城市自然与社会环境对新型冠状病毒肺炎发病率的影响

Identify the natural and socio-economic influencing factors of the new coronavirus pneumonia(COVID-19) incidence rates in Chinese cities

在城市尺度上探究了2020年1月1日~3月5日期间城市规模、医疗资源水平等城市自然与社会经济因素对新冠肺炎人群发病率的影响.基于多种传统统计模型与机器学习方法识别了新冠肺炎发病率的关键影响因子.基于新兴的可解释机器学习框架,探究了发病率与关键影响因子之间的非线性联系.结果表明:城市新冠肺炎发病率受到人口迁入、城市规模、城市医疗资源水平等方面的多要素影响,其中武汉迁入率具有最高的相关系数(相关系数达到0.43),其次为人口增长率(相关系数为0.38);人口迁徙、城市规模以及医疗服务资源水平均为关键影响因子;关键影响因子与人群发病率存在非线性关系:武汉迁入率对新冠肺炎发病率的影响曲线呈现S形,在迁入率大于2%进入平台期,人口密度的影响近似线性;人均GDP的影响呈现出明显的倒U型曲线特征,以人均GDP超过10万元为拐点;城市建设需要适当控制人口密度,避免城市人口分布过于紧凑.提升高经济水平地区的经济发展,从而带来更多健康收益.

This study explored the effects of both natural and socio-economic factors, such as city size and healthcare capacity, on the spreading of COVID-19 in China’s urban population from January 1 to March 5, 2020. Several statistical models and machine learning methods were used to identify the key determinants of the incidence rate of COVID-19. Based on the interpretable machine learning framework, possible nonlinear relationships between incidences and key impact factors were explored. The results showed that the incidence rate of COVID-19 in cities was influenced by several factors simultaneously. Among the factors, the population inflow rate from Wuhan was the factor that showed the highest correlation coefficient(0.43), followed by the population growth rate(0.38). Population migration size, city size and healthcare capacity were the key influencing factors. Nonlinear relationships existed between the key influencing factors and incidence rates. To be specific, the inflow rate from Wuhan had a S-shaped relationship and reaches an asymptote after 2%;the population density had an approximately linear relationship;the per capita GDP showed an evident inverted U curve with the per capita GDP over 100,000 yuan as the inflection point. City development needs to pay more attention to population density control and economic growth in order to bring more health benefits.