Smart Maternal Health Analyzer by AI

Abstract

Maternal health difficulties are currently one of the most difficult challenges in the world. Every year, many women die during pregnancy and after childbirth, which is a primary source of infant mortality. Maternal risk factors such as the mother\'s chronic illness, blood pressure, mental health, diet, and other medical care during pregnancy all play important roles. Pregnant women in remote locations confront several obstacles and challenges, including a scarcity of doctors, insufficient expertise, a lack of accessible clinics, infrastructural constraints, and transportation issues. The infant\'s poor health is mostly due to the mother\'s pregnancy, rather than any additional issues that may have occurred following childbirth. Using machine learning approaches, the study has predicted the maternal health risk level in previous due to avoid uncertain birth death or any inconvenience of a new born child. A variety of pre-trained advanced machine learning techniques were utilized in the study to find out the sustainable result. ANN, Ridge Classifier, SGD, XGBoost, Cat Boost, Random Forest, XGB, Decision Tree, and more algorithms were implemented. The recommended model was created, trained, and tested on the preprocessed dataset with the help of Hyper Parameter Tuning. The Cat Boost Classifier was the most accurate machine learning system for the study with a score of 97.4%.

Introduction

The text focuses on the use of machine learning for predicting maternal health risks, aiming to reduce maternal mortality and improve early detection of pregnancy-related complications, particularly in Bangladesh and other developing regions.

Problem Background

Maternal mortality remains a serious global health issue, with high rates in developing countries. In Bangladesh, maternal deaths and complications are still significant despite improvements in healthcare. Key risk factors include:

• Age
• Blood pressure
• Blood sugar levels
• Heart rate
• Blood disorders
• Environmental and demographic conditions

Limited access to timely healthcare, data, and early diagnosis contributes to preventable deaths and complications.

Objective

The main goal is to build a machine learning-based system that:

• Predicts maternal health risk early
• Classifies patients into risk categories
• Helps doctors make faster, data-driven decisions
• Reduces maternal mortality through early intervention

Literature Review Insights

Previous studies show that:

• ML models (SVC, Decision Trees, Random Forest, XGBoost, CatBoost, ANN) are effective for prediction.
• CatBoost often achieves the highest accuracy in maternal risk classification.
• CNNs and other AI methods improve medical prediction tasks.
  However, limitations include:
• Small or incomplete datasets
• Lack of real-world data integration
• Poor model interpretability
• Data privacy and accessibility issues

Proposed Methodology

The system uses a structured ML pipeline:

Data Collection & Preprocessing

• Data from hospitals, surveys, and electronic health records
• Handling missing values, encoding categorical data, and normalization
• Feature selection to identify key risk factors

Model Training

• Multiple algorithms used:
  • ANN
  • Naive Bayes
  • Decision Tree
  • SVC
  • SGD
  • XGBoost
  • CatBoost
  • Ridge Classifier
• Dataset: 1524 patient records with 12 features
• Stratified train-test split and cross-validation applied
• Hyperparameter tuning used to improve accuracy and prevent overfitting

Evaluation

Models are evaluated using:

• Accuracy
• Precision
• Recall
• F1-score
• AUC-ROC
• Confusion matrix

System Design

The system follows a full ML workflow:

• Data collection from clinical and survey sources
• Preprocessing and feature engineering
• Model training and validation
• Performance comparison of multiple algorithms

Key Findings & Challenges

• Machine learning can effectively predict maternal health risks
• CatBoost and ensemble methods perform strongly
• Major challenges include:
  • Limited access to medical data
  • Privacy concerns
  • Need for better interpretability
  • Requirement for real-world deployment validation

