Artificial intelligence and diagnosis and management of tuberculosis disease in children.
Authors
Affiliations (9)
Affiliations (9)
- Operational Research Center in Healthcare.
- Department of Biomedical Engineering, Near East University, Nicosia/TRNC, Mersin 10, Turkey.
- Department of Mathematical Sciences, Saveetha School of Engineering, SIMATS, Chennai, Tamil Nadu, India.
- Department of Medical Laboratory Science, College of Health Sciences, Federal University, Lokoja, Kogi State, Nigeria.
- Department of Medical Diagnostic Imaging, College of Health Science.
- Research Institute for Medical and Health Sciences, University of Sharjah, Sharjah, UAE.
- Department of Nursing Sciences, Faculty of Allied Health Sciences, Federal University Lokoja.
- Department of Histopathology, Medical Laboratory Services, Federal Teaching Hospital Lokoja, Kogi State, Nigeria.
- Department of Mathematics, Near East University, Nicosia/TRNC.
Abstract
The literature review is pertinent because diagnosing pediatric tuberculosis (PdTB) remains quite challenging, especially in areas with limited resources, due to complications caused by variable generalized symptoms, paucibacillary characteristics, vague clinical manifestations, and challenges associated with pediatric sputum sample production. Recent developments in artificial intelligence have the potential to enhance the accuracy of diagnoses and the effectiveness of treatments. Nineteen published studies between January 2024 and July 2025 were examined, which focused on artificial intelligence driven chest X-ray (CXR) examination and medical prediction. The reviewed studies utilized convolutional neural networks (CNN), transfer learning, and stacked ensemble machine learning (SEML) to achieve sensitivity values ranging from 76.0 to 98.2%, specificity of 70.0 to 98.0%, and area under the curve (AUC) values of as high as 0.98 in AI-CXR diagnosis for the detection of PdTB. Through continuous experiments and use of the AI-CXR triage in Ethiopia (2025), successfully identifying over 30% of patients, while prediction models indicate 82% hepatotoxicity concerns in Nigerian cohorts. Plasma proteomics and exhaled breath analysis are emerging methodologies that exhibit potential; however, pediatric datasets are limited, necessitating multicenter validation. Artificial intelligence enhances the diagnosis and treatment prediction of PdTB in resource-constrained settings. The integration of artificial intelligence with existing diagnostic tools like GeneXpert and telemedicine strategies can significantly improve the efficiency of screening processes. Future research efforts should prioritize the expansion of pediatric datasets and the evaluation of multimodal AI-PdTB approaches.