Machine Learning Applied to Health Information Exchange

José Machado, Filipe Miranda, Ana Sousa, Julio Duarte, António Abelha

OAI: oai:igi-global.com:298634 • DOI: 10.4018/ijrqeh.298634

The interest in artificial intelligence (AI) has grown in the last few years. The healthcare community is no exception. The present work is focused on the exchange of medical information, using the Health Level Seven (HL7) international standards. The main objective of the present work is to develop an AI model capable of inferring if for a given hour exists a peak in the number of exchanged messages. To accomplish that, two different deep learning models were created, an artificial neural networks (ANN) and long short-term memory (LSTM). The intention is to observe which is capable to perceive the situation better considering the environment and features of a healthcare facility. Using laboratory-generated data, it was possible to simulate variations and differences in “traffic.” Comparing the LSTM vs. ANN model, the first is capable of outputting peaks better but for considered mean values that do not happen. For the given context, predicting the peak is essential, so the LSTM is the right choice and uses fewer features that are good regarding performance.