A Linear Stochastic System Approach to Model Symptom Based Clinical Decision Support Tool for the Early Diagnosis for Psoriasis, Seborrheic Dermatitis, Rosacea and Chronic Dermatitis

Abstract

Prediction models provide the probability of an event. These models can be used to predict disease's outcomes, reccurencies after treatments. This paper presents an expert system called Symptom Based Clinical Decision Support Tool (SBCDST) for early diagnosis of erythemato-squamous diseases incorporating decisions made by Bayesian classification algorithm. This tool enables family practitioners to differentiate four types of erythemato-squamous diseases using clinical parameters obtained from a patient. In SBCDST, Psoriasis, Seborrheic Dermatitis, Rosacea and Chronic dermatitis diseases are described by means of well-classified set of attributes. Attributes are generated from the typical sign and symptoms of disorder. Based on our clinical results, tool yields 72%, 93%, 89% and 95% correct decisions on the selected dermatology diseases respectively. System proposed will provide the opportunity for early diagnosis for the patient and the expert medical doctor to take the necessary preventive measures to treat the disease; and avoid malpractice which may cause irreversible health damages.