Predictive Model Comparison for Predicting Condom Use - Comparison of Conventional Logistic Regression and Other Machine Learning

Aug 4, 2026

Upload by Skyler

Condom use at first sex remains a critical determinant of later sexual behavior. This study predicts condom use among male adolescents using five machine-learning classification models and compares them with conventional logistic regression. The dataset is drawn from the Indonesian Demographic and Health Survey (IDHS) 2017. Model performance is evaluated via accuracy, specificity, sensitivity, ROC curve, and AUC. Decision tree and naïve Bayes achieve the highest specificity and sensitivity, while KNN yields the highest AUC. Logistic regression further links condom use to education level, age at first sex, and attitudes toward condoms, leading to recommendations on equal education opportunities and improved knowledge and attitudes.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Background on condom use and HIV/STIs

Research gap and study objective

Materials and Method

Study population and dataset

Variables and features

Results and Discussion

Model performance and evaluation metrics

Interpretation of influential factors and recommendations

Predictive Model Comparison for Predicting Condom Use: Comparison of Conventional Logistic Regression and Other Machine Learning Fadhaa Aditya Kautsar Murti Department of Public Health, Universitas Indonesia, Indonesia E-Mail: [fadhaa.aditya@ui.ac.id](fadhaa.aditya@ui.ac.id) Received Jun 25th 2024; Revised Jul 15th 2024; Accepted Jul 22th 2024 Corresponding Author: Fadhaa Aditya Kautsar Murti Abstract Condom use at first sex remains an important issue as it shapes future sexual behavior. This study aimed to deploy and predict condom use using five different machine learning classification models. Dataset used for this study was from Indonesian Demographic and Health Survey (IDHS) 2017 with a population of interest was male adolescents. We evaluated five different models, namely logistic regression, naïve bayes, K-Nearest Neighbors, support vector machines, and decision tree. Performances of each model were assessed using metrics such as accuracy, specificity, sensitivity, ROC Curve, and AUC Score. Study found that different models exhibit different accuracy, specificity, sensitivity, ROC Curve, and AUC Score. The decision tree and naïve bayes models remained the models with the highest specificity and sensitivity, however the KNN model expressed the highest AUC score. Result from the conventional logistic regression also explained that condom use was associated with education level, age at first sex, and attitude towards condom use. The government is advised to create equal education opportunities for every adolescent and shape better knowledge and condom attitudes. Future studies are advised to enhance the performance of machine learning models using hyperparameter tuning and other methods. Keywords: Condom Use, Demographic Health Survey, First Sex, Indonesian Adolescents, Machine Learning 1. INTRODUCTION The reproductive health domain in Indonesia remains a significant issue. One of the main problems in this domain is the high number of HIV/AIDS cases and other Sexually Transmitted Infections (STIs) . The latest prevalence of HIV showed 540,000 people living with HIV/AIDS (PLWHA) in 2022 [1] . This problem is exacerbated by the age ofPLWHA in Indonesia, with a report from the first quarter of 2023 stating that more than 20% of new HIV cases are among adolescents [2] . It is well known that the prevalence and incidence of such issues are influenced by the consistency of condom use among adolescents. This is evident from the low number of adolescents who use condoms during their first sexual encounter [3]. Data from 2017 showed that less than 30% of adolescents used a condom during their first sexual experience [4] . It is, therefore, important to understand why condom use remains very low, given that condoms can reduce the probability of HIV infection by 90%[5] . Previous studies have explored why adolescents do not use condoms, citing factors such as low education levels, early sexual debut, peer pressure, and the nature of the partner at first sex [6], [7], [8], [9] . However, these studies primarily used conventional logistic regression to investigate the reasons for not using condoms. As statistical modeling and prediction methods have advanced, the application of machine learning in public health has increased. Previous studies have addressed the problem ofunmet needs in HIV prevention using machine learning, particularly among Indian and Chinese populations [10], [11] . However, the application of machine learning in Indonesia is relatively new. To our knowledge, other prediction methods besides conventional logistic regression have not been used to predict condom use in Indonesia. Therefore, this research aims to fill this gap by predicting condom use using machine learning methods, in comparison to conventional logistic regression. The novelty of using alternative prediction models is undoubtedly important, as it provides new perspectives for solving public health problems. This research will

FAQs

What data source and target population are used to predict condom use?

The study uses the Indonesian Demographic and Health Survey (IDHS) 2017 and focuses on unmarried male adolescents who had sex.

Which performance metrics are used to compare the prediction models?

Models are evaluated using accuracy, specificity, sensitivity, ROC curve, and AUC score.

How do the study findings compare decision tree, naïve Bayes, and KNN results?

Decision tree and naïve Bayes show the highest specificity and sensitivity, while the KNN model achieves the highest AUC score.

Aug 4, 2026

Upload by Skyler