Diagnosis Of Large For Gestational Age Fetus With An Expert-Driven Feature Selection Scheme

Faheem Akhtar,Jianqiang Li,Yan Pei,Bo Liu,Muhammad Azeem,Qing Wang,Ji-Jiang Yang
DOI: https://doi.org/10.1109/SSCI44817.2019.9002960
2019-01-01
Abstract:We propose to use an expert-driven feature selection scheme to diagnose and predict Large for Gestational Age (LGA) fetuses. A Fetus with excessive birth weight exhibits adverse neonatal and maternal complications. Early intervention can slow progression and prevent the upcoming complication of the disease. In this research, four well-known machine learning classifiers with ten-folds cross-validations are used to authenticate the proposed scheme. A Master feature vector is created, and an expert-driven feature selection scheme is proposed, which is later compared with existing published researches, master feature file created, and with an automated feature selection scheme. The best performance metrics (precision and AUC) scores are produced by random forest and logistic regression classifiers with the proposed expert-driven feature selection scheme. The proposed scheme played an essential role in elevating prediction precision and AUC scores from 0.71 and 0.70 to (0.9461 and 0.8172) and (0.9174 and 0.8281), respectively. Therefore, we recommend obstetrician's to update the prognosis process for LGA identification using expert-driven feature selection scheme.
What problem does this paper attempt to address?