There have been few independent evaluations of computer-aided detection (CAD) software for tuberculosis (TB) screening, despite the rapidly expanding array of available CAD solutions. We developed a test library of chest X-ray (CXR) images which was blindly re-read by two TB clinicians with different levels of experience and then processed by 12 CAD software solutions. Using Xpert MTB/RIF results as the reference standard, we compared the performance characteristics of each CAD software against both an Expert and Intermediate Reader, using cut-off thresholds which were selected to match the sensitivity of each human reader. Six CAD systems performed on par with the Expert Reader (Qure.ai, DeepTek, Delft Imaging, JF Healthcare, OXIPIT, and Lunit) and one additional software (Infervision) performed on par with the Intermediate Reader only. Qure.ai, Delft Imaging and Lunit were the only software to perform significantly better than the Intermediate Reader. The majority of these CAD software showed significantly lower performance among participants with a past history of TB. The radiography equipment used to capture the CXR image was also shown to affect performance for some CAD software. TB program implementers now have a wide selection of quality CAD software solutions to utilize in their CXR screening initiatives.

Across Asia, a large proportion of people with tuberculosis (TB) do not report symptoms, have mild symptoms or only experience symptoms for a short duration. These individuals may not seek care at health facilities or may be missed by symptom screening, resulting in sustained TB transmission in the community. We evaluated the yields of TB from 114 days of community-based, mobile chest X-ray (CXR) screening. The yields at each step of the TB screening cascade were tabulated and we compared cohorts of participants who reported having a prolonged cough and those reporting no cough or one of short duration. We estimated the marginal yields of TB using different diagnostic algorithms and calculated the relative diagnostic costs and cost per case for each algorithm. A total of 34,529 participants were screened by CXR, detecting 256 people with Xpert-positive TB. Only 50% of those diagnosed with TB were detected among participants reporting a prolonged cough. The study’s screening algorithm detected almost 4 times as much TB as the National TB Program’s standard diagnostic algorithm. Community-based, mobile chest X-ray screening can be a high yielding strategy which is able to identify people with TB who would likely otherwise have been missed by existing health services.

X-ray screening is an important tool in tuberculosis (TB) prevention and care, but access has historically been restricted by its immobile nature. As recent advancements have improved the portability of modern X-ray systems, this study represents an early evaluation of the safety, image quality and yield of using an ultra-portable X-ray system for active case finding (ACF). We reported operational and radiological performance characteristics and compared image quality between the ultra-portable and two reference systems. Image quality was rated by three human readers and by an artificial intelligence (AI) software. We deployed the ultra-portable X-ray alongside the reference system for community-based ACF and described TB care cascades for each system. The ultra-portable system operated within advertised specifications and radiologic tolerances, except on X-ray capture capacity, which was 58% lower than the reported maximum of 100 exposures per charge. The mean image quality rating from radiologists for the ultra-portable system was significantly lower than the reference (3.71 vs. 3.99, p < 0.001). However, we detected no significant differences in TB abnormality scores using the AI software (p = 0.571), nor in any of the steps along the TB care cascade during our ACF campaign. Despite some shortcomings, ultra-portable X-ray systems have significant potential to improve case detection and equitable access to high-quality TB care.

BackgroundTuberculosis has caused significant public health and economic burdens in Vietnam over the years. The Vietnam National Tuberculosis Program is facing considerable challenges in its goal to eliminate tuberculosis by 2030, with the COVID-19 pandemic having negatively impacted routine tuberculosis services at all administrative levels. While the turnaround time of tuberculosis infection may delay disease detection, high transportation frequency could potentially mislead epidemiological studies. This study was conducted to develop an online geospatial platform to support healthcare workers in performing data visualization and promoting the active case surveillance in community as well as predicting the TB incidence in space and time.MethodThis geospatial platform was developed using tuberculosis notification data managed by The Vietnam National Tuberculosis Program. The platform allows case distribution to be visualized by administrative level and time. Users can retrieve epidemiological measurements from the platform, which are calculated and visualized both temporally and spatially. The prediction model was developed to predict the TB incidence in space and time.ResultsAn online geospatial platform was developed, which presented the prediction model providing estimates of case detection. There were 400,370 TB cases with bacterial evidence to be included in the study. We estimated that the prevalence of TB in Vietnam was at 414.67 cases per 100.000 population. Ha Noi, Da Nang, and Ho Chi Minh City were predicted as three likely epidemiological hotspots in the near future.ConclusionOur findings indicate that increased efforts should be undertaken to control tuberculosis transmission in these hotspots.

Osteoporosis contributes significantly to health and economic burdens worldwide. However, the development of osteoporosis-related prediction tools has been limited for lower-middle-income countries, especially Vietnam. This study aims to develop prediction models for the Vietnamese population as well as evaluate the existing tools to forecast the risk of osteoporosis and evaluate the contribution of covariates that previous studies have determined to be risk factors for osteoporosis. The prediction models were developed to predict the risk of osteoporosis using machine learning algorithms. The performance of the included prediction models was evaluated based on two scenarios; in the first one, the original test parameters were directly modeled, and in the second the original test parameters were transformed into binary covariates. The area under the receiver operating characteristic curve, the Brier score, precision, recall and F1-score were calculated to evaluate the models’ performance in both scenarios. The contribution of the covariates was estimated using the Permutation Feature Importance estimation. Four models, namely, Logistic Regression, Support Vector Machine, Random Forest and Neural Network, were developed through two scenarios. During the validation phase, these four models performed competitively against the reference models, with the areas under the curve above 0.81. Age, height and weight contributed the most to the risk of osteoporosis, while the correlation of the other covariates with the outcome was minor. Machine learning algorithms have a proven advantage in predicting the risk of osteoporosis among Vietnamese women over 50 years old. Additional research is required to more deeply evaluate the performance of the models on other high-risk populations.