ItemArticle

Deep and Hybrid Ensemble Learning Methods for Enhanced Live-Birth Prediction in Fertility Treatments

Title
Deep and Hybrid Ensemble Learning Methods for Enhanced Live-Birth Prediction in Fertility Treatments
Creator
Jain, Rituraj; Shankar, Uma; Radhakrishnan, G V; Date, Saroj; Upreti, Kamal; Caroline, S.; Babu, P Ramesh; Dayal, Mohit
Description
The prediction of live birth outcomes using Assisted Reproductive Technologies (ART) remains a complex task owing to the high inter-patient variability and non-linear clinical interactions. This study presents a comparative evaluation of hybrid machine-learning models to improve in vitro fertilization (IVF) success prediction using a real-world anonymized dataset of 2,000 ART cases. After pre-processing (including missing value imputation, feature selection via Recursive Feature Elimination with Cross-Validation, and class balancing using SMOTE with k=5), four hybrid models were developed: stacking with XGBoost as the meta-learner, weighted ensemble, autoencoder-based feature fusion, and cascading classifiers. Models were evaluated using accuracy, AUC, precision, recall, and F1-score metrics, and compared against a baseline Random Forest classifier. The stacking model (XGBoost with Random Forest, MLP, and SVM base learners) achieved the best performance, with an accuracy and 0.999 AUC of 0.985. The weighted hybrid ensemble followed an accuracy of 0.953 and AUC of 0.994. The statistical significance of the improvements was confirmed using Wilcoxon Signed-Rank and McNemars tests (p < 0.05). To enhance model transparency, SHapley Additive exPlanations (SHAP) was applied to interpret base model contributions in the stacking architecture. These results support the application of AI-driven hybrid modelling for personalized IVF treatment planning. Future work will focus on prospective validation and clinical decision support system (CDSS) integration to assess deployment feasibility. (2025), (Slovene Society Informatika). All rights reserved.
Source
Informatica (Slovenia);Volume;49;Issue;7;pp.407-422
Date
01-01-2025
Publisher
Slovene Society Informatika
Subject
assisted reproductive technologies; hybrid machine learning; IVF prediction; live-birth estimation; predictive modelling
Coverage
Jain R., Department of Information Technology, Marwadi University, Gujarat, Rajkot, India; Shankar U., Faculty of Management and Social Sciences, Qaiwan International University, Kurdistan, Sulaimanyah, Iraq; Radhakrishnan G.V., Kalinga School of Management, Kalinga Institute of Industrial Technology, Bhubaneswar, India; Date S., Department of Artificial Intelligence and Data Science, CSMSS Chh. Shahu college of Engineering, Chh. Sambhajinagar, Maharashtra, India; Upreti K., Department of Computer Science, Christ University, Delhi NCR Campus, Ghaziabad, India; Caroline S., Department of Electronics and Communication Engineering, St. Xavier's Catholic College of Engineering, Tamil Nadu, Nagercoil, India; Babu P.R., Department of Computer Science & Engineering, Narsimha Reddy Engineering College, Telangana, Secunderabad, India; Dayal M., Department of Information Technology, Bharati Vidyapeeth's College of Engineering, New Delhi, India
Rights
Restricted Access; Hardcopy may be available in the library
Relation
ISSN: 3505596; CODEN: INFOF
Format
online
Language
English
Type
Article

Citation

Jain, Rituraj; Shankar, Uma; Radhakrishnan, G V; Date, Saroj; Upreti, Kamal; Caroline, S.; Babu, P Ramesh; Dayal, Mohit, “Deep and Hybrid Ensemble Learning Methods for Enhanced Live-Birth Prediction in Fertility Treatments,” CHRIST (Deemed To Be University) Institutional Repository, accessed June 18, 2026, https://archives.christuniversity.in/items/show/23482.

Output Formats