O uso de redes neurais artificiais na análise de dados clínicos e genéticos do câncer de pulmão

• Autor(a) principal: Lima, Débora Virgínia da Costa e
• Data de Publicação: 2020
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Repositório Digital da Universidade Federal Rural do Semi-Árido (RDU)
• Texto Completo: https://repositorio.ufersa.edu.br/handle/prefix/6671
• topic: Aprendizado de máquina; Inteligência artificial; Dados oncológicos; CIENCIAS BIOLOGICAS::BIOTECNOLOGIA

The health area produces a large volume of data, which when analyzed help in the prevention, diagnosis and implementation of public policies, but the analysis of so much data is complicated, so big data techniques appear as a possible solution to this problem. The present study aimed to analyze clinical and genetic data of lung neoplasms, using artificial neural networks as a method for data prediction. The steps of this study are: using clinical and genetic data present in international databases (GDC Data Portal) and local (Hospital Cancer Registry of the Mossoroense League for the Study and Fight against Cancer), a selection, organization and pre-processing of the data for insertion in the neural network, an implementation, training and testing of the artificial neural network algorithm and the results of the prediction and validation of the neural networks. The initial step was to recover the clinical and genetic data from databases, both international and local, and were filtered according to the variables to be used in neural networks. After a filtering and organizing step, was started the preprocessing of the data to allow insertion into the algorithm. Five networks were built, three implemented with international clinical data, a network implemented with international and local clinical data and a network with genetic data. The network that performed best was one of the networks with international clinical data that presented its validation accuracy = 86.06%. Demonstrating that the implementation of neural networks can be a good alternative for analysis in cancer patients.