Skip to Main content Skip to Navigation
Master Thesis

Création et validation d'un modèle de prédiction du nombre quotidien d'AVC en fonction de variables météorologiques

Abstract : Background: neurovascular events are part of diagnostic and therapeutic emergencies, and require recourse to emergency services. Cardiovascular risk factors are identified and accepted by consensus. However, the factors triggering these events are still imprecise. The influence of climate is plausible, supported by extensive work in the literature. The aim of this study is to establish a model for predicting the number of daily strokes based on meteorological data, in order to identify unfavorable climatic conditions and to adapt the health response according to the days at risk. Methods: 8,910 cases of ischemic, hemorrhagic and TIA strokes hospitalized at Bordeaux University Hospital, as well as meteorological variables were extracted over six consecutive years. To identify the determining climatic variables, an univariate analysis was performed using the Mann-Whitney-Wilcoxon test. The construction of the prediction model is based on the identified climatic parameters, using a generalized additive model in multiparametric linear regression. Its effectiveness was tested over the period from January 2019 to December 2019 by calculating the Pearson correlation coefficient and comparing the predictions with the observations. Results: in univariate analysis, there is a significant difference in maximum and average ambient temperature, and maximum and average relative humidity between days with "more than two daily cases of stroke" and those with "two daily cases or less", according to immediate or delayed effect up to six days. Likewise, there is a significant difference in the conditions of occlusion of the sky by cloud masses (minimum and average daily total cloudiness) with an immediate or delayed effect up to seven days before the event of interest when the threshold is considered to be plus or minus five cases of stroke daily. The maximum daily relative humidity is significantly different between these last two groups, with an immediate or delayed effect of five days. The model established makes it possible to predict the daily number of strokes for 62.1% of the days in 2019, especially those where the observed frequency is between three and six cases. The precision is acceptable in plus or minus two cases. It is faulted for observed frequencies less than three and greater than six per day, counting for one third of the days in 2019. Conclusion: the established model could be used in practice to identify days when the expected frequency would be between three and six per day. However, there is a large margin of error in a third of cases, limiting its reliability.
Complete list of metadata

https://dumas.ccsd.cnrs.fr/dumas-03085855
Contributor : Médecine Ub Connect in order to contact the contributor
Submitted on : Tuesday, December 22, 2020 - 9:18:18 AM
Last modification on : Wednesday, January 27, 2021 - 3:12:40 AM

File

Med_Spe_2020_Daligaud.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : dumas-03085855, version 1

Collections

Citation

Xavier Daligaud. Création et validation d'un modèle de prédiction du nombre quotidien d'AVC en fonction de variables météorologiques. Sciences du Vivant [q-bio]. 2020. ⟨dumas-03085855⟩

Share

Metrics

Record views

44

Files downloads

47