Using analytics to manage the uncertainty of a global pandemic
With advanced analytics and robust data, it becomes possible to study potential scenarios, optimise resources and better understand the impact of COVID-19 at a national, regional and local level.
In any major emergency, robust data and strong analytics are critical for the purposes of understanding and planning. Imagine the example of a huge wildfire – knowledge of the type of terrain where this is occurring, the wind direction and speed, any other important weather conditions and the exact makeup of medical and firefighting services in the area means planning becomes easier. It allows the fire itself to be fought more effectively, while also enabling better preparation around issues like how to care for the injured and what evacuation processes need to be implemented ahead of the fire’s course.
Clearly, then, advanced analytics, applied to reliable data, offers the world a good chance of tracking, slowing and fighting the COVID-19 pandemic. Of course, says Sherrine Eid, epidemiologist and principal industry consultant at SAS, doing so in the early stages of the pandemic was difficult, since there was little knowledge or data available around the disease.
“Analysts were able to base forecasts on the scientific knowledge gleaned from similar viruses, such the SARS outbreak in 2012, but as more data becomes available, predictive models are improving. With the right data in the analytical model, countries are able to prepare for the impact, ensure hospitals have enough capacity and begin working on vaccines,” she explains.
“When dealing with a global pandemic like this, it is important to adopt an interdisciplinary approach – which means that while you have data scientists undertaking the analysis, you also need to bring in industry experts who understand the data – in this case, epidemiologists – along with government and legal professionals who can use this information to more effectively shape public policy.”
Eid adds that a comprehensive approach to mitigating the pandemic entails leveraging multiple sources of data, such as risk models, clinical data sharing and epidemiological modelling, to create an early warning system to help experts understand critical trends more clearly in order to plan for the future.
“While a clear understanding of the data and a deep analysis thereof is vital, the ability to visualise metrics and trends also makes the understanding of various aspects of the disease, such as spread patterns, simpler to grasp.
“Analytics can assist in the creation of a COVID-19 dashboard that offers not only broad global or national data, but allows users to dig granularly into constantly updated statistics, something that is vital, since responses – such as the distribution of medical supplies – inevitably occur at a local level.”
She says the goal is to create modelling environments that let users visually explore trusted COVID-19 data and run different virus projection scenarios, or trend analyses around issues like fatality rates, total confirmed cases, or even the impact of a particular intervention, essentially a before and after analysis.
“Moreover, the ability to obtain broader collective insights is also necessary, since this enables users to compare their country’s situation with other nations that may currently be, or have recently been, in similar positions,” she adds.
Mark Lambrecht, a director at SAS Healthcare and Global Life Sciences division, points out that one can apply epidemiological models to project infection rates and peak hospital demand, or to simulate constraints within the system based on key attributes like social distancing or mask wearing.
“All kinds of scenarios can be run to determine all kinds of impact metrics, such as average time between infection and hospitalisation, or even to look at future scenarios, such as if the country is choosing to reopen schools, when would be the time to undertake this.
“One can even drill down to the healthcare system level, to determine things like whether hospitals have enough dialysis machines or ventilators available, and what the impact will be if non-essential businesses are reopened too quickly. Ultimately, advanced analytics and robust data are all about optimising resources as best as possible to reduce and manage the strain on the healthcare system. Equally, it is about enabling scenario planning to manage the significant remaining uncertainty around the pandemic as best as possible,” he concludes.