Knowledge-Based System for Diagnosis of Metabolic Alterations in Undergraduate Students

A knowledge based system to identify 10 main metabolic alterations in university students based on clinical and anthropometric parameters is presented. Knowledge engineering was carried out through unstructured expert interviews methodology, resulting in a knowledge base of 17 IF-THEN rules. A backward chaining machine engine was built in Prolog language; the attribute-values database about parameters of each student was also stored in Prolog facts. The system was applied to 592 cases: clinical and anthropometric parameters of the students stored in the database. Medical diagnoses and recommendations for each student, obtained from the system, were organized in individualized reports that the physicians gave to the students in personal interviews along only two days. The effectiveness of these interviews is largely attributed to the fact that physicians are the same experts who participated in the process of building the knowledge base.