A Data Driven and Interpretable Hybrid Quantum Classical Model for Male Infertility Risk Assessment

Abstract

Worldwide, nearly 15% of couples experience infertility and Male Infertility (MI) accounts for approx. 30-50% of these cases. Although the health impacts and multifactorial risk factors associated with MI remain insufficiently explored. Comprehensive integration of lifestyle, environmental and physiological factors is often lacking, and reliance on limited physical examinations or isolated medical parameters makes existing approaches inadequate. Thus, traditional approaches often miss subtle yet influential patterns and are not suitable for early detection. So, overcoming infertility remains a challenging task, despite the availability of massive Electronic Health Record (EHR) datasets from various sources. Thus, rapid development of Artificial Intelligence (AI) and Quantum Computing (QC) provides an opportunity to develop more accurate and integrative diagnostic systems for early identification, diagnosis and treatment for Assisted Reproductive Technologies (ART). Subsequently, this research proposes a data-driven model designed to improve transparency through predictive modelling and multivariate visualization to uncover hidden relationships among diverse factors such as age, trauma, alcohol consumption, fever history, seasonal variation, and surgical interventions and male fertility outcomes. For this, developed a reliable predictive model to identify key determinants of MI using EHR derived multivariate data and to assess the performance of quantum enhanced Deep Learning (DL) technique over classical LSTM architectures to capture complex nonlinear feature interactions associated with fertility outcomes and evaluate the feasibility of quantum enhanced sequential learning for small healthcare datasets more efficiently. Furthermore, both dataset level visualizations and model-level explainability were provided by SHAP based feature attribution.