Prediction of Tuberculosis: A Logistic Regression Model

Patients with tuberculosis (TB) often visit a physician without showing specific symptoms. Proper isolation, treatment, and preventive measures are essential to stop the transmission of the disease to others. Logistic regression is widely used statistical method for modeling the relationship between two variables (dependent variable and one or more independent variables), particularly when the outcome is categorical or binary. By applying the logistic (sigmoid) function, it transforms linear combinations of input variables into probabilities, making it ideal for classification tasks. Objectives: This study aims to identify the factors influencing tuberculosis (TB) using a logistic regression prediction model. Methods: This study was conducted in two groups consisting of 105 tuberculosis patients and 105 controls. The risk factors associated with tuberculosis (TB) by comparing a group of TB patients with a control group. By analyzing factors like age, sex, smoking habits, alcohol consumption, diabetic status, and asthma history, the goal seems to be identifying patterns or predictors that could help in understanding who might be at higher risk for developing TB. The use of logistic regression to model these influences and evaluate its predictive accuracy (using sensitivity, specificity, and ROC curve analysis) suggests the researchers are looking to quantify the effectiveness of these factors in predicting TB outcomes. Results: The sensitivity (61%) and specificity (64%) values suggest that the model is moderately effective at distinguishing between cases and non-cases of TB. This study employs a logistic regression model appears to be a solid choice for the prediction of tuberculosis, as indicated by the accuracy and predictive power, along with the area under the Receiver Operating Characteristic (ROC) curve of (AUC) of 0.702 indicate that the model is moderately effective at distinguishing between the presence and absence of tuberculosis.