Integration of AI and ML in Tuberculosis (TB) Management: From Diagnosis to Drug Discovery

Abstract

Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis. Despite the improvements in diagnostic techniques, the accuracy of TB diagnosis is still low. In recent years, the development of artificial intelligence (AI) has opened up new possibilities in diagnosing and treating TB with high accuracy compared to traditional methods. Traditional diagnostic techniques, such as sputum smear microscopy, culture tests, and chest X-rays, are time-consuming, with less sensitivity for the detection of TB in patients. Due to the new developments in AI, advanced diagnostic and treatment techniques have been developed with high accessibility, speed, and accuracy. AI, including various specific methodologies, is becoming vital in managing TB. Machine learning (ML) methodologies, such as support vector machines (SVMs) and random forests (RF), alongside deep learning (DL) technologies, particularly convolutional neural networks (CNNs) for image analysis, are employed to analyze diverse patient data, including medical images and biomarkers, to enhance the accuracy and speed of tuberculosis diagnosis. This study summarized the benefits and drawbacks of both traditional and AI-driven TB diagnosis, highlighting how AI can support traditional techniques to increase early detection, lower misdiagnosis, and strengthen international TB control initiatives.

Keywords: artificial intelligence; chest X-rays; culture tests; deep learning; machine learning; sputum smear microscopy; tuberculosis.

Figure 1

Percentage of diagnosed TB cases in seven different countries.

Figure 2

Overview of traditional diagnostic methods for tuberculosis. The image illustrates many approved methodologies for TB diagnosis: direct visualization of acid-fast bacilli via sputum smear microscopy; mycobacterial culture utilizing a liquid culture system (e.g., BACTEC MGIT tubes) typically employed for isolating Mycobacterium tuberculosis; chest X-ray (CXR) for detecting pulmonary abnormalities indicative of tuberculosis; and nucleic acid amplification tests (NAATs) for the molecular identification of Mycobacterium tuberculosis.

Figure 3

Conceptual framework illustrating critical data areas in TB management where AI can be applied. This illustration highlights potential integration points for AI and ML, highlighting the many data modalities involved in TB diagnosis and treatment, including (A) chest X-ray imaging, (B) CT scans, (C) genetic data, (D) bacteriological evidence, and (E) molecular diagnostics. The illustration is intended to provide a broad visual representation of the kind of data generated throughout TB care pathways where AI can be beneficial, rather than presenting analytical results.

Figure 4

Artificial intelligence (AI) in the anti-tuberculosis drug development process at integration. This image illustrates the enhancement of traditional drug development methods with artificial intelligence and machine learning (ML). Significant phases augmented by artificial intelligence and machine learning encompass AI-based target prediction, AI-based protein screening, AI-driven toxicity and side effects predictions, ML-based binding analysis, and AI in clinical trial design and patient stratification. These computational techniques aim to accelerate the testing, identification, and optimization of novel anti-TB therapies.