最近,一篇号称武汉从去年秋天就开始出现新冠疫情的论文出现在网上,而且还出自哈佛大学论坛,一时间引发外媒纷纷报道。只不过,这篇所谓的论文研究却漏洞百出,粗制滥造!
6月8日,美国哈佛大学的学术社区平台“DASH”上出现了这样一篇论文,标题为:通过对中国武汉医院交通和搜索引擎的数据分析,显示2019年秋天就出现早期疾病活动。
尽管标题中没有新冠字样,但是其摘要就很直白了,是这么写的:
“全球新冠肺炎疫情的源头与去年11 -12月武汉华南海鲜市场有关联。然而,最新证据显示,病毒在暴发前就已经在传播。我们通过已经确认过的数据流——医院停车场的卫星图和百度搜索相关疾病词汇——调查这一论断的可能性。我们发现,从2019年夏末秋初开始,医院交通车流量和相关疾病词汇搜索量都有增长。”
"The global COVID-19 pandemic was originally linked to a zoonotic spillover event in Wuhan’s Huanan Seafood Market in November or December of 2019. However, recent evidence suggests that the virus may have already been circulating at the time of the outbreak. Here we use previously validated data streams - satellite imagery of hospital parking lots and Baidu search queries of disease related terms - to investigate this possibility. We observe an upward trend in hospital traffic and search volume beginning in late Summer and early Fall 2019. "
那么他们选择的是什么样的词汇搜索呢?摘要里也写了:咳嗽(cough)和腹泻(diarrhea)。论文认为,这两样都是新冠肺炎的症状。
看完是不是有点蒙圈儿?停车场能和新冠有关系?百度搜索咳嗽和腹泻就说明是新冠了?论文里是这样“操作”的↓↓↓
这是论文里给出的图表数据:地点是武汉天佑医院(上排)和华南海鲜市场(下排)。其中红点是普通汽车,黄点是卡车。时间是2018年的10月,2019年的10月、12月,以及2020年2月。
另一张图表则是6家武汉医院停车场车辆数量变化的曲线图,以及百度搜索引擎检索相关关键词数量的曲线图等。
ABC 新闻网对这一论文做了详细的独家报道,报道中的有些细节不容忽视。
领导这篇论文研究的负责人Dr. John Brownstein,既是哈佛医学教授(Harvard Medical professor),同时也是该新闻网的供稿人(contributor)。而且,报道中的内容远比哈佛论坛里论文的内容详细多了。
比如在ABC新闻网的报道中,有武汉中南医院、同济医院、天佑医院和湖北妇幼保健院停车场车辆数在2018年10月和2019年10月变化的图片:
据ABC新闻网报道,通过一番对比分析,论文认为:医院停车场车辆增多,结论是病人明显增多;百度在去年某个时段关键词汇搜索增多,是因为大家都在那个时候得病了;同时,两个数据增长的时段一致!
然后论文给出结论,这和武汉早在去年8月就有新冠是有关联性的!
尽管这篇论文没有经过同行评议(peer-viewed),却早已成为各大外媒关注的对象,尤其是哈佛大学这块“金字招牌”特别引人瞩目↓↓↓
CNBC就在标题里突出了论文的“哈佛”属性:
哈佛研究发现:卫星图片和在线搜索显示中国在秋季已有新冠病毒
福布斯网站更简约:
哈佛研究显示:新冠病毒可能于去年8月已在中国传播
欧洲新闻网如出一辙:
哈佛研究人员表示,卫星图片显示新冠疫情于去年秋季开始,而非冬季
福克斯新闻更加详细:
哈佛研究使用卫星交通图像显示,新冠疫情比想象中更早出现在武汉
显然,这样的报道,就很容易成为把新冠疫情“甩锅”中国的利器,美国总统特朗普也在推特上转发福克斯新闻对这一事件的报道。
特朗普的儿子也一样,只不过转发的是ABC新闻网的内容:
不过,这样的证据确实可以得出论文的结论吗?《环球时报》英文版的一篇报道指出,论文提供的照片并不能说明问题。卫星专家在分析2018和2019年照片时认为,照片的拍摄选取角度并不相同(黄圈中显示)。以中南医院为例,2018年照片中的车辆看上去较少,因为拍摄角度有倾斜,车辆会被建筑物阴影遮挡;然而2019年照片更近似于垂直角度,所以更能看清车辆停放。
Analyzing satellite pictures taken in 2018 and 2019, satellite experts told the Global Times that some of those pictures were not taken from the same angle. Take the ones taken at Zhongnan Hospital as an example. Fewer vehicles can be seen on the 2018 photo because it was obviously taken from an angle of inclination. Thus, many cars were blocked by buildings and couldn't be seen in this picture, while the 2019 picture was taken from a vertical angle, which gives viewers a clearer view of vehicles.
其他两家医院的图片也都有拍摄角度上的不同:
同时,报道表示,图片中的车辆数字变化仅在几百辆左右,以至于有网友质疑对于武汉这样一座拥有1000万居民的大城市而言,这样的数字变化是否真的说明问题。
Comparing those satellite images, the number difference in the 2018 and 2019 pictures is only a few hundred. Chinese netizens questioned if a difference of a few hundred vehicles meant anything in a city with 10 million residents.
另外,报道援引武汉有关部门官方数据显示,2019年3月武汉拥有机动车320万辆,比2018年增加27.8万辆。
Wuhan authorities said the number of motor vehicles in the city was 3.2 million as of March 2019, an increase of 278,000 compared to that in 2018.
