Scholarlink

In diabetes, blood sugar is abnormally high. A chronic ailment observed in old and young people nowadays. Type 1 diabetes, Type 2 diabetes, prediabetes, gestational diabetes, Type 3c diabetes, neonatal diabetes, etc. No matter the type of diabetes, high blood glucose is the major cause. Diabetes symptoms include insulin resistance, autoimmune illness, pancreatic damage, and genetic mutations. Thus, early detection of this condition is crucial to provide diabetic patients with adequate medication and medical care. Many Machine Learning (ML)-based diabetes detection methods are used. Deep learning models for early diabetes prediction will be our emphasis in this effort. Deep learning models are trained and tested using the Diabetes likelihood Prediction dataset. The dataset has been pre-processed to eliminate noise and prevent overfitting. Three models—LSTM, CNN, and CNN-Deep LSTM—have been implemented. Hybrid CNN-Deep LSTM outperforms common ML models in predicting diabetes risk given fresh patient data. This study might help diagnose diabetes mellitus early so patients may get correct treatment.

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

Efficient Deep Learning-Based Approaches for Early Diagnosis of Diabetes Mellitus

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • G. Priyanka,
  • M. Nivaashini,
  • M. Thenmozhi,
  • G. Sathish Kumar

摘要

In diabetes, blood sugar is abnormally high. A chronic ailment observed in old and young people nowadays. Type 1 diabetes, Type 2 diabetes, prediabetes, gestational diabetes, Type 3c diabetes, neonatal diabetes, etc. No matter the type of diabetes, high blood glucose is the major cause. Diabetes symptoms include insulin resistance, autoimmune illness, pancreatic damage, and genetic mutations. Thus, early detection of this condition is crucial to provide diabetic patients with adequate medication and medical care. Many Machine Learning (ML)-based diabetes detection methods are used. Deep learning models for early diabetes prediction will be our emphasis in this effort. Deep learning models are trained and tested using the Diabetes likelihood Prediction dataset. The dataset has been pre-processed to eliminate noise and prevent overfitting. Three models—LSTM, CNN, and CNN-Deep LSTM—have been implemented. Hybrid CNN-Deep LSTM outperforms common ML models in predicting diabetes risk given fresh patient data. This study might help diagnose diabetes mellitus early so patients may get correct treatment.