Use of the evaluation of fuzzy logic for Human Papilloma Virus risk prediction

Abstract

The decrease in morbidity and mortality caused by cervical cancer (CC) is still a challenge in developing countries. As the main cause, the lack of an organized screening for patient recruitment obstructs the detection of precursor lesions in time for effective treatment. This paper evaluates the use of an intelligent computer system using fuzzy logic as a reading method to specialist predicting risk of pre neoplasic lesions. The system was constructed using six major cofactors in the development of the disease as input variables. The chosen cofactors were: age, age of the first sexual intercourse, number of sexual partners, tobacco smoking, time using hormonal contraception and concomitant presence of immunosuppression disease. The answers produced by the system were compared to those provided by an expert that evaluated 82 hypothetical cases formulated by the authors. The result was a degree of agreement above 81%, which certifies the proposed method. In addition, there was the calculation of sensitivity and specificity attributed to method with a result above 80% in most categories analyzed. We conclude, then, that the fuzzy logic is a suitable reader of specialist thinking, which can be used, if validated, by the Public Health Network for the execution of a schedule organized and consequent increasing recruitment of patients.