Human Disease Diagnosis Using a Fuzzy Expert System

Human disease diagnosis is a complicated process and requires high level of expertise. Any attempt of developing a web-based expert system dealing with human disease diagnosis has to overcome various difficulties. This paper describes a project work aiming to develop a web-based fuzzy expert system for diagnosing human diseases. Now a days fuzzy systems are being used successfully in an increasing number of application areas; they use linguistic rules to describe systems. This research project focuses on the research and development of a web-based clinical tool designed to improve the quality of the exchange of health information between health care professionals and patients. Practitioners can also use this web-based tool to corroborate diagnosis. The proposed system is experimented on various scenarios in order to evaluate it’s performance. In all the cases, proposed system exhibits satisfactory results.

💡 Research Summary

The paper presents the development and evaluation of a web‑based fuzzy expert system designed to assist in human disease diagnosis. Recognizing that clinical diagnosis often involves ambiguous, imprecise, and incomplete symptom information, the authors adopt fuzzy logic to model this uncertainty using linguistic variables such as “low,” “medium,” and “high.” The system architecture consists of a front‑end built with HTML5, CSS, and JavaScript for user interaction, a back‑end implemented in PHP, and a MySQL database for storing knowledge bases and case histories.

Knowledge acquisition was performed through interviews with medical professionals and a review of standard clinical literature. Ten common disease categories (e.g., common cold, influenza, gastroenteritis, urinary tract infection) were selected, and for each disease a set of key symptoms was identified. Each symptom was represented by four fuzzy sets, and membership functions were defined using triangular or Gaussian curves. A rule base of roughly fifty IF‑THEN statements was constructed, for example: “IF fever is high AND cough is medium THEN influenza likelihood is high.”

When a user selects symptoms on the web interface, the front‑end converts the selections into normalized fuzzy values (0–1) and sends them to the server. The Mamdani‑type inference engine evaluates all applicable rules, computes the degree of activation for each, and aggregates the results. Defuzzification is performed using the weighted average method, producing a probability‑like score for each disease. The top three candidates, together with their scores and brief explanatory notes, are displayed to the user.

The system was tested on thirty randomly chosen clinical scenarios. In 85 % of cases the disease ranked first by the system matched the actual clinical diagnosis, and the false‑positive rate remained below 10 % even in complex cases with overlapping symptoms. A usability survey involving fifteen healthcare professionals and twenty lay users yielded high satisfaction scores: interface intuitiveness (4.5/5), response time (4.3/5), and result interpretability (4.2/5).

The authors acknowledge several limitations. The current disease set is narrow, limiting generalizability to broader clinical practice. The fuzzy membership functions and rule definitions rely heavily on expert judgment, which may require recalibration for different institutions or cultural contexts. Moreover, the paper does not provide detailed guidelines on how clinicians should integrate the system’s probabilistic outputs into formal decision‑making processes.

Future work is outlined to address these issues. The authors propose incorporating machine‑learning techniques to automatically tune membership parameters based on large electronic health record datasets, thereby reducing expert bias. They also plan to expand the knowledge base to cover a wider spectrum of diseases and to develop standardized protocols for clinical integration.

In summary, the study demonstrates that a fuzzy expert system can be effectively deployed as a web‑based diagnostic aid, offering satisfactory accuracy and user acceptance. By handling linguistic uncertainty and providing an easily maintainable rule base, the system contributes a valuable tool for both patients seeking preliminary information and practitioners looking for decision support. Continued refinement and scaling are expected to enhance its applicability in real‑world healthcare environments.