Evaluation of Applied Artificial Neuronal Networks with a Timely Cervical Cancer Diagnosis in an Emerging Economy

Around the world, cancer is the main cause of deathDeaths. The most common types of cancer in women are breast, colorectal, lung, and cervical cancerCervical cancer (CC). CC has become relevant since, in emerging economiesEmerging economies such as Mexico, cases have continued to increase, caused by restrictions imposed on hospitals because of COVID-19, which complicated cancer prevention and patient care, as well as a low allocation of economic resources to public health systems that has limited diagnostic testing. For this reason, we propose the development of artificial neuronal networksArtificial Neuronal Networks (ANNs), which help give appropriate diagnoses. ANNs use information about real cases and with training discover the relationships among input variables, granting them a level of impact and forecasting a result. We propose five ANNs, whose basic input variables are the risk factorsRisk factors for CCCervical cancer, as well as the diagnostic tests: hybrid capture, cytology, colposcopyColposcopy, which were added according to the algorithm used for the detection of disease. The output variable was the biopsy. We evaluated the accuracyAccuracy and precisionPrecision of the networks to correctly classify cases as positive or negative. The ANN that had the best performance with these measurements was 5 with 84.85 and 100%, followed by 2 with a 78.79 and 50%. The difference between these ANNs is found in the diagnostic tests that were used, ANN 5 uses all three, while 2 just uses cytology. We recommend suing ANN 2 for a timely diagnosisDiagnosis, while ANN 5 requires the information from the colposcopy, which is a study that is done at the end.