Development of Artificial Intelligence Algorithms to Analyse Weather Conditions for the Prediction of Cerebrovascular Accidents

Abstract

The aim of this study is to investigate the influence of weather conditions on the occurrence of cerebrovascular accidents. To date, the literature has noted fluctuations in the incidence of spontaneous subarachnoid haemorrhage depending on external factors, but the strength of this relationship is not high. To progress beyond the state of the art, relevant meteorological and medical data were acquired, on the basis of which further considerations were carried out. This relationship was analysed using a classic statistical approach and machine learning methods, particularly using neural networks. In addition, the theoretical and statistical dependencies of the machine learning algorithms are described in this paper. The main output of this paper is a description of a recurrent neural network model that can detect weather conditions that increase the risk of a vascular incident with decent confidence. The presented technological solutions play a significant role in prediction, prevention, and a personalised approach to a patient. Moreover, the study is inextricably linked to the use of numerical head models and machine learning both in diagnosis and therapy. The confirmation of the existence of a relationship between specific weather conditions and vascular incidents can help not only medical facilities but also patients. Thanks to such forecasting methods, hospitals could prepare for periods of an expected increased number of patients requiring urgent help. On the other hand, patients from risk groups (or their families), knowing about the higher probability of, e.g., stroke or heart attack, could take necessary precautions, e.g., not plan strenuous work or stay constantly surrounded by family or caretakers.