MI pada Lead II menunjukkan sensitivitas hingga ~98% pada rekaman dengan manifestasi yang jelas, dan menurun menjadi 77,92% pada rekaman tanpa manifestasi morfologi, yang mendominasi data uji (58,1%), sehingga keterbatasan ini lebih disebabkan oleh konfigurasi single-lead dibandingkan kapasitas model. Perangkat ini menunjukkan performa yang menjanjikan, meskipun validasi lebih lanjut pada populasi yang lebih besar masih diperlukan sebelum dapat diterapkan secara klinis.
===============================================================================================================================
Cardiovascular disease is one of the leading causes of death worldwide, making continuous vital sign monitoring an important need. Current wearable devices are generally limited to basic parameter monitoring, while previous healthvest devices still had limitations in respiratory rate estimation range, PPG sensor integration, and absence of early myocardial infarction detection. This study develops a wearable health vest based on ECG and PPG capable of comprehensively monitoring vital parameters while providing early myocardial infarction (MI) detection, estimating heart rate, respiratory rate, heart rate variability, SpO2, blood pressure, and blood glucose, with an IoT-based transmission system. Testing used public datasets (MIT-BIH, BIDMC, PTB-XL) and primary data from 27 subjects. R-peak extraction achieved sensitivity of 99.322% and positive predictivity of 99.080%, exceeding the AAMI EC57 standard. Heart rate estimation obtained an MAE of 1.127 bpm, respiratory rate an MAE of 1.288 brpm, and SpO2 an MAE of 0.847%. Blood pressure estimation using a CNN+SE+BiLSTM+Attention model achieved an MAE of 8.62 mmHg for SBP and 6.40 mmHg for DBP, while blood glucose estimation using a CNN+SE model achieved an MAE of 14.52 mg/dL and RMSE of 18.75 mg/dL, with 76.5%% of data in Zone A and 100% in Zone A+B of the Clarke Error Grid; both remain preliminary given limited subject data. MI detection using a 1D-CNN with multi-scale convolution, Squeeze-Excite attention, and dilated convolution achieved 85.957% accuracy, 86.543% sensitivity, and an AUC-ROC of 0.930 on PTB-XL. Further analysis based on MI morphological manifestation on Lead II showed sensitivity up to ~98% on recordings with clear manifestation, decreasing to 77.92% without morphological manifestation, which dominated the test data (58.1%), indicating the limitation stems more from the single-lead configuration than model capacity. The device shows promising performance, though further validation on larger populations is needed before clinical application.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	wearable health vest, ECG, PPG, pemantauan tanda vital, deteksi infark miokard; wearable health vest, ECG, PPG, vital sign monitoring, myocardial infarction detection
Subjects:	R Medicine > R Medicine (General) > R856.2 Medical instruments and apparatus. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5102.9 Signal processing. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7878 Electronic instruments
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User:	Haykal Nadhif
Date Deposited:	31 Jul 2026 04:33
Last Modified:	31 Jul 2026 04:33
URI:	http://repository.its.ac.id/id/eprint/140327

Actions (login required)

View Item