An Explainable State-Space Transformer Framework for Predicting Clinical Deterioration in Non-Pulmonary Tuberculosis Patients
Ingénierie des systèmes d information · 30.06.2026 · 31 (6) · 2011-2022
DOI: 10.18280/isi.310622 ↗Kurz gefasst
Fachartikel in der Zeitschrift Ingénierie des systèmes d information, online erschienen am 30.06.2026. Verfasst von Ruqaia J. Kadhim und 4 weiteren Autor:innen.
Abstract (Originalsprache)
Non-pulmonary tuberculosis (NPTB) remains challenging to manage due to heterogeneous clinical manifestations and the difficulty of predicting patient deterioration from longitudinal clinical observations. Early identification of high-risk patients is particularly important in resource-limited settings, where timely intervention can improve clinical decision-making. This study proposes an explainable hybrid framework that integrates a state-space model (SSM) with a Transformer-based architecture for predicting clinical deterioration in NPTB patients. A real-world time-series dataset consisting of 125 patient records was used for model development and evaluation. The preprocessing pipeline included temporal feature extraction, class balancing using the Synthetic Minority Over-Sampling Technique (SMOTE), and sequence-based representation learning. The proposed hybrid model was compared with conventional machine learning methods, including Logistic Regression, Random Forest, and Long Short-Term Memory (LSTM) networks. Experimental results demonstrated that the proposed framework achieved superior predictive performance, with an area under the receiver operating characteristic curve (ROC) of 0.93, F1-score of 0.86, recall of 0.85, and Brier score of 0.14. Statistical analysis using DeLong and McNemar tests confirmed significant improvements over baseline models. Furthermore, SHapley Additive exPlanations (SHAP) analysis and Transformer attention visualization provided insights into the temporal features contributing to risk prediction. Although external validation using larger multicenter cohorts is required, the proposed framework provides a potential approach for interpretable artificial intelligence (AI)-assisted risk stratification in NPTB management.
Zitieren
APA
Kadhim, R. J., Alsaedi, H. J., Hussain, A. S., Tashtoush, M. A., & Almuttalibi, R. A. (2026). An Explainable State-Space Transformer Framework for Predicting Clinical Deterioration in Non-Pulmonary Tuberculosis Patients. Ingénierie des systèmes d information, 31(6), 2011-2022. https://doi.org/10.18280/isi.310622
Vancouver
Kadhim RJ, Alsaedi HJ, Hussain AS, Tashtoush MA, Almuttalibi RA. An Explainable State-Space Transformer Framework for Predicting Clinical Deterioration in Non-Pulmonary Tuberculosis Patients. Ingénierie des systèmes d information. 2026;31(6):2011-2022. doi:10.18280/isi.310622
BibTeX
@article{Kadhim2026,
author = {Kadhim, Ruqaia J. and Alsaedi, Hasanain J. and Hussain, Adel S. and Tashtoush, Mohammad A. and Almuttalibi, Rana A.},
title = {An Explainable State-Space Transformer Framework for Predicting Clinical Deterioration in Non-Pulmonary Tuberculosis Patients},
journal = {Ingénierie des systèmes d information},
year = {2026},
volume = {31},
number = {6},
pages = {2011--2022},
doi = {10.18280/isi.310622}
}
EUVIMED