Klasifikasi Depresi dengan Pendekatan Random Forest Berbasis Inter-Trial Phase Coherence (ITPC) Pada Multi Kondisi Stimulus Electroencephalogram

Novanti, Fitria (2026) Klasifikasi Depresi dengan Pendekatan Random Forest Berbasis Inter-Trial Phase Coherence (ITPC) Pada Multi Kondisi Stimulus Electroencephalogram. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221054-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Depresi merupakan gangguan mental dengan prevalensi tinggi yang dapat berdampak serius terhadap kualitas hidup dan berkontribusi terhadap risiko bunuh diri. Identifikasi depresi secara klinis masih banyak menggunakan instrumen self-report, seperti Beck Depression Inventory-II (BDI-II), yang dapat dipengaruhi oleh subjektivitas responden. Penelitian ini bertujuan menganalisis dan membandingkan karakteristik Inter-Trial Phase Coherence (ITPC) pada sinyal Electroencephalogram (EEG) antara individu depresi dan normal pada kondisi mata terbuka (open eyes) dan mata tertutup (close eyes), menilai kinerja algoritma Random forest dalam melakukan klasifikasi, serta mengidentifikasi kondisi stimulus dan region otak yang memberikan kontribusi terbesar terhadap proses klasifikasi. Penelitian melibatkan 24 responden yang terdiri atas 12 individu depresi dan 12 individu normal berdasarkan hasil skrining BDI-II. Perekaman EEG dilakukan menggunakan 21 elektroda berdasarkan sistem internasional 10–20. Sinyal EEG melalui tahap preprocessing dan segmentasi, kemudian transformasi Morlet Wavelet digunakan untuk memperoleh informasi fase pada pita frekuensi delta, theta, alpha, beta, dan gamma. Nilai ITPC dihitung dan dianalisis berdasarkan lima region otak, yaitu frontal, central, parietal, temporal, dan occipital. Klasifikasi dilakukan menggunakan algoritma Random forest dengan pendekatan bootstrap dan evaluasi berdasarkan prediksi Out-of-Bag (OOB) terakumulasi. Hasil penelitian menunjukkan bahwa kelompok depresi cenderung memiliki nilai ITPC yang lebih tinggi dibandingkan kelompok normal pada hampir seluruh region otak, yang mengindikasikan sinkronisasi fase aktivitas neural yang lebih kuat. Pemisahan distribusi ITPC antara kelompok depresi dan normal terlihat lebih jelas pada kondisi close eyes. Model Random forest menghasilkan accuracy sebesar 72,53%, precision 72,40%, recall 72,02%, F1-score 72,21%, dan Area Under the Curve (AUC) sebesar 78,86%, yang menunjukkan kemampuan klasifikasi dan diskriminasi yang cukup baik. Berdasarkan analisis feature importance, kondisi close eyes memberikan kontribusi dominan dalam proses klasifikasi, dengan fitur Close_Central_Alpha sebagai fitur paling berpengaruh dengan nilai importance sebesar 0,05. Region central dan frontal juga mendominasi fitur dengan kontribusi tertinggi. Hasil penelitian menunjukkan bahwa fitur ITPC pada kondisi close eyes, khususnya pada region central dan frontal, berpotensi menjadi biomarker EEG untuk mendukung pengembangan sistem deteksi depresi secara lebih objektif.
=======================================================================================================================================
Depression is a mental disorder with a high prevalence that can severely affect quality of life and contribute to the risk of suicide. Clinical identification of depression still largely relies on self-report instruments, such as the Beck Depression Inventory-II (BDI-II), which may be influenced by respondents' subjectivity. This study aims to analyze and compare the characteristics of Inter-Trial Phase Coherence (ITPC) derived from Electroencephalogram (EEG) signals between individuals with depression and healthy individuals under open-eyes and close-eyes conditions, evaluate the performance of the Random forest algorithm in classifying depression and normal conditions, and identify the stimulus conditions and brain regions that contribute most to the classification process. The study involved 24 participants, consisting of 12 individuals with depression and 12 healthy individuals, based on BDI-II screening results. EEG signals were recorded using 21 electrodes based on the international 1020 system. The preprocessing and segmentation, followed by Morlet Wavelet transformation to obtain phase information across the delta, theta, alpha, beta, and gamma frequency bands. ITPC values were calculated and analyzed across five brain regions: frontal, central, parietal, temporal, and occipital. Classification was performed using the Random forest algorithm with a bootstrap approach and evaluated based on accumulated Out-of-Bag predictions. The results showed that individuals with depression tended to exhibit higher ITPC values than healthy individuals across nearly all brain regions, indicating stronger neural phase synchronization. The separation of ITPC distributions between the two groups was more apparent under the close-eyes condition. The Random forest model achieved an accuracy of 72.53%, precision of 72.40%, recall of 72.02%, F1-score of 72.21%, and an Area Under the Curve (AUC) of 78.86%, indicating reasonably good classification and discrimination performance. Feature importance analysis revealed that the close-eyes condition contributed dominantly to the classification process, with Close_Central_Alpha identified as the most influential feature, achieving an importance value of 0.05. The central and frontal regions also dominated the features with the highest contributions. These findings indicate that ITPC features under the close-eyes condition, particularly in the central and frontal regions, have the potential to serve as EEG-based biomarkers to support the development of a more objective depression detection system.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Depresi, EEG, Feature Importance, ITPC, Random Forest
Subjects:	Q Science > QP Physiology > Q376.5 Electroencephalography (EEG)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Fitria Novanti
Date Deposited:	04 Aug 2026 06:55
Last Modified:	04 Aug 2026 06:55
URI:	http://repository.its.ac.id/id/eprint/143193

Actions (login required)

View Item