Analysis Group Researchers Identify Neighborhood Characteristics That May Predict Psoriasis in Quebec, Canada

June 18, 2024

Psoriasis, a chronic disease that causes an itchy, painful rash, affects 60 million people worldwide. Although the impact of individual-level health behaviors such as diet, alcohol consumption, smoking, and exercise are well understood, the impact of larger social, cultural, economic, and environmental conditions where patients live has not been studied.

To fill this gap, an Analysis Group team led by Principal Jimmy Royer, Vice President Irina Pivneva, Director of Data Science Maxime Leroux, and Data Scientist Kathleen Chen, in collaboration with researchers from McGill University, Centre de Recherche Dermatologique de Québec, Memorial University of Newfoundland, Exponent Inc., and the University of Manchester, analyzed the relationship between the incidence of psoriasis in different areas of Quebec and more than 400 neighborhood covariates available from the Canadian Urban Environmental Health Research Consortium, including air, noise, and light pollution measures; greenness indicators; “blue spaces,” or outdoor spaces with water; and climate metrics.

The investigators used advanced tree-based machine learning to model the complex, multivariable datasets, identifying 46 neighborhood factors that had an impact on predicting the probability of high psoriasis incidence. Top predictors such as higher ultraviolet radiation, maximum daily temperature, soil moisture, and urbanization all had a negative association with incidence of psoriasis, meaning that higher predictor values were associated with lower incidence of psoriasis in a given area. However, higher levels of nighttime light brightness had a positive association, and middle-class socioeconomic factors in a neighborhood also suggested higher incidence of psoriasis.

The study, “Tree‑Based Machine Learning to Identify Predictors of Psoriasis Incidence at the Neighborhood Level: A Populational Study from Quebec, Canada,” was published in March 2024 by the American Journal of Clinical Dermatology. Analysis Group services were provided pro bono.

Read the study