Artificial intelligence and machine learning applications in abdominal tuberculosis diagnosis: A scoping review and translational roadmap
Artificial Intelligence in Gastroenterology·
- DOI
- 10.35712/aig.123751
- PMID
- —
- PMCID
- —
- OpenAlex
- W7166723993
- Study type
- Systematic review
- Publisher
- Baishideng Publishing Group Inc.
- Article type
- journal-article
- Integrity
- current
Why this research matters now
Improving detection accuracy for abdominal tuberculosis in resource-constrained environments could mitigate diagnostic delays and support more standardized care delivery pathways.
Structured evidence summary
Research question
The review seeks to catalog current computational approaches for identifying abdominal tuberculosis and outline steps for fair clinical implementation.
Study design
Researchers performed a database search through April 2026, applied dual independent screening protocols, evaluated methodological quality using adapted risk assessment tools, and synthesized results narratively.
Population and setting
The analyzed cohorts were predominantly drawn from Chinese institutions, with sparse representation from other regions and insufficient focus on individuals living with HIV or harboring resistant pathogen strains.
Main findings
Computational algorithms successfully distinguished intestinal tuberculosis from Crohn’s disease across fourteen investigations, achieving discrimination metrics between 0.80 and 0.96, with combined data streams and transparent architectures yielding superior performance.
Public-health relevance
Improving detection accuracy for abdominal tuberculosis in resource-constrained environments could mitigate diagnostic delays and support more standardized care delivery pathways.
Important limitations
The compiled literature depends largely on composite reference standards, consists mainly of retrospective analyses, and suffers from narrow geographic distribution alongside modest participant numbers.
GIDS interpretation
This article supplies a curated taxonomy of algorithmic diagnostic strategies and a structured implementation framework, which may serve as searchable reference points for subsequent literature mapping initiatives.
Related GIDS surveillance
Literature context does not validate, explain, or change a surveillance signal. Exact and contextual relationships are shown separately.
Evidence relationships
This article has 8 auditable classifier relationships to diseases, places, topics, and study design.