Prayogi, Bagus (2025) Pengembangan Sistem Deteksi Stres dengan Induksi Aritmetika Menggunakan Wearable Photoplethysmography (PPG). Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5026211169-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy |
Abstract
Stres merupakan kondisi yang terjadi ketika homeostasis tubuh terganggu akibat tekanan mental maupun fisik, yang memicu reaksi fisiologis seperti peningkatan detak jantung, tekanan darah, dan kadar kortisol. Jika berlangsung dalam jangka panjang, stres dapat berdampak buruk terhadap kesehatan, termasuk meningkatkan risiko penyakit kardiovaskular. Untuk mencegah dampak tersebut, pengelolaan stres sebaiknya diawali dengan deteksi dini terhadap kondisi stres. Salah satu pendekatan yang dapat digunakan adalah deteksi stres berbasis sinyal fisiologis, seperti sinyal Photoplethysmogram (PPG) dari perangkat wearable. Deteksi ini dapat diotomatisasi dengan bantuan algoritma machine learning yang mampu mengenali pola fisiologis tertentu yang terkait dengan kondisi stres. Penelitian ini bertujuan untuk merancang metode deteksi stres yang diinduksi melalui tugas aritmetika dengan memanfaatkan sinyal PPG dari perangkat Polar Verity Sense, serta membandingkan performa dua algoritma machine learning, yaitu Support Vector Machine (SVM) dan Linear Discriminant Analysis (LDA), dalam mengklasifikasikan kondisi stres dan tidak stres. Data dikumpulkan dari 10 partisipan berusia 20–30 tahun melalui proses akuisisi sinyal, denoising, deteksi puncak, serta ekstraksi fitur HRV (domain time, frequency, dan non-linear), sebelum dilakukan pelatihan dan pengujian model. Hasil menunjukkan bahwa fitur-fitur seperti D2, shanEn, lftp, LFHF, dan hftp secara konsisten membedakan kondisi stres dan tidak stres. Model SVM menghasilkan akurasi rata-rata sebesar 73,33% dan F1-score 0,7293, sedangkan model LDA menghasilkan akurasi 74,44% dan F1-score 0,7161. Temuan ini menunjukkan bahwa deteksi stres berbasis sinyal PPG berpotensi untuk dikembangkan sebagai solusi non-invasif dan real-time dalam pemantauan kondisi stres, khususnya melalui perangkat wearable yang mudah digunakan.
=====================================================================================================================================
Stress is a condition that occurs when the body's homeostasis is disrupted due to mental or physical pressure, triggering physiological responses such as increased heart rate, blood pressure, and cortisol levels. When prolonged, stress can lead to serious health issues, including cardiovascular diseases. To mitigate these impacts, managing stress should begin with early detection. One promising approach is stress detection based on physiological signals, such as Photoplethysmogram (PPG) signals obtained from wearable devices. This detection process can be automated using machine learning algorithms capable of recognizing physiological patterns associated with stress conditions. This study aims to design a stress detection method induced through arithmetic tasks using PPG signals collected from a Polar Verity Sense wearable device. It also compares the performance of two machine learning algorithms—Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA)—in classifying stress and non-stress conditions. Data were collected from 10 participants aged 20–30 through a process involving signal acquisition, denoising, peak detection, and HRV feature extraction (including time domain, frequency domain, and non-linear features), followed by model training and evaluation. The results show that features such as D2, shanEn, lftp, LFHF, and hftp consistently distinguish between stress and non-stress states. The SVM model achieved an average accuracy of 73.33% and an F1-score of 0.7293, while the LDA model achieved an accuracy of 74.44% and an F1-score of 0.7161. These findings indicate that PPG-based stress detection holds potential as a non-invasive, real-time solution for stress monitoring, especially through user-friendly wearable devices.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Stres, PPG, HRV, Machine Learning, SVM, LDA, Wearable, Deteksi Dini, Stress, PPG, HRV, Machine Learning, SVM, LDA, Wearable, Early Detection |
Subjects: | T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis |
Depositing User: | Bagus Prayogi |
Date Deposited: | 25 Jul 2025 01:47 |
Last Modified: | 25 Jul 2025 01:48 |
URI: | http://repository.its.ac.id/id/eprint/121631 |
Actions (login required)
![]() |
View Item |