Intelligent diagnosis and prediction of pregnancy induced hypertension in obstetrics and gynecology teaching by integrating GA

TL;DR

This paper presents an intelligent system using genetic algorithms to improve the diagnosis and prediction of pregnancy-induced hypertension, helping medical professionals make better decisions.

Contribution

A novel prediction model integrating genetic algorithms and optimized feature selection for diagnosing pregnancy-induced hypertension is introduced.

Findings

The model achieved high performance metrics, including a recall of 0.768 and an area under the curve of 0.832.

The ratio of vascular endothelial growth factor receptor 1 to placental growth factor showed peak area under the curve values of 0.996 and 0.792 for early and late assessments.

Abstract

Advanced Diagnosis and Forecasting of Pregnancy-Induced Hypertension in Obstetrics and Gynecology Education through the Integration of Genetic Algorithms.

Pregnancy-induced hypertension represents a critical issue within the fields of obstetrics and gynecology, where precise diagnosis and forecasting are essential for effective management. The potential for misdiagnosis, often stemming from the inexperience of healthcare professionals, underscores the necessity for an advanced diagnostic system.

This research introduces an innovative sampling and feature selection technique grounded in F-scores optimization, alongside the development of a comprehensive prediction model that integrates genetic algorithms with various heterogeneous learners. The objective of this model is to maximize the utility of medical data and enhance treatment quality.

The refined intelligent feature selection approach identified several significant indicators of pregnancy-related hypertension, such as phosphor dehydrogenase deficiency, body mass index, gestational urinary proteins, vascular endothelial growth factor receptor 1, placental growth factor, thalassemia, and a familial history of diabetes mellitus or hypertension. The model achieved superior performance metrics, including the highest recall (0.768), F-score (0.728), and area under the curve (0.832) when compared to other prevalent models. Furthermore, the area under the curve for both early and late clinical assessments reached peak values of 0.996 and 0.792, respectively, when evaluated using the ratio of vascular endothelial growth factor receptor 1 to placental growth factor.

The intelligent diagnosis and prediction methodology for gestational hypertension proposed in this study exhibited remarkable efficacy and holds significant promise for implementation in both educational and clinical settings within obstetrics and gynecology, thereby advancing intelligent medical diagnostics in China.

Figures