An Explainable Rule-Based Expert System with Certainty Factor for Early Acne Detection

Abstract

Introduction — Early identification of acne-related skin conditions is important to support preventive care and increase user awareness before professional dermatological consultation. Early detection is often challenged by uncertainty in symptom perception and the limited availability of accessible digital decision-support tools. Many digital health systems rely on machine learning models that lack interpretability and transparency. This study proposes an explainable rule-based expert system incorporating a Certainty Factor (CF) approach to support early acne detection in a bio-digital context.

Methods — The objective of this research is to design and implement a digital expert system capable of handling uncertainty in user-reported symptoms while providing transparent diagnostic reasoning. Dermatological knowledge is represented using structured rule-based models enriched with expert-defined certainty values. User inputs are modeled with confidence levels ranging from 0 to 1 and processed using a forward-chaining inference mechanism combined with CF calculations. The system is implemented in Python and deployed through a web-based interface.

Results — Experimental evaluation was conducted using five structured test scenarios representing different symptom combinations. The system generated early acne condition outputs with confidence scores ranging from 35.0% to 72.0%. The highest confidence score (72.0%) was obtained in the Contact Dermatitis scenario, while incomplete symptom inputs produced no diagnostic output. Each valid case included explicit reasoning traces linking symptoms to computed confidence values.

Conclusion — The system demonstrates an interpretable and uncertainty-aware expert system framework for early acne detection. Its novelty lies in integrating CF reasoning within an explainable rule-based architecture for transparent digital decision support.