After were speaking about only a handful of infected patients a few weeks ago, we’ve now reached that stage where the most faithful representation of the epidemics is the number of casualties. Flattening that cursed curve isn’t as easy as it seems, and a massive global effort is underway, under the radar and unknown to many.

From now on, it’s very likely that you (will) know directly someone who is actually infected, quarantined, hospitalized… Despite the reassuring guidance from governments, we can’t help thinking that something is wrong: not enough tests, not enough masks, not enough beds, not enough respirators. And worst of all, hospital staff at breaking point. Nevertheless, our feeling is that: nobody is to blame – let’s face it, who could have been ready for that? But still, what are we really supposed to do if one really needs medical assistance in the coming days?

Global coordination in such an overloaded and blurry context is one of the most complex problems to tackle – ever. And, beyond the outstanding effort of clinicians and (para)medical staff, most countries lack a global vision which would rely on faithful and updated data. Most hospitals keep reporting their available beds by lengthy report emails, WhatsApp groups or phone calls which, needless to say, are time wasters and highly suboptimal. It is truly astonishing to see countries like France or Switzerland being so reluctant to provide an open live data hub for the Covid-19 pandemic. While some nice repositories with manually input or scraped data are emerging, they are based on the (hopefully not too tiny) fraction of reported cases. If you are a German speaker take a look at this article otherwise, we translate its title for you “We have taken the number of deaths from Wikipedia”. In a time of so-called high tech, this is astounding and shocking. We could do better, much better.

Healthcare was late in the data-driven decision landscape and particularly in Europe, but it is now widely taking off. We do believe that predictive and prescriptive tools could be scaled up to help cross-hospital as well as intra- and inter-country coordination.

Coordination vs. COVID-19

The past week, we’ve been scratching our heads about creating a territorial coordination tool and we came up with the following baseline prototype by reconciliation of different data sources. It is applied here to Switzerland:

Switzerland and the Coronavirus Crisis (COVID-19) as on 24.03.2020

In the above interactive map which shows the boundaries Switzerland and it’s “Cantons” (regions), we display 3 key indicators which you can hover on:

1. The Blue Circles are Healthcare facilities.
Their sizes scale with the number of beds reported in 2017 here.

2. The Red Circles (as you already guessed) illustrate the total reported Covid-19 cases per Canton centered on the corresponding “capital” (main city).
Their sizes, like in the John Hopkins map is sizable with the number of cases. e.g. “Vaud” is today the most critical spot.

3. The Yellow Lines are the central point of the tool.
We wrote an optimization algorithm to find the most relevant dispatching hospital for a given infection center depending on proximity, number of beds and number of new cases.

This basic tool gives an immediate interactive vision of the territorial configuration and allows to make quick decisions, especially in the early or late stages of epidemics. If fed by live data on bed capacity and cases with a higher granularity it would allow for government to take quick, global and targeted action in the service of a better territorial coordination. We see that for example Canton Zurich shows a compact aggregation of Hospitals while this is much less the case e.g. for Vaud.

Our tool also allows displaying forecasts for the next few days as shown below. The map depicts the situation to expect, 3 days ahead of today, namely on the 27th of March 2020 and the complexity that might entail in the most affected areas.

The same Switzerland facing the Coronavirus crisis (COVID-19) on 27.03.2020

Of course, we welcome any comments on the accuracy of the map data we represent. Indeed, since they come from several data sets, small errors can never be excluded. So do not hesitate to write to us if something is not right.

Anticipating the impact of COVID-19

Team CALYPS