Evaluación y aplicación de algoritmos de inteligencia artificial explicada para apoyar la toma decisiones médicas en la salud fetal .

Abstract

In recent decades, medicine has found in artificial intelligence (AI) a driver in the search for solutions. Thus, through machine learning algorithms, tasks as specific as patient diagnosis are favored by their ability to predict based on data. However, AI is presented as a "black box" that hides the learning process for the generation of results. In this research presented as a curricular degree work, the application of the Explained Artificial Intelligence (XAI) SHAP algorithm is carried out as an explanatory method for the evaluation of fetal health. For which the cardiotocography data obtained from the UCI Machine Learning repository is used for the application of AI in the area of obstetrics. To apply the XAI method, a selection of an AI algorithm was first made, for which 3 very popular machine learning methods were evaluated, such as Support Vector Machine, Random Forest and an Artificial Neural Network, which through the implementation and evaluation of A Systematic Literature Review was able to determine its wide use in the application of XAI. Random Forest was the algorithm that obtained the best results in the classification of fetal health, being the one selected for the application of XAI. The results have been evaluated with a specialist in the area of obstetrics who has validated the explanatory capacity of the SHAP algorithm and has found in this method a help to interpret the outputs of an AI algorithm, which is considered favorable for its implementation in medicine. specifically in the area of obstetrics.