Probabilistic Expert System based on Bayesian Networks for the prediction of cervical cancer

• Autores: Paulino Flores, Luis A., Huayna Dueñas, Ana M.
• Formato: artículo
• Fecha de Publicación: 2019
• Institución: Universidad Nacional Mayor de San Marcos
• Repositorio: Revistas - Universidad Nacional Mayor de San Marcos
• Lenguaje: español
• OAI Identifier: oai:ojs.csi.unmsm:article/16360
• Enlace del recurso: https://revistasinvestigacion.unmsm.edu.pe/index.php/rpcsis/article/view/16360
• Nivel de acceso: acceso abierto
• Materia: Predictive models; Bayes methods; Prediction algorithms; Probabilistic Computing; Modelos Predictivos; Método de Bayes; Algoritmos de Predicción; Computación Probabilística

Cervical cancer is currently the fourth most frequent type of cancer in women. A large number of techniques from the Artificial Intelligence (AI) such as Neuronal Networks, Support Vector Machines (SVM), Decision Trees and others; have been used to deal with the problem of predicting this disease. The following paper shows the cervical cancer risk prediction, by implementing a probabilistic model based on Bayesian Networks and using 322 instances where we could retrieve 15 different features that are known information from each patient. The tests were made using the 40% of the whole dataset, confusion matrix and AUC indicator. The results show that this work has raised a 96% of success rate as well as 0.9864 in terms of the AUC indicator, in addition to this, the results suggest that Bayesian Networks are able to reach a high performance and provide transparency during the inference process at the same time, something that does not happen in many other techniques, and that they are really efficient to face this sort of prediction problems.