Diabetes Risk Data Mining Method Based on Electronic Medical Record Analysis.

Area of Science:

• Medical Informatics
• Data Mining
• Public Health

Background:

• Electronic medical records (EMRs) are increasingly vital for clinical treatment and contain comprehensive personal health information.
• Data mining of EMRs is gaining traction in medical research for structured analysis and application in social development.
• Identifying diabetes risk early is crucial for effective public health interventions.

Purpose of the Study:

• To introduce a diabetes risk data mining method based on electronic medical record analysis.
• To provide research directions for diabetes risk prediction using EMR data.
• To evaluate the effectiveness of data mining techniques for identifying individuals at risk of diabetes.

Main Methods:

• Proposed a research strategy for diabetes risk data mining using EMR analysis.
• Employed data mining and classification rule mining techniques on EMR datasets.
• Utilized a decision tree algorithm for predictive modeling and analysis.

Main Results:

• The decision tree model achieved an average prediction accuracy of 91.21%.
• Similar results were observed between the training and test sets, indicating no overfitting.
• The method demonstrated robust performance in identifying diabetes risk from EMR data.

Conclusions:

• The proposed data mining method effectively identifies diabetes risk using electronic medical records.
• The high prediction accuracy suggests the potential for early detection and intervention.
• This approach offers a valuable framework for future research in EMR-based disease risk prediction.