Deep Learning to Predict Outpatient Visits by Respiratory Illnesses in a High PM10 Environment

Betetta-Gomez J.; Medina-Ramos C.; Tafur-Anzualdo I.; Diaz-Diaz F.D.M. · Proceedings of the 2020 IEEE 27th International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2020 (2020) · DOI: 10.1109/intercon50315.2020.9220211

Resumen

This study is focused on the implementation of a model based on artificial neural networks capable of predicting three respiratory diseases related to air pollution by 10-micrometers diameter particles, so-called Particulate Matter (PM10). In this way, the proposed model can successfully predict the number of hospital admissions related to asthma, bronchitis, and rhinopharyngitis. The successful predictive results make the model a useful tool to know the hospital requirements in advance. Furthermore, it is significant to have worked with Keras, a python deep learning library in the Google Colaboratory platform, with all the computing processes being performed in the cloud, and with minimal use of personal computer resources. The horizon of the data for PM10 corresponds to the measurements of SENAMHI and INEI for one year. Also, the data of the three respiratory diseases mentioned above corresponds to the cases registered in the area of pneumology at the National Hospital of Vitarte, during the year of study. Plus, the model includes temperature and wind speed as environmental factors that, together with the PM10, give rise to respiratory diseases which are the object of the study. Taking into account the predictive results (98.5%), the model is a promising tool for the prediction of diseases in the area of pneumology and the medical care necessary for their treatment.

Instituciones: Universidad Nacional de Ingeniería