Aplicación de técnica de inteligencia artificial para la detección de patrones en información ginecológica

Abstract

Abstract: In recent years there has been an increase in the use of artificial intelligence, including machine learning which has supported the development of systems that learn instinctively, processing millions of bits of data, to find patterns that allow generating knowledge to benefit health professionals and patients. For this reason, the present investigation intends to study how the application of machine learning techniques allows for the identification of patterns in gynecological information. The dataset utilized here contains 4505 records that are related to diseases in women. This dataset has 30 fields that describe aspects such as age, diseases, medications provided, treatments applied, among others. The incorporation of the CRISP-DM methodology in the development of the project allows for better management of key aspects such as data preprocessing. Three unsupervised algorithms have been used: k-means, DBSCAN and Hierarchical Clustering. For the validation of the results, the metric used is Silhouette. The results obtained provide an important contribution in the identification of groups that are related to one or several specific diseases.