Data

We obtain daily information from Shanghai Municipal Health Commission on: the number of confirmed cases and asymptomatic infections by district and county during the outbreak and Shanghai as a whole, the number of asymptomatic infections released from medical observation, and the number of cured cases, etc. Our codes and data can be found at https://github.com/Bai-Yu-Lan/SH-COVID19

 

我们每天从上海卫健委的官方网站获取:疫情期间各个区县以及上海市整体每日新增的确诊病例与无症状感染者数量与居住地、解除医学观察无症状感染者人数、治愈出院人数等信息。我们的代码与数据可以在https://github.com/Bai-Yu-Lan/SH-COVID19获取。

  1. Relevant Residence Information 患者居住地信息

  2. After manually obtaining the relevant residence information on that day from the Shanghai Municipal Health Commission, through update_address.ipynb, we update all_addresses.npy containing all addresses, and then obtain the district and street to which the new address belongs, the corresponding POI name and type, and the corresponding latitude and longitude coordinates through the Baidu Map API.

     

    从上海卫健委手动获取当日患者的居住地信息之后,通过update_address.ipynb,我们更新包含所有地址的all_addresses.npy,然后通过百度地图API获取新增地址所属区县与街道、对应的POI名称与类型、对应的经纬度坐标。

     

    For example, for No. 600 Yishan Road, Xuhui District, we can get: Shanghai, Xuhui District, Tianlin Street, Shanghai Sixth People's Hospital, Medical, (121.420···, 31.181···)

     

    例如,对于徐汇区宜山路600号,可以得到:上海市,徐汇区,田林街道,上海市第六人民医院,医疗,(121.420···, 31.181···)

     

    Since there are multiple address information corresponding to the same POI in the residence information, we encode each coordinate point in the format of [province and city]-[district]-[town]-[latitude and longitude] in order to facilitate the visualization.

     

    由于居住地信息中存在多个地址信息对应同一个POI的情况,为了便于进行可视化,我们以 [省市]-[区县]-[街道]-[经纬度] 的格式对每一个坐标点进行编码。

     

    For example, Shanghai, Xuhui District, Tianlin Street, (121.420···, 31.181···) -> 021-01-02-123.

    ,where Xuhui district corresponds to district code 01, Tianlin street corresponds to Xuhui district street code 02, (121.420···, 31.181···) corresponds to Tianlin street coordinate point code 123.

     

    例如,上海市,徐汇区,田林街道,(121.420···, 31.181···) -> 021-01-02-123,

    其中,徐汇区对应区县编码01,田林街道对应徐汇区街道编码02,(121.420···, 31.181···) 对应田林街道的坐标点编码123。

     

    With such coding, we can correctly count the number of times POI was notified within 21 days, and store all POI notified in Shanghai in daily_report_by_address.csv : if it was notified on the same day, the corresponding position will be set to 1.

     

    通过这样的编码方式,我们可以正确统计POI在21天之内被通报次数的情况,将上海所有POI被通报情况存入 daily_report_by_address.csv 中:若当日被通报,则对应位置将被置为1。

     

     

  3. Occupancy of Medical Resources 医疗资源使用情况

  4. In this section, we estimate the current occupancy of the module hospitals (Fangcang) by counting the number of new asymptomatic infections in the daily closed-loop management and social screening in Shanghai and the number of asymptomatic infections discharged from daily observation; we estimate the occupancy of hospitals by counting the number of new diagnoses per day and the number of discharges per day. Due to the lack of data on the number of asymptomatic infected persons released from daily observation in each district and county, we are unable to estimate the use of the pods at the district and county level for the time being.

     

    在这一部分中,我们通过统计上海每日闭环管理与社会面筛查中新增的无症状感染者、每日解除观察的无症状感染者来估计目前方舱的占用情况;通过统计每日新增确诊人数、每日出院人数来估计医院的占用情况。由于缺少各个区县每日解除观察无症状感染者的数据,我们暂时无法估计区县级别方舱使用情况。

     

Rt Analysis

How do we estimate the effective reproduction number Rt in Shanghai?

