STI : "Système Tutoriel Intelligent" in french

We worked on Analyze and Design an intelligent tutorial system for learning medical diagnosis

Abstract

In 2014, the WHO estimated that 8,000 people die every year in Cameroon as a result of medical errors, errors that are further accentuated with the lack of experience of our young doctors fresh out of school, who seek experience, but often acquire it by doing often irreversible damage in the process. Among the experience to be acquired by these young physicians is diagnostic work, which is a rather complex exercise that combines both knowledge and know-how. This document proposes a model of an ITS that could be a framework for experimentation in diagnosis for medical students and young physicians. The stage has been set by presenting the general concepts and a review of the literature about ITS and medical diagnosis. Then, an overall architecture of an ITS was described. There will be a learner model that will allow us to describe and follow the evolution of the learner's learning, his level, his gaps, his misconceptions; a tutor model that will allow us to describe our pedagogical strategy : training; an expert model that describes a representation of the knowledge to be transmitted to the learner including the identification of the disease ( in the form of a "if such symptoms, then such disease" relationship) and the diagnosis procedure ( using a base of facts and rules and Bayesian networks for inference); a virtual patient, modeled as an emotional and reactive agent, to simulate a disease and answer questions posed by the learner; and, finally the user interface through which the learner will interact with our ITS. The design of the user interface followed the user-centered design methodology presented in the ISO 13407 standard. There was an analysis allowing the identification and characterization of the target population and its environment; then, a design allowing the specification of the user requirements. Then, the tools, algorithms and interfaces that will be used for the implementation of each component of the ITS were presented. In this regard, in the Tutor module, JESS will be used to implement the rule base. The case base of the Expert module will be stored in a NoSQL database, MongoDB in this case. The Bayesian network will be implemented with the pyArgum tool. The user interface has been prototyped on the High Fidelity approach with the Figma tool.

Checkout our .