Mental health disorders such as suicidal ideation, bipolar disorder, depression, and anxiety pose public health challenges in Turkey. Approximately 20% of the population is estimated to be affected by some form of mental illness. Social media is used for sharing personal experiences and opinions in Turkey. This presents an opportunity to use textual data for detecting mental health disorders. This study aims to examine the predictive performance of various machine learning classifiers for identifying mental health conditions in Turkish language texts from social media. Given the lack of structured and unstructured datasets focused on Turkish mental health, we have mined and processed textual data from the social media platform Reddit. This led to the creation of the Turkish Mental Health Disorder (TMHD) Corpus. The TMHD dataset and the machine learning experiments conducted aim to improve the understanding of predicting mental health issues in Turkey using Artificial Intelligence.
Courage Armah, Z. Erdem, Halil Bisgin· Signal Processing and Commun...· 0 citations
Accurate prediction of protein-protein interaction interfaces is critical for understanding molecular recognition and guiding therapeutic design. This study presents a comprehensive machine learning pipeline for predicting interface residues in permanent homodimeric protein complexes. Using a curated dataset of 1311 homodimers, we benchmarked six widely used machine learning algorithms and identified multilayer perceptron and XGBoost as top performers, achieving Matthews correlation coefficients (MCC) exceeding 0.93. To enhance interpretability and efficiency, we employed recursive feature elimination to derive a minimal set of six biologically meaningful features, including solvent accessibility, surface roughness, planarity, and average protrusion index, that retained high predictive power (MCC > 0.90). Structurally stratified models tailored to α-helical, β-strand, and membrane proteins demonstrated comparable or improved accuracy relative to generalized models, particularly when utilizing the reduced feature subset. As a preliminary demonstration of generalizability, we applied our approach to an external heterodimer complex (PDB ID: 9ETL). While limited to a single case study, the structurally specialized models maintained high accuracy, suggesting potential applicability beyond the training domain. Furthermore, our residue-level feature-driven models demonstrated highly competitive performance when compared against the baseline established by the general-purpose ColabFold pipeline. The results highlight the importance of structural context in interface prediction and demonstrate that compact, structure-aware models can achieve high accuracy while reducing computational complexity. This work provides a scalable, interpretable, and biologically informed approach to protein interface prediction, with implications for large-scale structural descriptor, drug target characterization, and protein engineering applications.
Tayyip Topuz, Z. Erdem, Halil Bisgin et al.· Scientific Reports· 0 citations