  1. Estimation method

    In this work, we use the EpiEstim R package to estimate the effective reproduction number of COVID-19 in the Shanghai and each district. EpiEstim is an open-source R package to estimate the time-varying reproduction numbers of an epidemic from the incidence time series and has been used to estimate the transmission of Ebola virus, Zika virus, and SARS-CoV-2 in varying countries.

    我们采取了R语言工具包EpiEstim 来估计上海市及各行政区域的有效传染数变化趋势。EpiEstim是一个开源的R语言工具包,用来估计传染病传播过程中的时变传染数, 已经被应用于埃博拉、寨卡病毒、新冠病毒传播场景中。

    As the newly reported number of cases deviate from the actual number due to the delayed and limit accuracy of testing, the resulting of Rt estimates can be highly variable. Given this, EpiEstim provides a way to measure the reproduction number over a time period, Rt,τ, where τ is the size of the window in which the Rt is calculated. One can expect that Rt,τ would be less variable and more precise as the window size τ increases. In our case, we select τ as 5 days.

    由于核酸筛查与异常复查等因素,造成每日通报的新增感染数量具备不确定性。Rt估计的结果会出现不稳定性。 鉴于此,EpiEstim提供了在一段时间窗内估计稳定Rt的方法,即Rt,τ, 其中τ是我们可以预设的时间窗大小,随着时间窗τ的增大,我们估计的Rt,τ也会更加稳定。 在本项目中,我们设置时间窗τ为5天。

  2. Estimated Rt

    Time-varying infections and Rt in Shanghai and each district:

    上海市及各行政区域每日新增感染及Rt 动态变化(更新日期:2022-06-01):



















  3. What Is "R0"? Gauging Contagious Infections

    In epidemiology, the basic reproduction number, or basic reproductive number (sometimes called basic reproduction ratio or basic reproductive rate), denoted R0 (pronounced R nought or R zero), of an infection is the expected number of cases directly generated by one case in a population where all individuals are susceptible to infection [1].

    在流行病学中,感染的基本传染数表示为 R0 (读作 R nought 或 R zero), 是指在所有个体都易受感染的人群中,一个病例直接产生的感染人数的预期数量 [1]。

    R0 tells you the average number of people who will contract a contagious disease from one person with that disease. It specifically applies to a population of people who were previously free of infection and haven’t been vaccinated. For example, if a disease has an R0 of 18, a person who has the disease will transmit it to an average of 18 other people. That replication will continue if no one has been vaccinated against the disease or is already immune to it in their community.

    R0 告诉你平均有多少人将从一个患有该疾病的人那里感染该传染病。它特别适用于所有人都没有免疫力的人群。 例如,如果一种疾病的 R0 值为18,那么一个患有该疾病的人平均会传播给 其他18个人。如果社区中没有人接种过该疾病的疫苗或已经对该疾病有免疫力,那么这种传播将继续下去。

  4. What do Rt values mean?

    In general, we use the dynamic effective reproduction number, Rt, to track the time-varying spread of epidemic.
    Key measure: effective number of infections Rt < 1 for at least 2 weeks.

    一般情况下,我们可以使用动态变化的有效传染数, Rt, 来跟踪传染速度的变化。
    关键衡量标准:有效传染数 Rt < 1 至少达两周。

    In theory, an Rt (effective number of secondary cases per infected case in a population) of less than 1 is the best indication that the outbreak is under control and on a downward trend. In countries with large populations, Rt may vary across the population and should be estimated at the subnational level [2].

    从理论上讲,Rt (人群中每个感染病例导致继发病例的有效数量) 低于1,可以最好地表明疫情已得到控制并呈下降趋势。 在人口众多的国家中,Rt在全体人口中可能有差异,应在亚国家级进行估算 [2]。

    参考资料:
    [1] Basic reproduction number, from Wikipedia
    [2] WHO: 在COVID-19背景下调整公共卫生和社交措施时的公共卫生标准

仿真说明

SH-SEIR模型在SEIR模型的基础上,结合上海实际封控措施,增加社会面活动、居家隔离、 集中隔离三种封控状态,考虑了病毒传播、病情发展、核酸检测与转运、区域封控、 新增阳性患者的流行病学调查等机制,拟合上海自2月底以来的每日新增阳性病例曲线, 并针对疫情期间的若干关键环节,推演出不同措施下的疫情发展情况。

