Author(s): Pazo, María ; Boente, Carlos ; Albuquerque, M.T.D. ; Roque, Natália ; Gerassis, Saki ; Taboada, Javier
Date: 2022
Persistent ID: http://hdl.handle.net/10400.11/10122
Origin: Repositório Científico do Instituto Politécnico de Castelo Branco
Subject(s): Health data; Information theory; Ordinary kriging; Target analysis
The health status of the service sector workforce is a great unknown for medical geography. Despite the advances carried out by spatial epidemiology to predict spatial patterns of disease incidence, there are important challenges unsolved. In particular, the main issue resides in the ability to effectively simplify and visually represent the problem domain, given the need to cover very different service activities and, at the same time, consider the impact of numerous emerging risk factors such as those stemming from bioclimatic and socioeconomic variables. This article proposes a new approach that allows to consider, simplify, prioritise and visualise multiple occupational health risk factors giving rise to not healthy workers. For that, it is used a twofold approach based on an innovative synergy between Bayesian machine learning and geostatistics, to analyse up to 74.401 occupational health surveillance tests gathered between 2012-2016 in Spain. This solution allows to extract relevant patterns over those risk factors that cannot be further discriminated in the Bayesian network, such as spine or limbs observations, depicting distribution maps of key differentiating variables computed by an ordinary kriging approach.
