Development and use of a rule-based pathology expert consultation system

Area of Science:

• Pathology
• Medical Informatics
• Artificial Intelligence

Background:

• Histologic interpretation of complex tissues like tumors is subjective, leading to interobserver and intraobserver variability.
• Rule-based expert systems offer precise problem representation and efficient, accessible use for diagnostic support.

Purpose of the Study:

• To develop and discuss a rule-based expert system, the Pathology Expert Consultation System (PECS).
• To enhance diagnostic and prognostic accuracy in pathology, addressing subjectivity in interpreting complex tissue samples.

Main Methods:

• Development of a rule-based expert system (PECS) for pathology consultations.
• Integration of diverse data sources (database, worksheet, text files) with rule bases.
• Implementation of interactive features including graphical displays and explanations for user queries.

Main Results:

• PECS supports diagnosis, inference of stage, type, and grade, and checks for inconsistencies.
• The system can predict patient prognosis and integrates various data formats.
• Initial applications in lung, breast, and endometrial tumors demonstrate educational and clinical value.

Conclusions:

• The Pathology Expert Consultation System (PECS) shows significant potential to improve diagnostic consistency and provide prognostic information.
• PECS offers a valuable tool for both clinical practice and education in pathology.
• The system's ability to integrate data and provide explanations enhances its utility in complex diagnostic scenarios.