场景一:调控3月28-3月31日人均密接数

3月27日20:23,上海市新冠肺炎疫情防控工作领导小组办公室发布公告,上海将以黄浦江为界分两批实施封控管理, 进行全市范围内“切块式、网格化”核酸筛查,封控区域内,住宅小区实施封闭式管理,所有人员足不出户, 人员和车辆只进不出[1]。

3月28日5时起,浦东、浦南及毗邻区域先行实施封控,开展核酸筛查, 浦西地区重点区域继续实施封控管理;4月1日3时起,对浦西地区实施封控。3月28日至3月31日期间, 浦西非封控区域出现了较高的生活物资采购需求,人均密接数相较于28号前有所上升, 在我们的模型中这一时间段的人均密接数设置为15。

场景一展示了在其他政策与条件不变的情况下, 该时间段内非封控区的人均密接数对疫情走势、最终感染人数和最大隔离资源占用量的影响。


研究表明,封控时间节点和封控力度都对病毒传播具有极大影响[2,3]。 场景二模拟了其他不同封控时间点对于疫情走势、最终感染人数和最大隔离资源占用量的影响。

场景二:调整全市管控时间节点

方案B1:3月28日开始全市封控;

方案B2:3月28日浦东封控,3月30日全市封控;

方案B3:3月28日浦东封控,4月3日全市封控;

方案B4:3月28日浦东封控,4月5日全市封控;

方案B5*:3月28日浦东不封控,4月1日起开始全市封控;

真实情况:3月28日浦东封控,4月1日全市封控



6月1日起,上海市启动全市常态化核酸检测点免费检测服务, 要求进入有明确防疫要求的公共场所和搭乘公共交通工具的人员, 须持72小时内核酸检测阴性证明,重点行业人员做到一天一检。

场景三展示了自6月15号起核酸检测频率对于疫情发展趋势的影响**

场景三:6月15日起,核酸检测频率***对疫情发展的影响


自由传播下,Omicron病毒的R0约为9-12[3-4]。以自由活动下人均每日密接人数为15、 平均传播期为8天(潜伏期最后一天加传播期七天)计算,拟合模型中采用的 接触传染率为0.08。

考虑到未来可能发生的病毒变异,场景四和场景五模拟了接触感染率提高后全员核酸检测频率对于疫情发展的影响。

场景四:6月15日起,病毒发生变异,接触感染率提高至0.1,核酸检测频率**对疫情发展的影响

场景五:6月15日起,病毒发生变异,接触感染率提高至0.12,核酸检测频率对疫情发展的影响

注释

*:场景二方案5中,4月1日封控前,没有发生大量采购行为;

**:场景三至场景五中,由于模型参数在6月15日发生突变,人群中的阳性病例检出概率提高,导致6月15日新增阳性病例数量出现一个明显的高峰。

***:较高的核酸检测频率只是一种理想中的情况

参考资料

[1] 关于做好全市新一轮核酸筛查工作的通告
[2] Chang S, Pierson E, Koh P W, et al. Mobility network models of COVID-19 explain inequities and inform reopening[J]. Nature, 2021, 589(7840): 82-87.
[3] Yang Z, Zeng Z, Wang K, et al. Modified SEIR and AI prediction of the epidemics trend of COVID-19 in China under public health interventions[J]. Journal of thoracic disease, 2020, 12(3): 165.
[4] Liu Y, Rocklöv J. The effective reproductive number of the Omicron variant of SARS-CoV-2 is several times relative to Delta[J]. Journal of Travel Medicine, 2022, 29(3): taac037.
[5] Du Z, Hong H, Wang S, et al. Reproduction Number of the Omicron Variant Triples That of the Delta Variant[J]. Viruses, 2022, 14(4): 821.

About

The COVID-19 (Omicron variant) broke out in Shanghai in March, 2022. The total number of infections has surpassed 510,000 by April 24, seriously affecting social well-being and economic activities. To curb the spread of Omicron, the goverment has imposed different levels travel restriction to communities and carried out COVID testing at the population scale.

