FERRAMENTA COMPUTACIONAL PARA PREDIÇÃO DE RISCO DE INFECÇÃO POR PAPILOMAVIRUS HUMANO (HPV) PELA LÓGICA FUZZY

Abstract

Introduction: The Human Papillomavirus (HPV) is one of the most common sexually transmitted infections (STIs) and responsible for approximately 99% of cervical cancers in the world. Thus, there is a need to insert early search methods that help women to be referred for preventive examination of the uterine cervix. Therefore, one of the possible improvements for the active search in health came with the advent of technological evolutions. In this context, bioinformatics emerges as an instrument to aid in the prediction and early diagnosis of diseases. The objective was to develop a computational tool for HPV risk prediction using fuzzy logic. Methodology: This is a computational model using fuzzy logic tools to predict women with greater predisposition to HPV risk exposure. The research was authorized by the Research Ethics Committee under number 2392728. A semi-structured questionnaire was applied, in addition to the collection of vaginal smear samples in women over 18 years old who sought the Health Units. the selection of the choice of variables that were used as input to the software. From this perspective, 6 (six) cofactors considered as potential and important predictors for inputs and one output were selected. Among them are: age, education, smoking, sexual behavior, number of pregnancies, use of contraceptives. Then, for the development of the algorithm, two more phases took place. The first was the PHP language and MySQL database for entering the collected data, separating and standardizing information, which aimed at faster search and comparison through computer systems. The second started with the layout of the fuzzy sets to assemble the Fuzzy Logic in the system. In order to build the models, it was necessary to divide the input data into degrees of risk, as well as the output set, which represented the final fuzzy set. After defining the input and output sets, the base data was entered into the HPV Risk Calculator system. Based on the aforementioned indicators, the calculator made reference to the data reported in the interviews and exams of each research participant. First, risk cofactors available in the literature were listed, added to the collection results to construct the calculation with the determination of risk, after the analysis, the RISK set was concatenated into 3 sets. After the fuzzified data, the following items were obtained as variables for availability in the risk calculator: GREEN = [0 - 30%], low risk; YELLOW = [31 - 50%], medium risk; RED = above 50%, high risk of HPV infection. Results: 562 women were surveyed. From the results obtained in the epidemiological and cervical findings of the women participating in the research, 400 were used for software training and 162 for system validation. Conclusion: The software was effective in validation and testing. Its purpose was to search for women early to undergo the preventive examination of the uterine cervix in primary care, which is considered the point of entry for users in the Brazilian health system. Finally, we consider the need to increase the number of collections to achieve an accuracy of 95% for optimal validation of the calculator.