Algorithmic Determinants of Performance Heterogeneity in Whole-Genome Sequencing-Based Prediction of Drug Resistance in Mycobacterium tuberculosis: A Systematic Review and Meta-Analysis
Microorganisms·
- DOI
- 10.3390/microorganisms14081824
- PMID
- —
- PMCID
- —
- OpenAlex
- W7203721573
- Study type
- Systematic review
- Publisher
- MDPI AG
- Article type
- journal-article
- Integrity
- current
Why this research matters now
Reducing diagnostic uncertainty for clinical laboratories could facilitate more consistent implementation of genomic resistance testing. Standardized predictive accuracy may support uniform laboratory reporting practices and improve workflow reliability.
Structured evidence summary
Research question
What is the diagnostic accuracy of bioinformatic algorithms applied to whole-genome sequencing for detecting first-line drug resistance in Mycobacterium tuberculosis?
Study design
A prospectively registered, PRISMA-DTA-compliant systematic review and meta-analysis evaluating diagnostic accuracy studies. Primary outcomes were synthesized using a bivariate random-effects model, supplemented by subgroup analyses and meta-regression.
Population and setting
Information extracted from published diagnostic accuracy studies conducted in clinical laboratory environments managing tuberculosis specimens across varied operational contexts.
Main findings
Aggregated results indicated strong overall specificity and sensitivity for first-line drug resistance prediction. Rifampicin and isoniazid assays outperformed pyrazinamide testing. Hybrid computational models showed marginally higher sensitivity estimates than rule-based or machine learning alternatives, though these distinctions remained exploratory.
Public-health relevance
Reducing diagnostic uncertainty for clinical laboratories could facilitate more consistent implementation of genomic resistance testing. Standardized predictive accuracy may support uniform laboratory reporting practices and improve workflow reliability.
Important limitations
Categorical performance comparisons were restricted to seven analytical units, making it impossible to isolate algorithmic architecture from tool-specific features, training datasets, reference mutation libraries, or validation cohorts. Reported differences between computational categories must not be construed as causal.
GIDS interpretation
The publication appears in a peer-reviewed journal focusing on microbial diagnostics. Its systematic methodology provides a consolidated view of published algorithmic performance metrics, aiding researchers navigating the current literature landscape without implying real-time monitoring capabilities.
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 5 auditable classifier relationships to diseases, places, topics, and study design.