A neural network study of the correlation between metabolic–cardiovascular diseases and disability in elderly people

Numerous studies have affirmed the existence of a correlation between various cardiovascular diseases and functional decline in elderly people. Not much information, however, is available concerning the overall effect of various, possibly coexisting, cardiovascular pathologies, or metabolic conditions notoriously related to them, on determining disability. We wanted to verify if it were possible to assess: (1) The overall importance of various metabolic and cardiovascular diseases which elderly people often suffer from contemporaneously in determining a condition of not self-sufficiency; (2) The possibility of predicting a condition of not self-sufficiency in relation to the above-mentioned pathologies. In order to achieve this aim, we used an artificial neural network: a statistical–mathematical tool able to determine the existence of a correlation between series of data and, once ‘trained’, to predict output data given the input data. Although artificial neural networks have been applied in various areas of medical research, they have not been previously applied in geriatrics. We have applied this method to a sample of 179 elderly people, demonstrating that seven clinical–biological variables concerning their metabolic and cardiovascular conditions are strictly related, all together, to the presence or otherwise of a functional impairment. When tested on a sample of 20 ‘unknown’ elderly people, the trained network gave the correct answer — self-sufficiency or not self-sufficiency — in 95% of the cases. Despite the fact that the sample studied was relatively small, artificial neural networks are undoubtedly useful in predicting functional impairment in elderly people in relation to the presence of metabolic and cardiovascular diseases.