Additional menu

valueSTK

Data science lessons learned from experiencing COVID-19

by Filed Under: Also Published On..., Measurement

NOT MEDICAL ADVICE – personal lessons reflecting on what it tells us about business data science

Some people would ask, as a data scientist are you upset that the data actually failed you. Not really. Because I know in data science gaining insights from data is a combination of what the data tells you as well as a business person’s intuition and experience. It is a failure to rely too heavily on one or the other. You need to question everything. The data as well as our cognitive biases.

This spring I had a “not medically confirmed” case of COVID-19.

Meaning I can tell you anecdotally I had it. Doctors have told me to presume I did. But my test result was negative. I learned a lot from my experience, including insights that we can use in our businesses.

If you are more interested in the business lessons learned than my story, feel free jump to the lesson that’s relevant for you.

Otherwise, thank you for listening to my story first!

My anecdotal experience with the virus

I received official notice from the public health authority on March 21st, that someone who was at an event I attended March 12th had tested positive for COVID-19. I was already feeling off. No energy to go for a walk, cold as in chilled, scratchy throat. So, I take my “start” back to March 19th, which is typical, a week after exposure. The symptoms got progressively worse the next few days and I went for a test March 24th.

Crazy thing is I’m the person that doesn’t get sick. In yoga Ayurveda terms I describe myself as having a pitta dosha – a strong sporty constitution, fiery, and fast. I speculate I am more likely immune-tolerant, than having a strong immune system. Things just don’t floor me. THIS DID! But, from some sources, that characteristic may have been what saved me from it escalating to a severe case. Some severe cases have been an overactive immune response in the second week.

I was worried about at-risk people in my life and thank god had already social distanced from them before the pandemic was declared. I’m also thankful that the event on March 12th was my last, as I just stopped doing things after that. I went to less than a handful of stores the following days, but worried about the virus I wore gloves and kept a distance from people. Hopefully, I didn’t share it.

Getting tested was a nightmare. It was early days in the pandemic, processes were constantly changing, information to the public was unclear, and they were only testing people who traveled or had a known contact and had symptoms. I spent forever on hold with public health to be told to go to ER to be cleared for testing. Drive yourself. Wear a scarf. (I had no masks.)

My symptoms

That started my 3 week dance with the virus. I had distinct differences over each week of the 3 weeks. And a continuous up and down. Now in recovery, I continue to improve with the odd swing back slightly symptomatic days. Each day is better than the last (except for the swing backs). The strange thing about the virus is that it tracks differently for different people. There are some commonalities. Especially in terms of the fever, the lengthiness of the illness and the long time to recover. Here were my symptoms.

  • Fever
    • A 7-10 period of high fever and chills, always peaking at night. Disturbed fitful sleep with lots of dreams. Fully waking between 2-4am and feeling clammy.
    • Fever as a symptom lasted 3 weeks, finally noticeably breaking on April 11. At which point I was physically and emotionally exhausted.
    • The fever would go up and down, changing in intensity. Often impacted by activity – with activity being getting off the couch to try some work at my desk.
    • Throughout the 3 weeks of fever, I averaged 3-4 hours of sleep a night. Since the fever broke, I’ve returned to my usual 6.5-7.5 hours.
  • Low Blood oxygen
    • Shortness of breath, no energy, poor focus and concentration. Headaches. Summed up as the brain fog.
    • It seemed more noticeable as a remaining symptom the second week. The first week had so many symptoms, I may not have noticed as much. Though this second week seems to be the point where some people are admitted to hospital and put on oxygen, because blood-oxygen has fallen so low.
    • I don’t have an oxygen meter so don’t have exact figures. My Mom uses oxygen because she has lung scaring. I recognized her symptoms in myself. I recommend a meter should become a household item in your medicine cabinet since this virus will be with us a while. If you can get one!
    • It also seemed up and down, rather than a constant low. Now in week 5, I am still having headaches. My concentration is not yet what is usual for me. But getting better.
  • No taste, no smell and no appetite in the first 7-10 day period. I lost 7lbs. (Though now have found some of them). Taste slowly came back.
  • Every bodily function not my usual. My body feels like it has been through a major trauma.
  • A strange loss of time, in that I don’t really know how long it really all was that it happened. A symptom of psychological trauma.
  • Periods of time when I looked completely well with no visible symptoms. Especially mornings. It’s when I often talked to people – phone and video chat – who often said I didn’t seem ill.