2022年3月,上海市爆发了新冠肺炎Omicron变种,截至4月24日,总感染数已超过51万,严重影响了公众社会福祉与经济活动。 为了应对Omicron,政府已经采取了不同程度的管控措施,开展了大规模核酸检测。

The Shanghai COVID-19 Reopen is a research project jointly launched by OMNILab, AI Institute, Shanghai Jiao Tong University and Baiyulan Open AI in early April, 2022.

为向公众提供更直观的疫情信息,我们启动了Shanghai COVID-19 Reopen项目,该项目由上海交通大 学人工智能研究院 大数据智能实验室OMNILab 与上海白玉兰开源开放研究院联合发起。

This project aims at collecting, sharing, and visualizing the latest data released by the Shanghai Municipal Health Commission. The visualization was inspired by the "atlas of inequality" website designed by MIT Media Lab. We also developed a large-scale Agent-Based System (ABS) to simulate the spread of Omicron in the mega city. We expect that our simulation and analysis could assist the preparation of reopening strategies in Shanghai.

项目旨在收集、分享、和可视化由 上海市卫健委 发布的最新数据。可视化界面的设计与开发借鉴了MIT Media Lab开发的"atlas of inequality"项目。 此外,我们开发了大规模的智能体模拟系统(ABS),用于仿真上海市新冠肺炎传播演化过程。 期望该项目的仿真和分析能够辅助上海市在经济重启阶段的策略制定。

Principal Investigators:

  • 许岩岩
    Yanyan Xu

    AI Institute, SJTU

  • 金耀辉
    Yaohui Jin
    Baiyulan Open AI
    AI Institute, SJTU

  • 杨小康
    Xiaokang Yang
    AI Institute, SJTU

Researchers:

  • 杜沁益
    Qinyi Du
    AI Institute, SJTU
    Model development

  • 胡钊萍
    Zhaoping Hu
    AI Institute, SJTU
    Visualization

  • 黄劭煜
    Shaoyu Huang
    AI Institute, SJTU
    Model development

  • 徐晟元
    Shengyuan Xu
    AI Institute, SJTU
    Data preparation

Consultants:

  • 高丰
    Feng Gao
    Baiyulan Open AI
    Open Data China
    Open data consultant

  • 杨力
    Li Yang
    Koguan School of Law, SJTU
    Legal consultant

Contact:

AI_SJTU@sjtu.edu.cn

Acknowledgments:

  • cuebiq

    上海白玉兰开源开放研究院 致力于推进人工智能软件框架研发与开源,以开源文化精神为导向,坚持参与融入、蓄势引领、扎根上海、 面向全球, 积极推动中文开源规则和社区建设,推动国际规则互认,在重点领域形成“算力、算法、数据、场景、合规”一体化社区, 成为国际人工智能研发与应用生态网络的关键节点。
    Baiyulan Open AI Lab is committed to promoting the R&D and openness of artificial intelligence development. Guided by the spirit of open source culture, Baiyulan adheres to the principles of "Rooted in Shanghai, Facing the international community, Activtely Participating in, being an integrated part, and be ready to lead". It actively accelerates the growth of Chinese open source community and the development of community rules, promotes the mutual recognition of international rules and standards, builds a community integrating "computing power, algorithm, data, scenario and compliance" in key priority areas , and becomes a key node of the global AI development network.

  • CARTO

    人工智能正在深刻改变社会经济发展模式。面对国家和上海市的战略部署,上海交通大学抢抓人工智能发展的重大战略机遇, 集聚校内外资源,于2018年1月18日成立上海交通大学人工智能研究院。 人工智能研究院作为上海交通大学人工智能研究实体,将加快组建实体化研究团队,成为学校人工智能科研成果转化与 对外交流的统一出口,积极组织校内研究中心建设,统筹校内人工智能领域科研。
    Artificial Intelligence is profoundly changing the pattern of social and economic development. Faced with the strategic deployment of China and Shanghai, Shanghai Jiao Tong University gathers internal and external resources and founded the Institute of Artificial Intelligence on Jan.18, 2018, seizing the major strategic opportunity for AI development. As the AI research entity of Shanghai Jiao Tong University, the Institute accelerates the building of research team, actively organizes the construction of research center and coordinates the scientific research in AI field on campus, becoming a unified export of the school's foreign exchange and transformation of scientific research achievements.

最近通报 
点击颜色隐藏POI
>14天
7-14天
1-7天
1天