| Title: | Advancing AI-Based depression detection : a preliminary study on feature optimization and model robustness |
|---|
| Authors: | ID Zorko, Albert (Author) |
|---|
| Files: |  RAZ_Zorko_Albert_2025.pdf (12,52 MB)
MD5: 2EBEEB33F12E23AE224DC803CB606675
|
|---|
| Language: | English |
|---|
| Work type: | Unknown |
|---|
| Typology: | 1.08 - Published Scientific Conference Contribution |
|---|
| Organization: | FIŠ - Faculty of Information Studies in Novo mesto
|
|---|
| Abstract: | This study constitutes the second part of our investigation presented at ITIS 2023, which explores the search for objective physiological biomarkers for major depressive disorder (MDD). Moving beyond the established role of Heart Rate Variability (HRV), this preliminary research focuses on Pulse-Respiratory Coupling (PRC) – the coordination between cardiac and respiratory rhythms. We hypothesize that depression, characterized by autonomic nervous system (ANS) dysregulation, disrupts this coupling. A group of 73 subjects (healthy controls, untreated depressed patients, and patients treated with tricyclic antidepressants) were submitted to simultaneous electrocardiogram (EKG) and respiratory recording. Analysis revealed a distinct degradation of PRC in the depressed group, manifesting as a loss of synchronous patterns observed in healthy subjects. Machine learning models were trained on features derived from PRC timing. The k-Nearest Neighbors algorithm achieved a promising classification accuracy of 97.3% in distinguishing depressed from healthy individuals, outperforming other classifiers like Random Forest (95.9%) and Support Vector Machine (95.9%). While these results are preliminary and require validation in larger cohorts, they strongly suggest that PRC is a sensitive, non-invasive marker of ANS dysfunction in depression. This work underscores the potential of integrating multi-system physiological analysis with artificial intelligence to create objective aids for psychiatric diagnosis. |
|---|
| Keywords: | major depressive disorder, physiological biomarkers, pulse-respiratory coupling, heart rate variability, autonomic nervous system, machine learning |
|---|
| Publication status: | Published |
|---|
| Publication version: | Author Accepted Manuscript |
|---|
| Publication date: | 16.12.2026 |
|---|
| Year of publishing: | 2025 |
|---|
| Number of pages: | Str. [12-21] |
|---|
| PID: | 20.500.12556/ReVIS-13032  |
|---|
| COBISS.SI-ID: | 264894723 |
|---|
| UDC: | 616.89-008.454:612:004.8 |
|---|
| Note: | Nasl. z nasl. zaslona;
Opis vira z dne 15. 1. 2026;
|
|---|
| Publication date in ReVIS: | 22.01.2026 |
|---|
| Views: | 316 |
|---|
| Downloads: | 4 |
|---|
| Metadata: |    |
|---|
|
:
|
Copy citation |
|---|
| | | | Share: |  |
|---|
Hover the mouse pointer over a document title to show the abstract or click
on the title to get all document metadata. |
