This is a NYT article about using social media to predict outbreaks. An international team of scientists has developed an algorithm that could be predictive and get a two0week head start from Twitter, Google searches and mobility data from smartphones that would guide whether to relax or ramp up public health interventions. This is from a prepub from arXiv.org and the commentary is by the director of Machine Intelligence Lab at Boston Children's Hospital. With the caveat that the Google Flu Trends Algorithm from 2008 didn't work as expected but researchers have made adjustments to the program combining Google searches with other kinds of data. It is called a "harmonized" algorithm because it weights the data sources depending on how strongly it correlated to a coming increase in cases.
https://www.nytimes.com/2020/07/02/opinion/coronavirus-jail.html
This is a commentary from an anthropology Ph.D. candidate and medical student about the increase in the number of Covid-19 cases in jails. His research deals with the Cook County (Chicago) jails and was published in the Journal of Health Affairs (not a premier journal from the sound of the title). The study showed that "the cycle of arrests, jailings and releases was the most significant predictor of the spread of the coronavirus in Chicago and the rest of Illinois." A sixth of all of the cases in the city and state were linked to people who jailed and released from a single prison, according to data through April 19th. Nonviolent offenses account for 95% of the more than 10 million arrests in America every year. The conclusion from this data is that standard policing and incarceration policies are driving preventable spread of the virus. People are being released from prison daily with the belief they are negative when in reality the virus has begun to spread in their bodies. The authors suggest that policymakers should stop the daily cycling of thousands of people in and out of jails. During the pandemic issues of criminal justice reform should not be determined by partisan politics.
https://www.nytimes.com/2020/06/30/upshot/coronavirus-economists-dexamethasone.html
This is the Upshot and a discussion of the Recovery trial, a randomized controlled trial in Britain. It showed that natural experiments, such as when a shortage of one particular drug that was used to treat sepsis gave scientists the unique ability to assess the effects of the drug on mortality rates. Apparently, natural experiments allow a more causal interpretation of data, while associational studies, typical in medicine, do not. This points out the sometimes-abritary clinical cutoffs made to allocate particular treatments. For example, a hospital decides whether to provide a ventilator to patients based on whether they use 6 liters of oxygen per minute, but actually a 5-liter-per minute patient is pretty much the same. The commentary suggests that these questions must be asked and thresholds must be known and interrogated. Differences in treatment around the threshold must be examined to see whether they correspond to differences in clinical outcomes. In terms of the Recovery trial, after the results were announced, the data for critically ill Covid-19 patients were examined, with the expectation that patients hospitalized before the Recovery trail results were announced would fare worse than those hospitalized in the days afterward, assuming that doctors started using DEXA.
https://www.nytimes.com/2020/07/14/us/politics/trump-cdc-coronavirus.html
Article in NYT detailing the protracted rape of the CDC by the administration. So the HHS is now being tasked with collecting daily reports about the patients that each hospital is treating, number of available beds/ventilators, and other information vital to tracking the pandemic. S'posed to make it easier for decisions about allocating short-supply items like remdesivir, but in all likelihood this new policy will make it hard to access new information not open to the public, potentially affecting the work of scores of researchers, modelers and health officials who rely on data formerly sent to the CDC to make projections and crucial decisions. Just another "trump" in the road. Will this also be true of the Weather Bureau, the US Geological Service, NASA, FDA? Will the federal government be hiring new data "remodelers" trained to suppress information? What will be their civil service grades?
https://www.statnews.com/2020/07/16/hospital-data-reporting-covid-19/
This is from StatNews about an alarming new change in the way real-time Covid-19 data is reported. Accurate and timely reporting of data informs stop-on-a-dime policy changes. Covid-19 data,rather than being reported to CDC and HHS is now only being reported to the HHS, leading to concerns that the data would be eliminated, manipulated, falsified or all three. The rationale is data consolidation in order to make decisions about allocation of scarce resources, such as remdesivir. “In the midst of the worst public health crisis in a century, it is counter-productive to create a new mechanism which will be extremely complicated to build and implement,” public health officials, including former CDC director Tom Frieden, wrote. The data is openly available on the CDC’s website and allows some of the most consulted Covid-19 tracking sites, like covidexitstrategy.org, to provide numbers on how many hospital beds hospitals are open in different states.
No comments:
Post a Comment