Scholarlink

We have developed our previous work by focusing on the types of fetal malformation cases collected from the Children’s Hospital of the Ministry of Health in the Syrian Arab Republic during the period 2020–2023. These cases were classified into four categories: the first is moderate to severe malformations, which represent about 4% of cases. The second type is multiple and overlapping malformations, accounting for approximately 14% of the targeted cases. The third type is complex and life-threatening malformations, representing about 46% of the total cases. The last type is benign cases with no malformation, making up around 36% of the total cases. The network input consisted of ten features: father’s age, mother’s age, smoking during pregnancy, medication use during pregnancy, diseases and infections during pregnancy, genetic factors, family history, consanguinity between the couple, exposure to radiation during pregnancy, and previous miscarriages or infant deaths. The output was the classification of the fetal malformation type, the associated diseases, and their severity, ultimately aiding in the decision-making process regarding whether to continue the pregnancy. We built six predictive models to classify fetal malformations: Random Forest, SVM, KNN, Logistic Regression, Decision Tree, and Naïve Bayes. The accuracy of these models was as follows: 73.3%, 70.5%, 69.2%, 69.2%, 66.4%, and 61.0%, with an overall accuracy of 68.2667%. Our designed model is effective and helpful for parents in making informed decisions regarding the birth of a healthy and safe baby.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Using Machine Learning Techniques to Classify Fetal Congenital Malformations Early

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Majida Albakoor,
  • Khalid Saeed,
  • Akram Massouh,
  • Batoul Aljneidi,
  • Yanal Alshoufi,
  • Qais Ayman Abohadir,
  • Farah Ziad Zaineddin

摘要

We have developed our previous work by focusing on the types of fetal malformation cases collected from the Children’s Hospital of the Ministry of Health in the Syrian Arab Republic during the period 2020–2023. These cases were classified into four categories: the first is moderate to severe malformations, which represent about 4% of cases. The second type is multiple and overlapping malformations, accounting for approximately 14% of the targeted cases. The third type is complex and life-threatening malformations, representing about 46% of the total cases. The last type is benign cases with no malformation, making up around 36% of the total cases. The network input consisted of ten features: father’s age, mother’s age, smoking during pregnancy, medication use during pregnancy, diseases and infections during pregnancy, genetic factors, family history, consanguinity between the couple, exposure to radiation during pregnancy, and previous miscarriages or infant deaths. The output was the classification of the fetal malformation type, the associated diseases, and their severity, ultimately aiding in the decision-making process regarding whether to continue the pregnancy. We built six predictive models to classify fetal malformations: Random Forest, SVM, KNN, Logistic Regression, Decision Tree, and Naïve Bayes. The accuracy of these models was as follows: 73.3%, 70.5%, 69.2%, 69.2%, 66.4%, and 61.0%, with an overall accuracy of 68.2667%. Our designed model is effective and helpful for parents in making informed decisions regarding the birth of a healthy and safe baby.