How Data Science Is Helping To Combat Coronavirus Pandemic
The rapid expansion and global effect of COVID-19 will make people feel uncertain and fearful as the novel coronavirus continues to escalate and causes them to alter many aspects of their daily lives. The increasingly rapid spread of COVID-19 has put out innovative tools for big data analytics, with institutions from all healthcare industry fields looking to track and mitigate the impact of this virus. Social distance and home-stay restrictions in the United States have reduced the SARS-CoV-2 infection rate, the pathogen that triggers COVID-19. This stopped the imminent threat to the U.S. healthcare system, but the consensus on a long-term strategy or solution to the issue remains elusive. Given that the fact is that there have been no easy solutions and that medications and vaccinations will take countless months, if not years, to discover, test, and mass-produce drugs. Now is the best time to think about another topic: how will data science and technology support us with the pandemic when creating treatments and vaccinations?
Before policymakers re-open their markets, they should be assured that the corresponding feature. COVID-19 cases would not cause local healthcare systems to return to crisis care levels. This includes mitigation and elimination of the virus, continuous measurement of the viral infection activity, evaluation of the effectiveness of suppression measures, and prediction of near-term demand for local healthcare systems. Considering the population's demographics, the incidence of pre-existing conditions and population size, and socio-economics, this demand is enormously variable.
Data science could already offer comprehensive, reliable estimates of health system demand, which would be needed in almost all re-opening plans. We have to step beyond that to a flexible approach to research, interpretation, and prediction to inform real-time policy decisions and refine public health guidelines for re-opening recursively. While most re-opening proposals suggest comprehensive research, contact tracking, and community mobility monitoring, hardly any of them consider developing such a complex feedback loop. Such feedback may be used to assess the degree of virus activity expected in the area provided the regional health system capability and modify population distances accordingly.
For data science to progress, first and foremost, we have to address limitations in data collection and accessibility caused by current reporting processes. Most public health agencies currently compile and monitor metrics that are not beneficial and report 48-hour delays and sometimes irregularities. While there are instances of regional success in such reporting, the healthcare IT community's guidelines for reliable and timely public health reporting remain widely ignored. Consider the number of COVID-19 hospitalized patients, which is the most substantial morbidity and mortality rate on the regional health system. At present, compared to limited time lags in verifying and documenting cases and the inability to differentiate between current and accumulated hospitalizations, even regions reporting hospitalization data also provide a distorted image of the burden on the local healthcare system. Ideally, regions must record all reported and proven hospital cases and specify the date of admission in relation to the date of the report or verification.
Although with perfect reporting, there are significant uncertainties in what such data will teach us. For example, today's new hospitalization rates indicate virus behavior from 9 to 13 days ago (which depends, in turn, on social distancing interventions from up to 17 days prior). Failure to figure in such factors has led to substantial over-estimation of hospitalization requirements across the world. Therefore, we have to calculate the behavior of the virus by potential predictors that are predictive development life cycle of the virus. These should be evaluated against the growth of emerging and complete COVID-19 hospitalizations and, preferably, against the number of new infections, ensuring that they are correctly measured by large-scale monitoring.
Accessible proxy measures comprise test positivity ratings in health systems, case statistics, fatalities, and likely seroprevalence rates. Ongoing symptom monitoring through mobile apps, regular web or phone surveys, or cough sounds may spot potential zones with high transmission rates for viruses. Contact tracing, which needs tremendous human effort, can also help monitor possible cases if it can be successfully applied utilizing technology under development by major American technology firms.
With accurate monitoring and performance analysis in place, we can quantify the incidence of infection and the steady growth and transmission rates, which are essential for evaluating whether policies are working. This is an issue not only of data gathering and also of data interpretation. Sensitivity problems, frequent variability, time lags, and uncertainty need to be analyzed before these data could be used consistently. For example, symptom monitoring is non-specific and may have trouble spotting the virus's activity at low prevalence. Other evolving data sources, including wastewater and smart thermometer data, carry similar commitments and deal with the same problems.
It is then essential to analyze the regional effects of policy measures, like shelter-in-place instructions (through mobility reduction) and touch tracking (through reductions in cases reported), first as necessary projections and, finally, by evolving to what-if analyses. Several attempts have measured mobility's effect on the propagation of viruses, and some have proposed "safe" types of mobility. While there are several possible ways to estimate population mobility—such as through traffic patterns, Internet bandwidth used by address, and card swiper location—the most efficient method for calculating mobility appears to be through smartphone monitoring. Groups, including the COVID-19 Mobility Data Network, only provide data regularly in unredacted, distributed reports.
Implementing such technologies and data science to maintain the healthcare needs expected far below the accessibility threshold in an area involves several trade-offs on confidentiality, which would require thoughtful regulation so that strategies invented to last the current pandemic will not contribute to a permanent loss of privacy. However, given the enormous economic and secret medical cost of the shutdown, we rapidly need to develop a warning system that informs us to improve containment measures if the next peak of the COVID-19 breakout is likely to overtake our regional healthcare system. We must concentrate our efforts on using data science to predict and control local health system resource requirements depending on specific virus activity measures and community distancing effects.
If you found this Article interesting, why not review the other Articles in our archive.