REFEROCOD: a probabilistic method to medical coding support.

Area of Science:

• Medical Informatics
• Health Services Research

Background:

• Accurate medical coding is crucial for healthcare evaluation and funding.
• French physicians face challenges with non-intuitive diagnosis coding, leading to frequent errors.
• Coding errors have significant consequences for healthcare systems.

Purpose of the Study:

• To develop and evaluate REFEROCOD, a tool designed to support medical coding.
• To minimize diagnosis coding errors and improve efficiency for physicians.
• To provide a system that suggests appropriate diagnosis codes based on clinical context.

Main Methods:

• The REFEROCOD tool utilizes probabilistic knowledge to suggest diagnosis codes.
• It considers physician activities, patient medical context, procedures, age, and sex.
• Information is extracted from discharge abstracts to inform code suggestions.

Main Results:

• The proposed method aims to minimize errors in diagnosis code assignment.
• REFEROCOD provides a list of probable diagnosis codes.
• The system enhances accuracy by considering multiple patient and procedural factors.

Conclusions:

• REFEROCOD offers a promising solution to reduce medical coding errors.
• The tool supports physicians in assigning accurate diagnosis codes efficiently.
• Probabilistic knowledge integration improves the reliability of coding suggestions.