Conclusion

In today\'s world, we are all dependent on technology. We cannot survive a day without the use of technology. However, technology and appliances offer both advantages and downsides. In the world of technology, the implementation of machine learning is rather difficult. The medical and technological sectors are interdependent. Today, we observe a revolutionary technology participation in the medical business. The proposed method predicts maternal health risks using data from patient medical checkups. To predict maternal health risk, I evaluated the effectiveness of several classifiers and boosting strategies. When comparing the accuracy of different machine learning algorithms, CatBoost performs remarkably well across a variety of criteria, whereas Naive Bayes has the lowest accuracy. Python and Google Colab were used to conduct the research. Pregnancy death can result from a significant risk of maternal health. This automated technique might be a useful tool in raising awareness of whether maternal health risks are high or low. This tool can produce the most rapid and accurate prediction results in the shortest amount of time and at no cost.

References

[1] “Global Health Observatory,” Who.int, 2019. https://www.who.int/data/gho.“Bangladesh Bureau of Statistics,” Dghs.gov.bd, 2023. [2] A. Rajkomar, J. Dean, and I. Kohane, “Machine Learning in Medicine,” New England Journal of Medicine, vol. 380, no. 14, pp. 1347-1358, Apr. 2019. [3] A. Raza, H. U. R. Siddiqui, K. Munir, M. Almutairi, F. Rustam, and I. Ashraf, “Ensemble learning-based feature engineering to analyze maternal health during pregnancy and health risk prediction,” PLOS ONE, vol. 17, no. 11, p. e0276525, Nov. 2022. [4] H. B. Mutlu, F. Durmaz, N. Yucel, Cengi lE., and M. Yildirim, ¨ “Prediction of Maternal Health Risk with Traditional Machine Learning Methods,” NATURENGS, vol. 4, no. 1, pp. 16– 23, Jun. 2023. [5] N. Krupa, M. MA, E. Zahedi, S. Ahmed, and F. M. Hassan, “Antepartum fetal rate feature extraction and classification using empirical mode decomposition and support vector machine,” Biomedical Engineering OnLine, vol. 10, no. 1, p. 6, 2011. [6] S. Nanda, M. Savvidou, A. Syngelaki, R. Akolekar, and K. H. Nicolaides, “Prediction of gestational diabetes mellitus by maternal factors and biomarkers at 11 to 13 weeks,” Prenatal Diagnosis, vol. 31, no. 2, pp. 135–141, Dec. 2010. [7] L. Pawar, J. Malhotra, A. Sharma, D. Arora, and D. Vaidya, “A Robust Machine Learning Predictive Model for Maternal Health Risk,” 2022 3rd International Conference on Electronics and Sustainable Communication Systems (ICESC), Aug. 2022. [8] S. Chaudhuri and B. Mandal, “Predictive behaviour of maternal health inputs and child mortality in West Bengal – An analysis based on NFHS3,” Heliyon, vol. 6, no. 5, pp. e03941– e03941, May 2020. [9] M. Assaduzzaman, A. A. Mamun, and M. Z. Hasan, “Early Prediction of Maternal Health Risk Factors Using Machine Learning Techniques,” IEEE Xplore, Jan. 01, 2023. [10] E. F. Elia and J. Ayungo, “Socio-demographic influence on the pregnant women’s comprehension of maternal health information in Tanzania,” Heliyon, vol. 9, no. 12, p. e22448, Dec. 2023. [11] L. C. Kenny, W. B. Dunn, D. I. Ellis, J. Myers, P. N. Baker, and D. B. Kell, “Novel biomarkers for pre-eclampsia detected using metabolomics and machine learning,” Metabolomics, vol. 1, no. 3, pp. 227– 234, Jul. 2005. [12] R. Yarlapati, S. Roy Dey, and S. Saha, “Early Prediction of LBW Cases via Minimum Error Rate Classifier: A Statistical Machine Learning Approach,” 2017 IEEE International Conference on Smart Computing (SMARTCOMP), May 2017.

Copyright

Copyright © 2026 Ch. Lahari , Mrs. V. Swapna , M. Mastan Vali , B. Kavyanjali , K. Ravi Raju. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.