CAD4TB software updates: different triaging thresholds require caution by users and regulation by authorities

Fehr, Jana; Gunda, Resign; Siedner, Mark J.; Hanekom, Willem A.; ́u, T. Ndung; Grant, Alison D.; Lippert, Christoph; Wong, Emily

doi:10.5588/ijtld.22.0437

Cited by 6 publications

(6 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In conclusion, we have demonstrated the need for context-specific threshold determination, supporting recent calls for regulation of companies to provide guidance for effective adoption of CAD with appropriate oversight by WHO [ 30 , 42 ]. In our setting, a CAD4TBv7 threshold of 13 was close to WHO target product profile criteria.…”

Section: Discussionsupporting

confidence: 75%

“…However, the study population was very different, being people who presented or were referred to tuberculosis screening centres in Bangladesh with 98% having symptoms. In the second, a community-based study in South Africa restricted to participants with sputum test results, a threshold of 20 was estimated (to most closely match radiologist sensitivity of 81%, with specificity 57%) [ 42 ]. This highlights the importance of threshold determination needing to be context specific.…”

Section: Discussionmentioning

confidence: 99%

“…In the second, a community-based study in South Africa restricted to participants with sputum test results, a threshold of 20 was estimated (to most closely match radiologist sensitivity of 81%, with specificity 57%). 42 This highlights the importance of threshold determination needing to be context specific.…”

Section: Discussionmentioning

confidence: 99%

See 2 more Smart Citations

Computer-aided detection thresholds for digital chest radiography interpretation in tuberculosis diagnostic algorithms

Vanobberghen,

Keter,

Jacobs

et al. 2023

ERJ Open Res

View full text Add to dashboard Cite

ObjectivesUse of computer-aided detection (CAD) software is recommended to improve tuberculosis screening and triage, but threshold determination is challenging if reference testing has not been performed in all individuals. We aimed to determine such thresholds through secondary analysis of the 2019 Lesotho national tuberculosis prevalence survey.MethodsSymptom screening and chest X-ray (CXR) were performed in participants aged ≥15 years; those symptomatic or with abnormal CXR provided samples for Xpert MTB/RIF and culture testing. CXRs were processed using CAD4TB version 7. We used six methodological approaches to deal with participants who did not have bacteriological test results to estimate pulmonary tuberculosis prevalence and assess diagnostic accuracy.ResultsAmong 17 070 participants, 5214 (31%) had tuberculosis status determined; 142 had tuberculosis. Prevalence estimates varied between methodological approaches (0.83–2.72%). Using multiple imputation to estimate tuberculosis status for those eligible but not tested, and assuming those not eligible for testing were negative, a CAD4TBv7 threshold of 13 had sensitivity 89.7% (95% confidence interval 84.6–94.8) and specificity 74.2% (73.6–74.9), close to WHO target product profile criteria. Assuming all those not tested were negative produced similar results.ConclusionsThis is the first study to evaluate CAD4TB in a community-screening context employing a range of approaches to account for unknown tuberculosis status. The assumption that those not tested are negative – regardless of testing eligibility status – was robust. As threshold determination must be context-specific, our analytically-straightforward approach should be adopted to leverage prevalence surveys for CAD threshold determination in other settings with a comparable proportion of eligible but not tested participants.

show abstract

Section: Discussionsupporting

confidence: 75%

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Computer-aided detection thresholds for digital chest radiography interpretation in tuberculosis diagnostic algorithms

Vanobberghen,

Keter,

Jacobs

et al. 2023

ERJ Open Res

View full text Add to dashboard Cite

show abstract

“…Dangers of changes in thresholds with version updates have also been pointed in the recent literature. 15 …”

Section: Resultsmentioning

confidence: 99%

Tweaking algorithms. Technopolitical issues associated with artificial intelligence based tuberculosis detection in global health

