A new resource on artificial intelligence powered computer automated detection software products for tuberculosis programmes and implementers.

Recently, the number of artificial intelligence powered computer-aided detection (CAD) products that detect tuberculosis (TB)-related abnormalities from chest X-rays (CXR) available on the market has increased. Although CXR is a relatively effective and inexpensive method for TB screening and triaging, a shortage of skilled radiologists in many high TB-burden countries limits its use. CAD technology offers a solution to this problem. Before adopting a CAD product, TB programmes need to consider not only the diagnostic accuracy but also implementation-relevant features including operational characteristics, deployment mechanism, input and machine compatibility, output format, options for integration into the legacy system, costs, data sharing and privacy aspects, and certification. A landscaping analysis was conducted to collect this information among CAD developers known to have or soon to have a TB product. The responses were reviewed and finalized with the developers, and are published on an open-access website: www.ai4hlth.org. CAD products are constantly being improved and the site will continuously be updated to account for updates and new products. This unique online resource aims to inform the TB community about available CAD tools, their features and set-up procedures, to enable TB programmes to identify the most suitable product to incorporate in interventions.

Deep learning, computer-aided radiography reading for tuberculosis: a diagnostic accuracy study from a tertiary hospital in India.

In general, chest radiographs (CXR) have high sensitivity and moderate specificity for active pulmonary tuberculosis (PTB) screening when interpreted by human readers. However, they are challenging to scale due to hardware costs and the dearth of professionals available to interpret CXR in low-resource, high PTB burden settings. Recently, several computer-aided detection (CAD) programs have been developed to facilitate automated CXR interpretation. We conducted a retrospective case-control study to assess the diagnostic accuracy of a CAD software (qXR, Qure.ai, Mumbai, India) using microbiologically-confirmed PTB as the reference standard. To assess overall accuracy of qXR, receiver operating characteristic (ROC) analysis was used to determine the area under the curve (AUC), along with 95% confidence intervals (CI). Kappa coefficients, and associated 95% CI, were used to investigate inter-rater reliability of the radiologists for detection of specific chest abnormalities. In total, 317 cases and 612 controls were included in the analysis. The AUC for qXR for the detection of microbiologically-confirmed PTB was 0.81 (95% CI: 0.78, 0.84). Using the threshold that maximized sensitivity and specificity of qXR simultaneously, the software achieved a sensitivity and specificity of 71% (95% CI: 66%, 76%) and 80% (95% CI: 77%, 83%), respectively. The sensitivity and specificity of radiologists for the detection of microbiologically-confirmed PTB was 56% (95% CI: 50%, 62%) and 80% (95% CI: 77%, 83%), respectively. For detection of key PTB-related abnormalities 'pleural effusion' and 'cavity', qXR achieved an AUC of 0.94 (95% CI: 0.92, 0.96) and 0.84 (95% CI: 0.82, 0.87), respectively. For the other abnormalities, the AUC ranged from 0.75 (95% CI: 0.70, 0.80) to 0.94 (95% CI: 0.91, 0.96). The controls had a high prevalence of other lung diseases which can cause radiological manifestations similar to PTB (e.g., 26% had pneumonia, 15% had lung malignancy, etc.). In a tertiary hospital in India, qXR demonstrated moderate sensitivity and specificity for the detection of PTB. There is likely a larger role for CAD software as a triage test for PTB at the primary care level in settings where access to radiologists in limited. Larger prospective studies that can better assess heterogeneity in important subgroups are needed.

A bibliometric analysis of tuberculosis research, 2007-2016.

BACKGROUND: Tuberculosis (TB) research is a key component of the End TB Strategy. To track research output, we conducted a bibliometric analysis of TB research from the past decade. METHODS: The Web of Science database was searched for publications from January 2007 to December 2016 with "tuberculosis" in the title. References were analysed using the R bibliometrix package. A year-stratified 5% random subset was drawn to extract funding sources and identify research areas. FINDINGS: The annual growth rate of publications was 7.3%, and was highest (13.1%) among Brazil, Russia, India, China and South Africa (BRICS). The USA was the most productive country, with 18.4% of references, followed by India (9.7%), China (7.3%), England (6.5%), and South Africa (3.9%). In the subset analysis, the most common research area was 'fundamental research' (33.8%). Frequently acknowledged funders were US and EU-based, with China and India emerging as top funders. Collaborations appeared more frequently between high-income countries and low/medium income countries (LMICs), with fewer collaborations among LMICs. CONCLUSION: The past decade has seen a continued increase in TB publications. While USA continues to dominate research output and funding, BRICS countries have emerged as major research producers and funders. Collaborations among BRICS would enhance future TB research productivity.

Performance of the Xpert HIV-1 Viral Load Assay: a Systematic Review and Meta-analysis.

Viral load (VL) is the preferred treatment-monitoring approach for HIV-positive patients. However, more rapid, near-patient, and low-complexity assays are needed to scale up VL testing. The Xpert HIV-1 VL assay (Cepheid, Sunnyvale, CA) is a new, automated molecular test, and it can leverage the GeneXpert systems that are being used widely for tuberculosis diagnosis. We systematically reviewed the evidence on the performance of this new tool in comparison to established reference standards. A total of 12 articles (13 studies) in which HIV patient VLs were compared between Xpert HIV VL assay and a reference standard VL assay were identified. Study quality was generally high, but substantial variability was observed in the number and type of agreement measures reported. Correlation coefficients between Xpert and reference assays were high, with a pooled Pearson correlation (n = 8) of 0.94 (95% confidence interval [CI], 0.89, 0.97) and Spearman correlation (n = 3) of 0.96 (95% CI, 0.86, 0.99). Bland-Altman metrics (n = 11) all were within 0.35 log copies/ml of perfect agreement. Overall, Xpert HIV-1 VL performed well compared to current reference tests. The minimal training and infrastructure requirements for the Xpert HIV-1 VL assay make it attractive for use in resource-constrained settings, where point-of-care VL testing is most needed.