Time Series Forecasting to Predict the Evolution of the Functional Profile of the Elderly Persons

There are many pathologies and capacity losses that progress with a similar evolution profile in certain groups of people. Health professionals are becoming increasingly knowledgeable in anticipating the development of these pathologies through preventive medicine. However, the increasing amount of data, coming from the collection of information from a larger number of patients, makes it difficult to analyse it manually. In the case of gerontology, it is difficult to classify in groups the evolution of the elderly for common pathologies in that age group. To be able to do this would make it possible to know in advance how a pathology or capacity loss will progress in an ageing person and to apply the corresponding preventive procedures. There are already works that try to improve the results of preventive medicine, but these are focused on analysing the current state of the patient and not their foreseeable future. In this article, time series forecasting by means of recurrent neural networks is used to analyse the evolution of the functional profile of ageing people as a time series. Based on the patterns contained in the patient’s time series and in the training of a model with data from previous patients, it is possible to determine the future evolution in patients with a similar history. To do this, functional profile data collected on an assessment platform developed by the authors of this article is used.