SeekDoc: Seeking eligible doctors from electronic health record

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Data Mining

Background:

• Online medical platforms offer vast resources but complicate doctor selection.
• Electronic Health Records (EHR) analysis is crucial for accurate patient treatment.
• Identifying suitable doctors requires predicting their performance for specific conditions.

Purpose of the Study:

• To develop a novel approach, SeekDoc, for identifying the most effective doctor for a specific disease.
• To leverage patient reviews and doctor-disease interactions for improved medical service discovery.

Main Methods:

• Constructed a doctor-disease heterogeneous information network.
• Integrated patient reviews and rating data.
• Employed autoencoder for unsupervised feature learning and XGBoost for performance prediction.

Main Results:

• SeekDoc effectively predicts doctor experience scores for specific diseases.
• The proposed method demonstrates strong performance compared to existing algorithms.
• The approach efficiently identifies suitable doctors from complex medical information.

Conclusions:

• SeekDoc offers a robust solution for matching patients with appropriate doctors.
• The integration of EHR data, patient feedback, and network analysis enhances medical decision-making.
• This method improves the efficiency and accuracy of doctor selection in online healthcare platforms.