The medical assessment

The test came back negative. After a week of waiting, I contacted my public health authority. It was about the time that here in Ontario they decided to clear up the backlog – contacting only people with positive test results and sending negative results by letter. I’ve never received a letter. I did contact my doctor’s office, who was able to pull down the result from the hospital because it was done in ER. It was negative.

So, just as I was left alone to recover at home, I was left to decide when it was finished. Positive test people must undergo 2 negative tests to be cleared. The public health site advice to people who have negative results is to remain self-isolating for 48 hours after symptoms end. An oxymoron if I ever heard one! But their catch-all for the false negatives.

At the start of the fourth week I had a virtual appointment with my doctor. A “report in” and to ask what I was supposed to do to consider myself cleared.

She said that testing isn’t perfect, it can be the sample wasn’t taken well, the test inconclusive or lots of other reasons why a false negative could happen. They would have only tested me again if I had ended up having to return to hospital with worsening symptoms. But not if I just had symptoms at home. I should expect to continue having some symptoms for some time, even though I’m now recovering, as that’s the nature of viruses. She said to consider still having fevers as a fever above 38. At this point I shouldn’t have a fever, so if I do than return to ER. Take things slowly… as in if you go for a walk, walk 10 minutes initially. They haven’t confirmed that people gain immunity from having it, so continue to protect yourself with social distancing and hand washing. Wear a mask if/when you go out.

So, now I’m in the same situation as everyone else. Just worn out and not yet back to my usual self.

As a data scientist, the lessons I learned….

Predictive Analytics can be more informative than incomplete current data

For this virus, all over the world, the number of cases reported is an inaccurate and incomplete number. The need for testing overwhelmed the usual public health authorities. (Note, I talk about capacity below, but a big part of the reasons is a lack of capacity).

Partly this is just the nature of pandemics. They swing up fast and furious. It’s hard to get a handle on what is happening. Partly, this is because of the nature of this particular virus, where some people have a severe case, some people the at-home version I had, some people no symptoms at all. Making it hard for public health to decide who was to be tested an do adequate contact tracing.

Authorities put conditions on who was to be tested, which means it wasn’t a comprehensive sampling. The tests themselves have been flawed, with some people speculating about 20% false negatives. The test isn’t really that flaky. Since they weren’t doing a random sampling, only testing people who probably had the virus, then there was probably a higher instance of false negatives in the people they tested. The point is a positive test result means you positively have it. A negative test result doesn’t necessarily mean you don’t have it.

As a result, number of cases is actually not a reliable number to look at. The real number to track is number of deaths. Grim as it is, it is the most conclusive statistic.

The predictive models, though by nature always incorrect, are more reliable at predicting the number of deaths than extrapolating from just number of cases. They take in a variety of factors. Including how the virus performed in other regions. How behaviours like social distancing may impact it. Prediction versus recent actual. This is what we should be focusing on, not obsessing on current number of cases.

There are lessons for our own data science projects, when we want to predict much more positive outcomes of revenue, customer engagement and pipeline. People often focus on low level lagging indicators when we don’t have the mechanisms in place to gather it all and get a clear picture. We should be looking at our leading indicators and predictive models. Identifying actions and adjusting behaviours as they meet reality.

Capacity is key

The pandemic has been a look behind the current of the health care industry. Internationally. The main reason our public health organizations, hospitals, and governments have been looking at predictive models has been to perform capacity planning.

If the health care system doesn’t have the capacity to handle the pandemic, then the number of deaths will increase, making it more and more unmanageable. It’s a two-pronged response both in terms of gearing up the medical response and dampening a surge in potential hospitalizations. That’s why the early emphasis to use social distancing to flatten the curve to a point to a point where increased hospital capacity could still deal with potential admissions. Though as things progressed, we found other capacity issues than just hospital beds. Ventilators. Personal protective equipment. Testing capacity. And for us in Ontario, a really serious issue with also supporting long term care facilities.

It’s also why I was sent home to go through it alone without continuing support. Unless it escalated. Even though it was the worst illness I’ve ever gone through!  (That and the fact it’s so contagious). The health care system didn’t have the capacity to support non-severe cases. I did reach out to my doctor a few times, who was supportive. And so were friends and family. So I was lucky.

