[Paper Review] ICD 10 Based Medical Expert System Using Fuzzy Temporal Logic

P. Chinniah, S. Muttan|arXiv (Cornell University)|Jan 12, 2010

Artificial Intelligence and Decision Support Systems2 references3 citations

TL;DR

This paper presents an ICD-10-based medical expert system that leverages fuzzy temporal logic to enhance diagnostic accuracy by modeling symptom severity and disease likelihood over time. It uses a fuzzy weighted symptom-disease relationship, applies Elders' algorithm with a modified Euclidean metric for clustering, and employs minimum similarity as a decision threshold to recommend possible diseases, reducing reliance on physician-led initial diagnosis time and cost.

ABSTRACT

Medical diagnosis process involves many levels and considerable amount of time and money are invariably spent for the first level of diagnosis usually made by the physician for all the patients every time. Hence there is a need for a computer based system which not only asks relevant questions to the patients but also aids the physician by giving a set of possible diseases from the symptoms obtained using logic at inference. In this work, an ICD10 based Medical Expert System that provides advice, information and recommendation to the physician using fuzzy temporal logic. The knowledge base used in this system consists of facts of symptoms and rules on diseases. It also provides fuzzy severity scale and weight factor for symptom and disease and can vary with respect to time. The system generates the possible disease conditions based on modified Euclidean metric using Elders algorithm for effective clustering. The minimum similarity value is used as the decision parameter to identify a disease.

Motivation & Objective

• To reduce time and cost associated with initial physician-led diagnosis by automating symptom-based disease inference.
• To model evolving symptom severity and disease probability over time using fuzzy temporal logic.
• To integrate ICD-10 disease classification with a fuzzy logic-based knowledge base for improved diagnostic reasoning.
• To apply clustering via Elders' algorithm and modified Euclidean metric to group similar symptom patterns and identify candidate diseases.
• To establish a decision parameter based on minimum similarity value for reliable disease recommendation.

Proposed method

• The system uses a knowledge base containing symptoms and disease rules mapped via fuzzy logic to represent symptom severity and disease likelihood.
• Fuzzy temporal logic models the dynamic evolution of symptoms and their impact on disease probability over time.
• A fuzzy severity scale and weight factor are assigned to symptoms and diseases, adjustable based on temporal progression.
• The system computes disease similarity using a modified Euclidean metric to cluster symptom patterns.
• Elders' algorithm is applied to cluster symptom sets and identify the most similar disease patterns.
• The minimum similarity value among clusters is used as the decision parameter to select the most likely disease.

Experimental results

Research questions

• RQ1 How can fuzzy temporal logic improve the modeling of symptom progression and disease likelihood in medical diagnosis?
• RQ2 What is the optimal method for clustering symptom patterns to support diagnostic inference in an ICD-10 framework?
• RQ3 How does incorporating time-varying severity and weight factors enhance diagnostic accuracy in expert systems?
• RQ4 Can a similarity-based decision parameter using modified Euclidean distance effectively rank candidate diseases?
• RQ5 To what extent can such a system reduce the burden of initial physician diagnosis through automated symptom analysis?

Key findings

• The system successfully models symptom-disease relationships using time-varying fuzzy weights and severity scales.
• Clustering via Elders' algorithm with a modified Euclidean metric enables effective grouping of symptom patterns.
• The minimum similarity value across clusters serves as a reliable decision parameter for disease identification.
• The integration of ICD-10 with fuzzy temporal logic enhances diagnostic reasoning in dynamic clinical scenarios.
• The approach reduces dependency on manual initial diagnosis, offering a scalable framework for automated medical advice.
• The system demonstrates feasibility in recommending diseases based on symptom similarity, though no specific numerical accuracy rate is reported.