Meteorological factors for subarachnoid hemorrhage in the greater Düsseldorf area revisited: a machine learning approach to predict the probability of admission of patients with subarachnoid hemorrhage.

Abstract

Reported data regarding the relation between the incidence of spontaneous subarachnoid hemorrhage (SAH) and weather conditions are conflicting and do so far not allow prognostic models. Admissions for spontaneous SAH (ICD I60.*) 2009-2018 were retrieved form our hospital data base. Historical meteorological data for the nearest meteorological station, Düsseldorf Airport, was retrieved from the archive of the Deutsche Wetterdienst (DWD). Airport is in the center of our catchment area with a diameter of approximately 100 km. Pearson correlation matrix between mean daily meteorological variables and the daily admissions of one or more patients with subarachnoid hemorrhage was calculated and further analysis was done using deep learning algorithms. For the 10-year period from January 1, 2009 until December 31, 2018, a total of 1569 patients with SAH were admitted. No SAH was admitted on 2400 days (65.7%), 1 SAH on 979 days (26.7%), 2 cases on 233 days (6.4%), 3 SAH on 37 days (1.0%), 4 in 2 days (0.05%), and 5 cases on 1 day (0.03%). Pearson correlation matrix suggested a weak positive correlation of admissions for SAH with precipitation on the previous day and weak inverse relations with the actual mean daily temperature and the temperature change from the previous days, and weak inverse correlations with barometric pressure on the index day and the day before. Clustering with admission of multiple SAH on a given day followed a Poisson distribution and was therefore coincidental. The deep learning algorithms achieved an area under curve (AUC) score of approximately 52%. The small difference from 50% appears to reflect the size of the meteorological impact. Although in our data set a weak correlation of the probability to admit one or more cases of SAH with meteorological conditions was present during the analyzed time period, no helpful prognostic model could be deduced with current state machine learning methods. The meteorological influence on the admission of SAH appeared to be in the range of only a few percent compared with random or unknown factors.