Juma M. Shija
Computational Ecology and Software
International Academy of Ecology and Environmental Sciences (IAEES)
15 (2) , 30
This study presents a machine learning model for the early detection of sexually transmitted infections using symptom-based clinical data. Five supervised learning algorithms were evaluated: AdaBoost, Decision Tree, Random Forest, Stochastic Gradient Descent, and Support Vector Machine. AdaBoost achieved the best performance and was selected for deployment. The model was integrated into a Django-based web application to provide real-time prediction through a secure and user-friendly interface. The study demonstrates how machine learning can support early disease screening, improve healthcare accessibility, and provide intelligent decision support in health-related systems.
To design and develop a machine learning model capable of supporting early detection of sexually transmitted infections based on symptom data.
To compare multiple supervised learning algorithms and identify the best-performing model.
To integrate the trained model into a web-based application for real-time prediction and user interaction.
The study used symptom-based clinical records related to sexually transmitted infections. Data preprocessing included cleaning inconsistent entries, correcting typographical errors, handling missing values, and applying feature engineering. SMOTE was used to address class imbalance. The dataset was split into training, testing, and validation sets using a 70:15:15 ratio. Five algorithms were evaluated: AdaBoost, Decision Tree, Random Forest, Stochastic Gradient Descent, and Support Vector Machine. The best model was serialized using joblib and integrated into a Django web application.
AdaBoost achieved the highest performance among the evaluated models, with an accuracy of 97.45%, precision of 97.53%, and F1-score of 97.7%. The results show that ensemble learning can perform well on structured healthcare prediction tasks and can support real-time predictive analytics when integrated into a web application.
The study demonstrates that machine learning can be used effectively to support early screening of sexually transmitted infections. The integration of the model into a Django-based web application shows the practical value of combining predictive analytics with full-stack web systems for accessible, real-time decision support.
Future work should focus on expanding the dataset, incorporating additional predictors, optimizing models, validating the system in clinical settings, and deploying the system across multiple platforms including USSD for users without smartphones.
Institution:
Mbeya University of Science and Technology; Nelson Mandela African Institution of Science and Technology
Supervisors:
Dr. Elizabeth Mkoba and Dr. Judith Leo