A Journey in Approachable Data Science

It’s been about a month since the entire world was forced to come to an abrupt stop. In Canada we’d say, the global community looked like “Canadian deer in headlights.”  The synchronized lockdowns around the world slowed down propagation velocity and saved thousands of lives.

deer

These lockdowns have allowed everyone to “buy” precious time. Time to tame the curve, of course, but as importantly, time to plan for the complex execution to protect those most at risks when we start relaxing isolation rules. There is no “blueprint” to follow, we must use our best judgment and making the wrong call would come at a hefty price.

This paper looks at the key public policy options to minimize avoidable casualties and other adverse impacts until a vaccine becomes part of the equation.

This paper also attempts to get a little closer to the truth on some of the metrics on which we will base the most important decisions moving forward: proper estimation of the risk of death by age profile / health condition as well as developing a capacity to estimate accurately Sars-coV-2 infection levels in the population.  

Hindsight is 20/20 and one day we’ll know for sure what those numbers were. In the meantime, we need to get our estimates as close as we need them to be to make these high-stake decisions – such as identifying the proper timing to shift from “generalized” to “targeted” lockdowns.

The fact that we don’t have a count of infected individuals – the “true case number” – is not an excuse for using the wrong datasets for making these go/no-go decisions. We can derive “directionally correct” estimates by bridging data gaps with proper evidence triangulation. It’s not easy and it’s not perfect but it’s significantly more reliable than relying on “Case” and “CFR” data to make these decisions.

During the initial lockdowns, tracking casualties & hospitalization intake were key measures to get a grip on the situation. Key measures moving forward will shift to IFR risk management, flare-up advanced detection and propagation tracking.

This analysis covers:
●. A comparison of Case Fatality Rate (CFR) and Infection Fatality Rate (IFR)
and the use and limitations of these metrics; 
●. Propagation dynamics;
●. Methodology to Improve the risk of death estimation accuracy by age
group and health profile;
●. Methodology to improve infection propagation estimation accuracy;
●. Key considerations for Public Policy as the world embark on the long wait
for the availability of a vaccine.  

You can download the report here: https://www.outhiink.com.com/covid-19-special-report or via this linkedin post: https://www.linkedin.com/pulse/covid-19-data-could-speak-perspectives-inform-public-policy-de-bane

Hope you find value in reading Outhiink’s take on Covid-19 and as always, don’t hesitate to get back to us if you have any comments or questions. If you feel like it, you can read the prologue – totally facultative –  which talks about how this “special edition” analysis came to be.

PROLOGUE

At Outhiink we’re in the business of finding solutions to complex problems via “Forward Thinking”; our methodology which guides all our assignments.

Let’s be clear, at Outhiink we’re not epidemiology experts, our area of expertise is complex problem framing and creative solution space analysis.

We think we get to a better place when we bring together domain experts and problem framing experts to address our biggest challenges such as defining the future of Healthcare, education, or how to respond to crises such as this one. Only relying on domain experts tends to limit the solution space to past approaches and makes it difficult to explore new ways of framing the problems in an attempt to expand the solution space.

Enough self-advertising, back to the story. A week ago, one of my children called me and said: “Dad, people don’t understand what’s going on with Covid-19. You should analyze this and make a video.”  Thank you my son for considering your dad a “good figurer” but YouTube is not my thing so I offered him this challenge instead:

I will invest three days to perform a quick dive and feed him the result of my analysis so he can prepare an explainer video. We had a virtual handshake deal. The deal was a no brainer for me. What I had planned to do in April just went through the hopper and I was looking for something useful to do. Clearly, there is nothing more important right now than finding ways to help in shaping our way out of this. On a more personal level, for many years I kept telling my children that “thinking on your feet” can be learned and is a great life skill. Delivering such an analysis in three days is a way to demonstrate to them that it can be done.

I did deliver my analysis in 3 days flat, 3 heavy 18hr day: 
● ½ day to frame the issues and develop working hypotheses / questions to
be tested;
● ½ day to build the model and preliminary storyline;
● 1 day to identify & validate relevant data;
●1 day to make sense of the information, draw implications, populate the
model and finalize the story line / key takeaways;
● I’ve added 2 days of wordsmithing and slide layout tweaks as I am not
holding my breath for my son to deliver the video that triggered all this.

Three days is very short to conduct such an analysis; five days would have been a little less hectic. This being said, I am quite confident that the essence of my analysis, key takeaways and conclusions would have remained the same whether I spent another week or more evaluating the data available as of April 8th. Stated differently, at this point in time, these were the most reasonable conclusions I could reach and adding more time would not have changed that.

Now back to my son video story. The video may eventually be produced or it may not, but at least my son knows that it is possible to conduct quickly a pretty decent analysis if you know what to look for and how to properly assess what you find.

Now back to my son’s video story. The video may eventually be produced or it may not, but at least my son knows that it is possible to conduct a pretty decent analysis quickly if you know what to look for and how to properly assess what you find. To nudge him a little to deliver his part of the deal, I created this 30-second Coronavirus “explainer clip” and told him I was going to put his name on it if he doesn’t deliver his part of the deal. I would never do that, of course, but I hope the tactic works.  If you really want to see it here it is; https://youtu.be/9ZSGUWjzJs4

You can download the report here: https://www.outhiink.com.com/covid-19-special-report or via this linkedin post https://www.linkedin.com/pulse/covid-19-data-could-speak-perspectives-inform-public-policy-de-bane

Leave a Reply

Your email address will not be published. Required fields are marked *