报道还发现,ABC新闻网在援引几张数据图时,最后的时期标注为2019年5月,不清楚是否是无意中标错了,还是有意为之,另有目的。
至于在论文中被引用的词汇搜索,百度方面10日也发布声明:这是非常牵强和不严谨的。同时,百度给出了时间段更长和更详细的对比数据,百度指数显示,2019年12月左右“腹泻“搜索量还有轻微的下降。百度同时呼吁相关研究需要更加科学严谨的引用百度数据。
BBC驻京记者John Sudworth则认为,研究人员使用的数据很有局限性:比如说,每年的同一天拍摄卫星图片是不可能的,因为云层在某些照片中会有遮挡。
The BBC's John Sudworth in Beijing says there were limits on the data set used by researchers - for example, they could not always compare satellite images taken on the same day in consecutive years due to cloud cover in some of the photos.
有科学家也表示,对于这篇论文的结论持怀疑态度。英国诺丁汉大学传染病流行病学教授Keith Neal在接受路透社采访时表示,论文的车流量数据包括至少一家儿童医院。儿童确实会患流感生病,但是并非会得新冠肺炎。
Keith Neal, a professor of the epidemiology of infectious diseases at Britain’s Nottingham University, said the study included traffic around at least one children’s hospital and that while children do get ill with flu, they do not tend to get sick with COVID-19.
来自美国斯克里普研究院的Eric Topol博士则表示该论文的研究方法并不可行,“很间接,很不准确”。Topol博士没有参与该项研究,他表示根据目前他掌握的证据,他对新冠在8月份暴发的证据表示怀疑。他和其他人员的基因研究指出,病毒可能在去年秋季的某一时段,从动物宿主转到了人类身上。
Dr. Eric Topol, director of the Scripps Research Translational Institute, said the research method is not validated and is “very indirect and imprecise.” Topol, who was not involved with the research, said he doubts the outbreak began in August, based on the evidence he has seen so far. He and others pointed to genetic evidence suggesting the virus made the leap from animal host to humans some time in the fall.
6月10日,世卫组织卫生紧急项目负责人迈克尔瑞安表示,现在越来越多研究使用卫星图像,如跟踪气候变化和人口移动,但重要的是不要过多推测。
图源:中报视频
瑞安说,“重点是,我们不要对停车场车辆数量的意义进行过多猜测,然后一下子跳跃两三步就匆忙得出结论。因为根本就没有证据表明那是确实发生过的。但是,我们乐意和该研究团队跟进,并评估他们的研究方式以及到底产生了什么影响。”
“It’s really important that we don’t speculate too much regarding what the implications of cars in the car parking lot and then make a jump, 2 or 3 steps forward into what it represents. Because it there’s no evidence per se that’s what was suppposed to that auctually happen. But we’ll be very happy to follow up with the team that have done this work and look at how they did their study what the implications are.”
视频来源:中报视频
外交部发言人华春莹于11日在外交部例行记者会上做出回应。她表示觉得特别诧异(surprised):“本能的反应是怎么能通过停车场汽车数量来推断新冠肺炎疫情发生的时间呢?我觉得这种联想非常不可思议(ridiculous)。”
华春莹表示,论文除了拿停车场的汽车数量来推断新冠肺炎可能最初发生的时间之外,还有几个明显和低级的漏洞:
图源:外交部网站
哈佛大学的学术平台,只是一个开放型的、搜集和保存教研人员一些研究资料的资料库,不是有同行评议的刊物,这个论文到底代表了哈佛大学医学院的正式观点和它的水准,还是只是个人的或者几个人的运作?
First, DASH is Harvard's open-access repository to collect, save and publish research by members of the Harvard community, not an academic journal that requires rigorous peer review. Does the said study truly reflect Harvard's viewpoint and standard? I'm afraid there's a big question mark over it.
论文的作者之一,恰巧是独家报道这篇论文的美国广播公司的撰稿人,而且恰巧在这篇论文甚至还没有预发布之前就已经拿到了更多的数据并且进行了报道。
Second, one of the authors of the study "happens" to be a contributor of ABC News that exclusively reported the study, and ABC News just "happened" to have covered it with a lot more data even before the study was pre-released.
论文当中的一些图表上标注的时间居然是2019年5月份,不知道这是无心的疏忽,还是有什么别的问题?
Third, a diagram in the ABC News report was marked "May 2019". I'm not sure if it's an unintentional mistake or due to some other reason.
论文认为是关键证据之一的、比如说咳嗽腹泻的检索量,我注意到中方有些媒体也做了一些深度的调研,就发现论文当中引述的2019年9月份对咳嗽和腹泻两个关键词的检索量,还不如2017年和2018年同期。
Fourth, regarding the search data of "cough" and "diarrhea", which is considered key evidence in the study, I noticed that some Chinese media, after in-depth research, found that the increase in "cough" and "diarrhea" search data cited by the study is actually less than that over the same period in 2017 and 2018.
华春莹表示:“事实上,将这么严肃的科学问题进行如此不严肃的处理,实在是让人感到奇怪。这么一个漏洞百出、粗制滥造的所谓论文,却让美方一些政客和媒体如获至宝,大肆传播。”
In fact, it is odd that such a serious scientific issue should be taken so lightly. However, this so-called research, full of loopholes and shoddy work, is widely circulated by some US politicians and media as new "evidence" of China's concealment of the epidemic.
“这种非常可笑的现象背后,他的用意恐怕并不那么可笑。”
I'm afraid behind this ridiculous phenomenon is not that a ridiculous motive at all.
综合来源:路透社,CNN,Abc News,观察者网,环球时报英文版网站,外交部网站,央视新闻