Visual analytics can help to combat COVID-19
The best way to understand the disease and to forecast future impacts is through the implementation of visual analytics, allowing scientists to visualise trends, infection rates and ‘what if’ scenarios.
As the world struggles to deal with the global health challenge created by COVID-19, it has become clear that we need a new, multi-angled approach to tackling this disease. Analytics techniques like optimisation, forecasting and epidemiological modelling can play a big role in helping public health and life sciences companies fight the pandemic.
With viral outbreaks like the coronavirus, robust information is absolutely vital in order to foster accurate decision-making. As Mark Lambrecht, director of the Global Health and Life Science Practice at SAS, points out, when the crisis began, the world lacked knowledge of the natural history of this disease, apart from the existing understanding it had of other, similar viruses like SARS.
To more clearly understand, track and analyse the disease, he says it’s important to clearly understand everything from foundational parameters to trends, mortality rates and overall case counts, in order to get an idea of the velocity of its global spread.
“In analytics, parameters are always important and are particularly crucial here, if we are to effectively understand whether measures like lockdown are working. To this end, it’s necessary to look at areas like peak incidences, peak infections and inflection points. In addition, a good analytics tool should also allow you to run ‘what if’ scenarios, enabling the user to compare multiple possible models, in order to provide more accurate forecasts and future planning,” he says.
“Perhaps most crucially, the analytics tool needs to be able to provide visualisation of the data – this is particularly critical when tracking a disease that is highly infectious and globally dispersed. Visualisation makes it easier to not only understand what is happening in the present, but more vitally, also forces you to ask the right questions for the future. For example, a visual representation of how many people have tested positive and how many are dying can be overlaid with a map breaking this down by aspects like age, the amount of testing being conducted, or the regions where this is occurring. Such an approach will give a much more accurate and understandable reading of the situation, especially compared to what could be understood from mere figures on a page.”
Lambrecht adds that this type of visualisation can then be combined with other details, such as where COVID-19 incidences are highest, in order to more clearly understand which areas need the most help, and how things like schooling will be impacted in this region. Similarly, visualising fatality rates will enable scientists to improve their testing strategies as well as understand the disease’s impact on different age groups.
“Visualisation tools are also beneficial to healthcare providers offering critical care, as they can use advanced epidemiological techniques to model and predict peak ICU bed occupation rates and how these may change in a day, week or month from now, as well as understanding more clearly how many nursing staff will be needed and what equipment will be required for them to remain operational.
“Of course, analytics tools are at their best when having access to as much data as possible, so you really want insight into biological and healthcare data, epidemiologists' work, data from public health specialists, and any relevant publicly available data. We are currently looking at how something like text analytics can play a role in studying doctors’ notes and various scientific literature to draw additional insights from these.”
To this end, he adds, SAS has made several data discovery environments free to access on their Web site in order to assist other healthcare organisations as well as governments that either do not have a dedicated data science team, or whose teams want to compare their models to the SAS one.
“This is, after all, a global challenge that no single entity can solve, which is why collaboration is so vital. While there are multitudes of efforts and initiatives, what is needed is to harmonise the best of these while also putting in place some type of standards – at present, harmonising data is difficult because different countries and regions have different methodologies for testing, data entry, case counting and so on,” he points out.
Nonetheless, continues Lambrecht, the coronavirus challenge presents a great opportunity to strengthen global healthcare systems, to better understand at-risk populations and to ensure far greater levels of preparedness for the next such crisis.
“Pandemics have happened before and will happen again, so the one key lesson to draw from this is that in a world beset by radical uncertainties, we need the capacity to control this uncertainty as far as possible. Analytics is a crucially important factor to control this crisis, as it offers the means to ensure we deploy the right technology, at the right stage of a crisis, to solve the right problem at the right time,” he concludes.