Tesis

Optimization 2D & 5D measurements for depression diagnosis: an evaluation of analytical techniques and detection accuracy by machine learning, and deep learning / Al Fathjri Wisesa

Abstrak
Measurement is the cornerstone of scientific development including in the design of instruments capable of objectively assessing depressive conditions. In this context an accurate precise and reliable measurement system is essential for validly mapping such conditions. Principles of measurement in physics such as calibration accuracy and precision can be adopted to design more objective tools for assessing mental health conditions. This study aims to develop and evaluate a measurement system for depression levels in students through the integration of non-invasive physical parameters and psychological parameters. A predictive model was built using machine learning (ML) and deep learning (DL) approaches resulting in accurate and holistic measurements. This research method uses a quantitative approach with secondary data analysis from Data in Brief publications which contain physical parameters including heart rate (BPM) oxygen saturation (SpO ) body temperature and sleep duration and psychological parameters obtained through the Perceived Stress Scale (PSS-10) questionnaire. The research process includes data standardization data relationships modeling using algorithms and performance evaluation. A 2D and 5D input configuration was used to assess the effectiveness of feature combinations in depression level classification. The algorithms used in this study include Logistic Regression Random Forest and Support Vector Machine (SVM) as representations of the Machine Learning (ML) approach as well as Artificial Neural Networks (ANNs) as representations of the Deep Learning (DL) approach. Model performance evaluation was conducted using metrics. ML/DL is a computational approach capable of analyzing data without explicit synthetic data programming. The analysis results showed that the 5D configuration provided a significant improvement in accuracy compared to 2D with Random Forest demonstrating the most consistent and superior performance. This research contributes to the development of an integrated physiological and psychological depression measurement system leading to the development of a more objective accurate and easy-to-use depression measurement tool.

Informasi Detail

DDC

Rt 530.76 WIS o

Prodi

Universitas Negeri Malang. Program Studi Fisika, 2025.

Deskripsi Fisik

xiii, 187 lembar. : ilus. ; 30 cm.

Bahasa

No Reg

00682/RT/25

Edisi

Tesis (Pascasarjana)--Universitas Negeri Malang. 2025

Subjek

1. FISIKA - MODEL PEMBELAJARAN
2. DIAGNOSIS DEPRESI ( 2D & 5D ) - EVALUASI TEKNIK ANALITIS
3. PHYSICS - LEARNING MODEL

Pembimbing

1. Dr. Eny Latifah, S.si, M.si; 2. Dr. Hari Wisodo, S.pd, M.si

Lampiran Berkas