What type of study is this?

September 10, 2025Open Access

Predicting Tuberculosis Treatment Outcome Using Machine Learning Techniques

ADA DeborahCollege of Saint Elizabeth ADAdeniji Oluwashola DavidUniversity of Ibadan AOAdeyemi Samuel OladeleUniversity College Hospital, Ibadan

Key Points

Logistic regression shows a treatment outcome accuracy of 75%, indicating effective prediction.
Random forest also predicts outcomes but with a lower accuracy of 55%, suggesting a performance gap.
Machine learning techniques are used to enhance treatment completion and avoid antibiotic resistance risks.
GridSearchCV improves model performance, signaling the importance of hyperparameter tuning in machine learning.

Abstract

The cause of tuberculosis can be dangerous and even be a fatal disorder, the mainstream of patients are able to recover with prompt diagnosis and treatment. After a few weeks of treatment, you won't be contagious, and you could start feeling better and as a result most people don’t take their TB medications as prescribed by their doctor. Also taking TB medications or not completing the entire therapy could lead to the bacteria still alive in them to develop antibiotics resistance, which is far more dangerous and difficult to treat. In this research, two (2) machine learning algorithms; Logistic Regression (LR) and Random Forest (RF) were employed for predicting tuberculosis treatment outcome in order to ensure treatment completion for favorable outcome. GridSearchCV was used to improve the models performance and of the two developed model, both models performed very well with LR having an accuracy of 75%, and RF an accuracy of 55%.

AI से पूछें

Bookmark

View Full Paper