Deep Learning Classification of Active Tuberculosis Using Chest X-Rays: Efficacy of Transfer Learning and Generalization Performance of Cross-Population Datasets

Chest X-ray based diagnosis of active Tuberculosis (TB) is one of the oldest ubiquitous tests in medical practice. Artificial Intelligence (AI) based automated detection of abnormality in chest radiography is crucial in radiology workflow. Most deep convolutional neural networks (DCNN) for diagnosing TB by transfer learning from natural images and using the same dataset to evaluate the model performance and diagnostic accuracy. However, dataset shift is a known issue in predictive models in AI, which is unexplored. In this work, we fine-tuned, validated, and tested two benchmark architectures and utilized the transfer learning methodology to measure the diagnostic accuracy on cross-population datasets. We achieved remarkable calcification accuracy of 100% and area under the receiver operating characteristic (AUC) 1.000 [1.000 – 1.000] (with a sensitivity 0.985 [0.971 – 1.000] and a specificity of 0.986 [0.971 – 1.000]) on intramural test set, but significant drop in extramural test set. Accuracy on various extramural test sets varies 50% - 70%, AUC ranges 0.527 – 0.865 (sensitivity and specificity fluctuate 0.394 – 0.995 and 0.443 – 0.864 respectively). Diagnostic performance on the intramural test set observed in this study shows that DCNN can accurately classify active TB and normal chest radiographs, however the external test set shows DCNN is less likely to generalize well on models trained on specific population dataset.

Non-Volatile Organic Compounds in Exhaled Breath Particles Correspond to Active Tuberculosis

Human breath contains trace amounts of non-volatile organic compounds (NOCs) which might inform non-invasive methods for evaluation of individual health. In previous work, we demonstrated that lipids detected in exhaled breath aerosol (EBA) could be used as markers of active tuberculosis (TB). Here, we advanced our analytical platform in characterizing small metabolites and lipids in EBA samples collected from participants enrolled in clinical trials designed to identify molecular signatures of active TB. EBA samples from 26 participants with active TB and 73 healthy participants were processed using a dual-phase extraction method, and metabolites and lipids were identified via mass spectrometry (MS) database matching. In total, 13 metabolite and 9 lipid markers were identified with optimized relative standard deviation values that were statistically different between individuals diagnosed with active TB and the healthy controls. A feature ranking algorithm reduced this number to 10 molecules, with the membrane glycerophospholipid, phosphatidylinositol 24:4, emerging as top driver of segregation between the two groups. These results support the utility of this approach to identify consistent NOC signatures from EBA samples in active TB cases and suggest the potential to apply this method to other human diseases which alter respiratory NOC release.

Epidemiological trends in tuberculosis in Ukraine at the beginning of the COVID­-19 pandemic

Objective — to examine the tendencies of the epidemiological situation on tuberculosis in Ukraine at the beginning of the COVID-19 pandemic in 2020. Materials and methods. The analysis of the data of official statistics in 2015—2020, a survey of 157 medical experts and a telephone questionnaire survey of 69 residents of the Volyn region. Results and discussion. It was established that the disease of COVID-19 has found its imprint on the epidemiological situation of tuberculosis and on the functioning of the TB care service. In 2020, the rate of decline in the incidence of pulmonary and extrapulmonary tuberculosis was significantly more intensive than in previous years. However, the incidence of tuberculosis among contacts with bacterial excretion in tuberculosis foci increased 1.85 times in each year. The frequency of tuberculosis recurrences is steadily decreasing.The prevalence and mortality from all forms of active tuberculosis among the entire population of Ukraine has a tendency similar to a decrease in morbidity.The effectiveness of treatment of patients of all categories is proportional to the severity of the tuberculosis process, but still insufficient. Mortality in TB care institutions increased in the previous year against the decrease of this indicator in previous years. Against the background of the inaccessibility of the population to TB care, day hospitals are underused, and the sanatorium network needs to be reorganized. The surgical method of treatment of patients is extremely insufficiently realized. Conclusions. The dispensary contingent of TB care institutions is decreasing every year. There is a reduction in the infrastructure of the TB care service, and in each year, the largest reduction was in the bed stock of TB care institutions and sanatoriums, as well as a decrease in the number of TB doctors.

A Systematic Review of Economic Evaluations of Active Tuberculosis Treatments

Background: The disease burden of active tuberculosis (TB) is considerable, but systematic reviews of economic evaluations of active TB treatments are scarce.Methods: PubMed, EMBASE, and the Cochrane Library databases were used to search for articles on cost-effectiveness analysis or cost-utility analysis that economically evaluated active TB treatments, which were then systematically reviewed by two independent reviewers. We extracted vital components of the included studies, such as country, population, intervention/comparator, primary outcome, values of outcomes, thresholds, model type, time horizon, and health states included in the model.Results: Seventeen studies were included in this systematic review. Thirteen dealt with interventions of medications, and the remaining four compared care strategies. The Markov model was the most commonly used tool to compare medications, whereas studies on care plans mainly used decision trees. The most commonly used primary outcome was disability-adjusted life years, followed by quality-adjusted life years. For treatment-naïve TB, the 4-month regimen was more cost-effective than the 6-month regimen mainly in low- and middle-income countries. For multidrug-resistant TB, a bedaquiline-based regimen was cost-effective. For multidrug-resistant TB, decentralized care that employed the use of home or mobile devices was more cost-effective than hospital-based centralized care in low- and middle-income countries.Conclusion: New treatment strategies to improve therapeutic outcomes by enhancing treatment adherence, such as regimens with shorter durations (2 or 4 months) and decentralized care, or new anti-TB agents (e.g., bedaquiline) have been suggested as cost-effective interventions for active TB. This review provides information on the economic evaluation of active TB from good-quality studies, thus aiding the future economic evaluation of active TB.