Aplikasi Pemantauan Emosi Berbasis Digital Phenotyping Multimodal untuk Intervensi Stres dan Kecemasan

Zahirah, Regina Salsabila (2026) Aplikasi Pemantauan Emosi Berbasis Digital Phenotyping Multimodal untuk Intervensi Stres dan Kecemasan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5023221026-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Gangguan kesehatan mental seperti stres dan kecemasan terus meningkat secara global, sementara metode pemantauan emosi konvensional masih bergantung pada laporan subjektif yang sering tidak merefleksikan kondisi nyata pengguna. Seiring perkembangan teknologi, pendekatan digital phenotyping memberikan cara baru untuk memantau perubahan emosional secara objektif melalui pemanfaatan sensor perangkat digital yang digunakan sehari-hari. Penelitian ini bertujuan mengembangkan sebuah aplikasi mobile yang mampu mendeteksi dan memantau kondisi stres dan kecemasan pengguna secara pasif melalui integrasi data multimodal dari perangkat wearable berbasis TTGO ESP32. Sistem menggabungkan tiga sinyal fisiologis, yaitu Heart Rate Variability (HRV), Electrodermal Activity (EDA), dan Skin Temperature (ST), dengan self report harian pengguna sebagai sumber informasi tambahan. Proses penelitian meliputi tahap preprocessing sinyal, ekstraksi fitur numerik, seleksi fitur berbasis mutual information, dan klasifikasi emosi ke dalam empat kelas kondisi menggunakan model machine learning berbasis XGBoost, yang menghasilkan akurasi sebesar 84,38% dan ROC-AUC macro sebesar 0,9310 pada data uji independen. Hasil klasifikasi fisiologis dan laporan diri digabungkan melalui mekanisme fusi berbobot berbasis reliabilitas, dengan seluruh alur data dikelola melalui backend Python menggunakan REST API agar proses sinkronisasi berlangsung stabil dan konsisten. Hasil penelitian ini adalah terbangunnya sistem pemantauan emosi terpadu yang menyajikan visualisasi kondisi harian yang informatif, menyediakan intervensi ringan melalui chatbot bernama Noxi, serta dilengkapi website pemantauan admin untuk memantau kondisi emosi pengguna secara real time. Pengujian usabilitas menggunakan System Usability Scale terhadap 43 responden menghasilkan rata-rata skor sebesar 80,41 yang masuk dalam kategori Excellent.
=====================================================================================================================================
Mental health disorders such as stress and anxiety continue to rise globally, while conventional emotion monitoring methods still rely on subjective reports that often fail to reflect users' actual conditions. With advances in technology, the digital phenotyping approach offers a new way to objectively monitor emotional changes through the use of sensors in everyday digital devices. This study aims to develop a mobile application capable of passively detecting and monitoring users' stress and anxiety conditions through the integration of multimodal data from a wearable device based on TTGO ESP32. The system combines three physiological signals, namely Heart Rate Variability (HRV), Electrodermal Activity (EDA), and Skin Temperature (ST), with users' daily self reports as an additional source of information. The research process included signal preprocessing, numerical feature extraction, feature selection based on mutual information, and emotion classification into four condition classes using a machine learning model based on XGBoost, which achieved an accuracy of 84.38% and a macro ROC AUC of 0.9310 on the independent test set. The physiological and self report classification results were combined through a weighted fusion mechanism based on reliability, with the entire data pipeline managed through a Python backend using REST API to ensure a stable and consistent synchronization process. The result of this research is an integrated emotion monitoring system that presents informative daily condition visualizations, provides light intervention through a chatbot named Noxi, and is equipped with an admin monitoring website to track users' emotional conditions in real time. Usability testing using the System Usability Scale on 43 respondents yielded an average score of 80.41, which falls into the Excellent category.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Digital Phenotyping, Kesehatan Mental, Multimodal, Smartphone, Wearable Device, REST API, Stres, Kecemasan Digital Phenotyping, Mental Health, Multimodal, Smartphone, Wearable Device, REST API, Stress, Anxiety
Subjects:	R Medicine > R Medicine (General) > R856.2 Medical instruments and apparatus. T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing T Technology > T Technology (General) > T58.6 Management information systems
Divisions:	Faculty of Electrical Technology > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User:	Regina Salsabila Zahirah
Date Deposited:	31 Jul 2026 07:41
Last Modified:	31 Jul 2026 07:41
URI:	http://repository.its.ac.id/id/eprint/141088

Actions (login required)

View Item