Related Experiment Video
Updated: Jan 7, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Prediction of menstrual literacy among female university students using tree-based machine learning algorithms: A
Md Mahadi Hassan1, Noushin Nohor1, Anika Bushra Boitchi1
1Department of Public Health and Informatics, Jahangirnagar University, Dhaka, Bangladesh.
Background:
Menstrual literacy is crucial for safeguarding the health, dignity, and rights of women. However, in low- and middle-income countries (LMICs), such as Bangladesh, it remains an overlooked issue. Female university students, positioned at a pivotal developmental stage, require accurate and comprehensive knowledge about menstruation. Despite overall improvements in education, significant gaps persist in menstrual knowledge and hygiene practices among this group.
Objectives:
This study aimed to predict menstrual literacy levels among female university students in Bangladesh using tree-based machine learning models. It also sought to identify the most influential predictors shaping menstrual knowledge.
Design:
A cross-sectional analytical design was employed to assess menstrual literacy and apply predictive modeling.
Methods:
A total of 576 female students from a public university in Bangladesh participated in a structured survey. Data were collected on sociodemographic and menstrual characteristics. Five tree-based machine learning models-Random Forest, Decision Tree, Extra Trees, LightGBM, and XGBoost-were trained to classify participants based on their level of menstrual knowledge. Model performance was evaluated using accuracy, F1-score, and receiver operating characteristic (ROC)- Area Under the Curve metrics.
Results:
Overall, 76% of participants demonstrated good menstrual knowledge, while 24% had poor knowledge. The Random Forest model outperformed others with an accuracy of 81%, followed closely by Extra Trees, LightGBM, and XGBoost (each at 80%). Key predictors identified across models included menstrual duration, type of hygiene materials used, permanent residence, and whether participants received early menstrual education. All models performed better in predicting participants with good knowledge, revealing a class imbalance.
Conclusion:
Tree-based machine learning approaches effectively predict menstrual literacy and uncover critical influencing factors. These findings can guide targeted educational interventions and public health strategies to improve menstrual literacy, particularly in LMIC settings such as Bangladesh.
