A deep learning-based algorithm for pulmonary tuberculosis detection in chest radiography.

In tuberculosis (TB), chest radiography (CXR) patterns are highly variable, mimicking pneumonia and many other diseases. This study aims to evaluate the efficacy of Google teachable machine, a deep neural network-based image classification tool, to develop algorithm for predicting TB probability of CXRs. The training dataset included 348 TB CXRs and 3806 normal CXRs for training TB detection. We also collected 1150 abnormal CXRs and 627 normal CXRs for training abnormality detection. For external validation, we collected 250 CXRs from our hospital. We also compared the accuracy of the algorithm to five pulmonologists and radiological reports. In external validation, the AI algorithm showed areas under the curve (AUC) of 0.951 and 0.975 in validation dataset 1 and 2. The accuracy of the pulmonologists on validation dataset 2 showed AUC range of 0.936-0.995. When abnormal CXRs other than TB were added, AUC decreased in both human readers (0.843-0.888) and AI algorithm (0.828). When combine human readers with AI algorithm, the AUC further increased to 0.862-0.885. The TB CXR AI algorithm developed by using Google teachable machine in this study is effective, with the accuracy close to experienced clinical physicians, and may be helpful for detecting tuberculosis by CXR.

Impact of medications on outcomes in patients with acute myocardial infarction and chronic obstructive pulmonary disease: A nationwide cohort study.

BACKGROUND: Various inhaled bronchodilators have been associated with cardiovascular safety concerns. This study aimed to investigate the long-term impact of chronic obstructive pulmonary disease (COPD) and the safety of COPD medications in patients after their first acute myocardial infarction (AMI). METHODS: This nationwide cohort study was conducted using data from the Taiwan National Health Insurance Research Database. Patients hospitalized between 2000 and 2012 with a primary diagnosis of first AMI were included and divided into three cohorts (AMI, ST-elevation myocardial infarction [STEMI], and non-STEMI [NSTEMI]). Each cohort was propensity score matched (1:1) with patients without COPD. A Cox proportional hazards regression model was used to estimate hazard ratios (HRs) with 95% CIs. RESULTS: A total of 186 112 patients with AMI were enrolled, and COPD was diagnosed in 13 065 (7%) patients. Kaplan-Meier curves showed that patients with COPD had a higher mortality risk than those without COPD in all cohorts (AMI, STEMI, and NSTEMI). The HR of mortality in AMI, STEMI, and NSTEMI patients with COPD was 1.12 (95% CI, 1.09-1.14), 1.20 (95% CI, 1.14-1.25), and 1.07 (95% CI, 1.04-1.10), respectively. Short-acting inhaled bronchodilators and corticosteroids increased mortality risk in all three cohorts. However, long-acting inhaled bronchodilators reduced mortality risk in patients with AMI (long-acting beta-agonist [LABA]: HR, 0.87; 95% CI, 0.81-0.94; long-acting muscarinic antagonist [LAMA]: HR, 0.82; 95% CI, 0.69-0.96) and NSTEMI (LABA: HR, 0.89; 95% CI, 0.83-0.97; LAMA: HR, 0.80; 95% CI, 0.68-0.96). CONCLUSION: This study demonstrated that AMI patients with COPD had higher mortality rates than those without COPD. Using inhaled short-acting bronchodilators and corticosteroids reduced survival, whereas long-acting bronchodilators provided survival benefits in AMI and NSTEMI patients. Therefore, appropriate COPD medication for acute AMI is crucial.