[A diagnostic model for alcoholic liver disease based on a generalized regression neural network]

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Hepatology

Background:

• Alcoholic liver disease (ALD) diagnosis relies on expert clinical consultation.
• Developing objective diagnostic tools for ALD is crucial for timely intervention.

Purpose of the Study:

• To evaluate a generalized regression neural network (GRNN) model for diagnosing ALD.
• To assess the integration of multiple disease indicators within the GRNN model for enhanced diagnostic accuracy.

Main Methods:

• A GRNN model was developed using seven selected ALD indicators.
• Data from literature (40 patients) and clinical records (135 patients) were used for training and testing the model.
• Model performance was validated against expert clinical classifications.

Main Results:

• The GRNN model achieved high diagnostic accuracy: 94.1% for reported patients and 100% for clinical patients in the training set.
• The model demonstrated 100% accuracy on the testing sets for both patient groups.
• Results showed strong agreement between the model's diagnoses and expert clinical consultations.

Conclusions:

• The GRNN model, utilizing multiple ALD indicators, provides accurate and comprehensive diagnosis.
• This AI-driven approach shows significant feasibility for clinical application in ALD diagnosis.