Pertama, Nayla Pramudhita Putri (2025) Deteksi Obstructive Sleep Apnea Menggunakan Wearable Photoplethysmography Berbasis Support Vector Machine. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5023211012-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5023211012-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy |
Abstract
Obstructive Sleep Apnea (OSA) adalah gangguan tidur kronis yang dapat memicu berbagai komplikasi kesehatan serius, seperti hipertensi, penyakit jantung, dan diabetes. Metode deteksi standar saat ini, yaitu Polysomnography (PSG), memiliki kelemahan berupa biaya tinggi, waktu pengumpulan data yang lama, serta penggunaan alat yang kompleks, sehingga tidak ideal untuk pemantauan harian. Penelitian ini bertujuan untuk mengembangkan sistem deteksi OSA berbasis sensor Photoplethysmography (PPG) memanfaatkan algoritma Support Vector Machine (SVM) untuk klasifikasi. Sistem ini mengukur interval peak-to-peak sinyal PPG dan fitur respirasi. Perancangan perangkat keras melibatkan penggunaan
mikrokontroler Teensy 4.1 dan sensor PPG MAX30105 dengan LED merah dan inframerah. Perangkat lunak dikembangkan untuk mengolah data PPG dan mengklasifikasikan hasil menggunakan Support Vector Machine (SVM). Model SVM yang telah dibangun memperoleh nilai akurasi 93.8%, F1-Score 92.6%, dan AUC 98.4%. Kemudian untuk hasil pengujian data PPG menggunakan model mendapat nilai confidence level rata-rata sebesar 79.7%. Sistem juga diuji menggunakan pembanding ECG dan respiration belt. Diperoleh RMSE sebesar 1.91 BPM dan 64.4 ms untuk HRV dan 0.3962 BPM untuk respirasi. Hasil yang
diharapkan dari penelitian ini adalah solusi alternatif yang lebih terjangkau dan praktis dibanding PSG, memungkinkan deteksi OSA yang lebih efisien serta diagnosis dan perawatan tepat waktu bagi pasien.
============================================================================================================================================
Obstructive Sleep Apnea (OSA) is a chronic sleep disorder that can trigger various serious health complications, including hypertension, cardiovascular disease, and diabetes. The current standard detection method, Polysomnography (PSG), has limitations including high cost, lengthy data collection time, and complex instrumentation, making it unsuitable for daily monitoring applications. This study aims to develop an OSA detection system based on Photoplethysmography (PPG) sensors utilizing Support Vector Machine (SVM) algorithm for classification. The system measures peak-to-peak intervals of PPG signals and respiratory features. The hardware design involves the implementation of Teensy 4.1 microcontroller and MAX30105 PPG sensor with red and infrared LEDs. Software is developed to process PPG data and classify results using Support Vector Machine (SVM). The developed SVM model achieved an accuracy of 93.8%, an F1-Score of 92.6%, and an AUC of 98.4%. Furthermore, testing on PPG data using the model yielded an average confidence level of 79.7%. The system was also evaluated using comparison data from ECG and a respiration belt. The results showed an RMSE of 1.91 BPM and 64.4 ms for HRV, and 0.3962 BPM for respiration. The expected outcome of this research is an alternative solution that is more affordable and practical compared to PSG, enabling more efficient OSA detection and facilitating timely diagnosis and treatment for patients.
Actions (login required)
 |
View Item |