Azka, Nadiya (2025) Rekonstruksi Sinyal Vital Sign Pada Sistem Monitoring Nonkontak Menggunakan Radar Berbasis BI-LSTM Untuk Mengurangi Efek Gerakan Tubuh Acak. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5023211040-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (19MB) | Request a copy |
Abstract
Pemantauan vital sign tubuh secara tepat dan akurat, seperti respiration rate variability (RRV) dan heart rate variability (HRV), sangat penting dalam mendukung diagnosis penyakit serta deteksi dini penurunan kondisi pasien. Meskipun metode konvensional seperti electrocardiogram (ECG) dan photoplethysmograph (PPG) memiliki akurasi yang tinggi, metode ini membutuhkan kontak langsung dengan tubuh pasien, yang dapat menyebabkan ketidaknyamanan, terutama untuk pemantauan jangka panjang. Pendekatan nonkontak menggunakan radar continuous wave (CW) menawarkan solusi melalui deteksi gerakan mikro dari dada. Namun, gerakan tubuh acak (random body movement/RBM) menjadi tantangan utama karena radar CW sangat sensitif terhadap perubahan posisi. Penelitian ini mengusulkan metode rekonstruksi sinyal menggunakan regresi Bi-LSTM untuk memulihkan sinyal vital sign yang cacat akibat gangguan RBM, sehingga dapat meningkatkan akurasi pemantauan vital sign nonkontak berbasis radar. Tahapan pemrosesan sinyal melibatkan tiga tahapan utama, yaitu rekonstruksi sinyal, demodulasi sinyal, serta estimasi vital sign. Penelitian ini berhasil meningkatkan akurasi sistem pemantauan vital sign nonkontak, meskipun terdapat gangguan RBM. Hasil pengujian pada 15 subjek menunjukkan peningkatan akurasi estimasi setelah rekonstruksi sinyal menggunakan Bi-LSTM. Pada pemantauan RRV selama lima menit, rata-rata nilai eror AAPE menurun dari 35,03% sebelum rekonstruksi menjadi 11,37% setelah rekonstruksi, APPE dari 710,41 milidetik menjadi 69,86 milidetik, RMSE dari 2,256 detik menjadi 0,560 detik, dan P-RMSE dari 65,75% menjadi 17,07%. Untuk pemantauan HRV, AAPE menurun dari 9,06% menjadi 5,89%, APPE dari 27,68 milidetik menjadi 6,72 milidetik, RMSE dari 0,089 detik menjadi 0,058 detik, dan P-RMSE dari 11,78% menjadi 7,65%.
===================================================================================================================================
Accurate and precise monitoring of vital signs, such as respiration rate variability (RRV) and heart rate variability (HRV), is crucial for supporting disease diagnosis and the early detection of patient deterioration. Although conventional methods like electrocardiography (ECG) and photoplethysmography (PPG) offer high accuracy, they require direct physical contact, which can cause discomfort, particularly during long-term monitoring. A non-contact approach using continuous wave (CW) radar presents a viable alternative by detecting micro-motions of the chest wall. However, random body movements (RBM) pose a significant challenge as CW radar is highly sensitive to positional changes. This paper proposes a signal reconstruction method based on Bidirectional Long Short-Term Memory (Bi-LSTM) regression to restore vital sign signals corrupted by RBM artifacts, thereby enhancing the accuracy of non-contact radar-based vital sign monitoring. The signal processing pipeline consists of three main stages: signal reconstruction, signal demodulation, and vital sign estimation. This study successfully improves the accuracy of the non-contact vital sign monitoring system in the presence of RBM disturbances. Experimental results from 15 subjects demonstrate a significant improvement in estimation accuracy following signal reconstruction with the Bi-LSTM model. For five-minute RR monitoring, the mean Absolute Average Percentage Error (AAPE) decreased from 35.03% to 11.37%, the Absolute Peak-to-peak Error (APPE) from 710.41 ms to 69.86 ms, the Root Mean Square Error (RMSE) from 2.256 s to 0.560 s, and the Percentage RMSE (P-RMSE) from 65.75% to 17.07%. For HRV monitoring, the AAPE decreased from 9.06% to 5.89%, APPE from 27.68 ms to 6.72 ms, RMSE from 0.089 s to 0.058 s, and P-RMSE from 11.78% to 7.65%.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Bi-LSTM, Gerakan Tubuh Acak, Nonkontak, Radar, Rekonstruksi Sinyal, Tanda Vital, Random Body Movement, Non-contact, Signal Reconstruction, Vital Sign |
Subjects: | T Technology > T Technology (General) > T174 Technological forecasting |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Biomedical Engineering > 11410-(S1) Undergraduate Thesis |
Depositing User: | Nadiya Azka |
Date Deposited: | 04 Aug 2025 05:00 |
Last Modified: | 01 Oct 2025 03:15 |
URI: | http://repository.its.ac.id/id/eprint/125513 |
Actions (login required)
![]() |
View Item |