David,

Onno,

Pourraz

et al. 2024

DIGITAL HEALTH

View full text Add to dashboard Cite

Computer-aided detection algorithms based on artificial intelligence are increasingly being tested and used as a means for detecting tuberculosis in countries where the epidemic is still present. Computer-aided detection tools are often presented as a global solution that can be deployed in all the geographical areas concerned by tuberculosis, but at the same time, they need to be adjusted and calibrated according to local populations’ characteristics. The aim of this article is to analyze the tensions between the standardization of computer-aided detection algorithms and their local adaptation and the political issues associated with these tensions. We undertook a qualitative analysis of practices associated with tuberculosis detection algorithms in different contexts, contrasting the perspectives of various stakeholders. Algorithms embed the promise of standardization through automation and the bypassing of variable human expertise such as that of radiologists, they are nonetheless objects of local practices that we have characterized as “tweaking.” This work of tweaking reveals how the technology is situated but also the many concerns of the users and workers (insertion in care, control over infrastructure, and political ownership). This should be better considered to truly make computer-aided detection innovative tools for tuberculosis management in global health.

show abstract

“…Any participant who answered yes to any of the WHO-4SS questions, who had a chest x-ray score ≥ 25 (CAD4TBv5), who had abnormal lung fields (as determined by the radiologist), or who was unable to undergo chest x-ray was offered sputum collection. A CAD4TB score reflects the degree of abnormality and likelihood of TB, but there is no universally accepted CAD4TBv5 score to use in a TB screening setting [ 13 ]. In a process previously detailed by Fehr et al [ 14 ], a CAD4TBv5 score of 25 threshold was selected to maximize TB case-finding in this study.…”

Section: Methodsmentioning

confidence: 99%

Subclinical tuberculosis linkage to care and completion of treatment following community-based screening in rural South Africa

Sifumba,

Claassen,

Olivier

et al. 2024

BMC Global Public Health

View full text Add to dashboard Cite

Background Tuberculosis (TB), a leading cause of infectious death, is curable when patients complete a course of multi-drug treatment. Because entry into the TB treatment cascade usually relies on symptomatic individuals seeking care, little is known about linkage to care and completion of treatment in people with subclinical TB identified through community-based screening. Methods Participants of the Vukuzazi study, a community-based survey that provided TB screening in the rural uMkhanyakude district of KwaZulu-Natal from May 2018 – March 2020, who had a positive sputum (GeneXpert or Mtb culture, microbiologically-confirmed TB) or a chest x-ray consistent with active TB (radiologically-suggested TB) were referred to the public health system. Telephonic follow-up surveys were conducted from May 2021 – January 2023 to assess linkage to care and treatment status. Linked electronic TB register data was accessed. We analyzed the effect of baseline HIV and symptom status (by WHO 4-symptom screen) on the TB treatment cascade. Results Seventy percent (122/174) of people with microbiologically-confirmed TB completed the telephonic survey. In this group, 84% (103/122) were asymptomatic and 46% (56/122) were people living with HIV (PLWH). By self-report, 98% (119/122) attended a healthcare facility after screening, 94% (115/122) started TB treatment and 93% (113/122) completed treatment. Analysis of electronic TB register data confirmed that 67% (116/174) of eligible individuals started TB treatment. Neither symptom status nor HIV status affected linkage to care. Among people with radiologically-suggested TB, 48% (153/318) completed the telephonic survey, of which 80% (122/153) were asymptomatic and 52% (79/153) were PLWH. By self-report, 75% (114/153) attended a healthcare facility after screening, 16% (24/153) started TB treatment and 14% (22/153) completed treatment. Nine percent (28/318) of eligible individuals had TB register data confirming that they started treatment. Conclusions Despite high rates of subclinical TB, most people diagnosed with microbiologically-confirmed TB after community-based screening were willing to link to care and complete TB treatment. Lower rates of linkage to care in people with radiologically-suggested TB highlight the importance of streamlined care pathways for this group. Clearer guidelines for the management of people who screen positive during community-based TB screening are needed.

show abstract

CAD4TB software updates: different triaging thresholds require caution by users and regulation by authorities

Cited by 6 publications

References 11 publications

Computer-aided detection thresholds for digital chest radiography interpretation in tuberculosis diagnostic algorithms

Computer-aided detection thresholds for digital chest radiography interpretation in tuberculosis diagnostic algorithms

Tweaking algorithms. Technopolitical issues associated with artificial intelligence based tuberculosis detection in global health

Subclinical tuberculosis linkage to care and completion of treatment following community-based screening in rural South Africa

Contact Info

Product

Resources

About