A united model for diagnosing pulmonary tuberculosis with random forest and artificial neural network

Predicting the risk of pulmonary embolism in patients with tuberculosis using machine learning algorithms

BACKGROUND

This study aimed to develop predictive models with robust generalization capabilities for assessing the risk of pulmonary embolism in patients with tuberculosis using machine learning algorithms.

METHODS

Data were collected from two centers and categorized into development and validation cohorts. Using the development cohort, candidate variables were selected via the Recursive Feature Elimination (RFE) method. Five machine learning algorithms, logistic regression (LR), random forest (RF), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), and support vector machine (SVM), were utilized to construct the predictive models. Model performance was evaluated through nested cross-validation and area under the curve (AUC) metrics, supplemented by interpretations using Shapley Additive explanations (SHAP) and line charts of AUC values. Models were subjected to external validation using an independent validation group, facilitating the early identification and management of pulmonary embolism risks in tuberculosis patients.

RESULTS

Data from 694 patients were used for model development, and 236 patients from the validation group met the enrollment criteria. The optimal subset of variables identified included D-dimer, smoking status, dyspnea, age, sex, diabetes, platelet count, cough, fibrinogen, hemoglobin, hemoptysis, hypertension, chronic obstructive pulmonary disease (COPD), and chest pain. The RF model outperformed others, achieving an AUC of 0.839 (95% CI 0.780-0.899) and maintaining the highest average performance in external fivefold cross-validation (AUC: 0.906 ± 0.041).

CONCLUSIONS

The RF model demonstrates high and consistent effectiveness in predicting pulmonary embolism risk in tuberculosis patients.

Identification of diagnostic biomarkers correlate with immune infiltration in extra-pulmonary tuberculosis by integrating bioinformatics and machine learning

The diagnosis of tuberculosis depends on detecting Mycobacterium tuberculosis (Mtb). Unfortunately, recognizing patients with extrapulmonary tuberculosis (EPTB) remains challenging due to the insidious clinical presentation and poor performance of diagnostic tests. To identify biomarkers for EPTB, the GSE83456 dataset was screened for differentially expressed genes (DEGs), followed by a gene enrichment analysis. One hundred and ten DEGs were obtained, mainly enriched in inflammation and immune -related pathways. Weighted gene co-expression network analysis (WGCNA) was used to identify 10 co-expression modules. The turquoise module, correlating the most highly with EPTB, contained 96 DEGs. Further screening with the least absolute shrinkage and selection operator (LASSO) and support vector machine recursive feature elimination (SVM-RFE) narrowed down the 96 DEGs to five central genes. All five key genes were validated in the GSE144127 dataset. CARD17 and GBP5 had high diagnostic capacity, with AUC values were 0.763 (95% CI: 0.717-0.805) and 0.833 (95% CI: 0.793-0.869) respectively. Using single sample gene enrichment analysis (ssGSEA), we evaluated the infiltration of 28 immune cells in EPTB and explored their relationships with key genes. The results showed 17 immune cell subtypes with significant infiltrations in EPTB. CARD17, GBP5, HOOK1, LOC730167, and HIST1H4C were significantly associated with 16, 14, 12, 6, and 4 immune cell subtypes, respectively. The RT-qPCR results confirmed that the expression levels of GBP5 and CARD17 were higher in EPTB compared to control. In conclusion, CARD17 and GBP5 have high diagnostic efficiency for EPTB and are closely related to immune cell infiltration.