Validation of a metabolic prototype to assist in the treatment of insulin-dependent diabetes mellitus

Abstract

This paper describes the principles and prototyping of a computer system to assist in the treatment of patients with insulin-dependent (type 1) diabetes mellitus. The system adopts a mixed approach involving rule-based qualitative algebra and a dynamic mathematical model to define the relationships between insulin dosage, diet and glycaemic response. The rule-based system (KBS), implemented in PROLOG, can be used to generate qualitative therapeutic advice. These suggestions are quantified and rank-ordered by the use of a mathematical model of glucose—insulin interaction in type 1 diabetes mellitus, with parameters adjusted for individual patients. In this paper an overview of the integrated prototype, linking the KBS and model, is provided and a case study used to demonstrate the principles of the system in operation. The results of verification and validation work performed on the KBS are described.

Key Words:

• Knowledge-based system
• Insulin advisory therapy
• Diabetes mellitus
• Dynamic simulation
• Verification
• Validation