Di Stefano Mirko, Aragão Leonardo, Ambrosio Giuseppe, Ciangottini Diego, Duma Cristina, Lubrano Pasquale, Martelli Barbara, Salomoni Davide, Sergi Giusy, Spiga Daniele, Stracci Fabrizio, Ronchieri Elisabetta, Storchi Loriano, Cutini Sara
Department of Statistical Sciences "Paolo Fortunati", University of Bologna, Via Belle Arti 41, Bologna, 40126, Italy.
CMCC Foundation, Euro Mediterranean Center on Climate Change, Viale C. Berti Pichat 6/2, Bologna, 40127, Italy.
Sci Rep. 2025 Jun 5;15(1):19851. doi: 10.1038/s41598-025-04887-4.
Although a worldwide health crisis, the COVID-19 pandemic affected several geographical areas in Italy in very different ways in terms of infection rate, morbidity, and death. In the present work, we carefully studied the incidence rate in several Italian provinces and propose a complete data analysis strategy to explore, preprocess, and analyse the time series of COVID-19 positive and hospitalised cases with a daily cadency. We applied a new procedure, developed for unevenly sampled data (Discrete Correlation Function), to perform the cross-correlation analysis looking at possible correlation between COVID-19 positive and hospitalised cases with the air quality during the first pandemic wave. It is a completely new approach, that makes use of techniques used in transversal fields, such as signal processing and astronomy. The study suggests some plausible correlations between COVID-19 time series and NO related air pollutants. Instead, differently from what often has been reported, we did not find any specific correlation between COVID-19 infection and PM[Formula: see text] air pollutant. We further corroborate the results using a Machine Learning approach that uses Random Forest and the Permutation Feature Importance Analysis to include a wider set of possible risk factors, founding the same dependence between COVID-19 cases and NO related air pollutants.
尽管新冠疫情是一场全球健康危机，但在感染率、发病率和死亡率方面，它对意大利的几个地理区域产生了截然不同的影响。在本研究中，我们仔细研究了意大利几个省份的发病率，并提出了一种完整的数据分析策略，用于探索、预处理和分析每日新冠确诊病例和住院病例的时间序列。我们应用了一种针对不均匀采样数据开发的新程序（离散相关函数），进行互相关分析，以研究第一波疫情期间新冠确诊病例和住院病例与空气质量之间的可能相关性。这是一种全新的方法，利用了信号处理和天文学等横向领域中使用的技术。该研究表明，新冠时间序列与与一氧化氮相关的空气污染物之间存在一些合理的相关性。相反，与经常报道的情况不同，我们没有发现新冠感染与细颗粒物空气污染物之间存在任何特定相关性。我们使用机器学习方法进一步证实了结果，该方法使用随机森林和排列特征重要性分析来纳入更广泛的可能风险因素，发现新冠病例与与一氧化氮相关的空气污染物之间存在相同的相关性。