Being sick made me question my own capacity. As the pandemic was declared I was about to release an online course and start a Facebook Ad campaign to promote it. Initially I paused the week of March 16 to reconsider the messaging and create a COVID-19 limited time free offer around it. Then when I got sick all plans were put on hold. I work independently, so it was a big hit to my business’ capacity.  Luckily, I’ve just picked it up again in my recovery. I now have a Facebook Ad campaign running with reasonable results. But my business, which can operate in the pandemic, had to be put on hold during the illness.

The lesson to consider is to look at our own business’s capacity risks. Do we know them? What would surprise you, like everyone was about the need for PPE. Do we have contingency plans and backup plans? Can we identify behaviours that could affect the slant of our business growth curve? Would our capacity to deliver be strained by an unexpected surge in customer demand? A positive, but things may need to be put in place to meet that.

Cultivate anticipation

My spidey sense was tingling the week of March 9th. Enough so that I had stopped seeing my mother because she’s a person at risk.  But I didn’t listen to it enough to not go to a 65 person event, when they were telling us to stay away from 250 person events that week. It was an event important to me and brought me joy. I did carry on that spidey sense by cancelling other things that week. (I also had some tingling about the market the week before, watched rather than make changes, and as a result lost equity.)

The biggest lesson here is around cultivating an ability to anticipate things. At the start of the epidemic you almost needed to be anticipating what was coming next to get ahead of it. The government here in Ontario seemed to be lagging a week. I know other governments lagged longer. It’s the nature of governments, they move slowly. You were most at risk if you waited for someone to tell you what to do or how to respond. Possibly as part of re-opening we should lag the government instructions. Anticipate a resurgence of the virus.

Strategy comes up as my number one skill in Gallop Strengths finder. Intuition as my first function as a Myer’s Brig ENTP. If you read the description of Strategy in Gallop Strengths finder, they say it’s really strategic thinking and not a skill someone can learn in a traditional way. 

I’m still working on whether it can be taught, because I’d like to teach it.  As a person with a growth mindset, I think we may have tendencies, but we can always learn other things. Strategic thinking, in my opinion is something that we can develop or cultivate, rather than learn. It’s a worthy goal. Like all such things.

Lesson learned is to consider what you need to anticipate in your business. Consider the areas of risk.

Insights are a combination of data and intuition. Anecdotal evidence can be as important as hard data.

No one is going to tell me I didn’t have COVID-19. Even with the negative test result. My experience with it, is enough for me. While I waited for the test result, my need to have a test answer dwindled. It really didn’t matter what the test result was, I knew how sick I was.

Some people would ask, as a data scientist are you upset that the data actually failed you. Not really. Because as a data scientist, I know that gaining insights from data is a combination of what the data tells you and a business person’s intuition and experience. It is a failure to rely to heavily on one or the other. You need to question everything. The data as well as our cognitive biases.

In truth, I think public health should have followed up with me anyway. And put me into some probable category given the symptoms. Maybe tested me again. Possibly still contact traced any contacts I might have had. As above, they don’t mainly because of capacity to do so. It’s also a negative characteristic of traditional health care. There are lots of people with other diseases, not confirmed, that struggle to be diagnosed by means other than testing. It’s one of the failures of health care.

The lesson for business is the value of anecdotal evidence as a type of data. Our user stories are as important as our tracking of our users. We need a way to capture both. Often the hard data is easy to collect, and the anecdotal hard, so we forget about it.

Listen to inform, not to obsess

All of us have been consuming some news about the pandemic. Unless you’ve taken stay at home to mean be a hermit. Some of it is informative. Other parts are just plain unhelpful. People have many responses in a pandemic. Some people bury their heads in the sand. Others obsess too much.

For me, feeling left alone to take care of myself and being a person who likes figuring out how things work, I did seek out reliable sources for information. I listened to other people’s anecdotal stories for confirmation of my own. I had to turn off the firehose of negative horrible stories of escalations. They were just too hard.

The goal is to have a balance. Listen to reliable sources to be informed. But don’t obsess over the junk food part of the media.

There are lessons here for our own customer research. We need to listen to and respond to our customer’s experiences. As well, we need to not go down rabbit holes with squeaky wheels. Find balance in what we take in and filter out what isn’t relevant.

People are people, you meet them where they are.

There is a lot of fear around the virus. As well, as I had an early case there was less information out at the time. I had different reactions from people when I told them I had the virus. Magnified by the confusion created by the false negative test. My usual believing mirrors were there with me as they ever are. Others doubted I had the virus.

It’s a phenomenon I’ve come to recognize in engagements with people. Meaning I’ve seen this before. How people interact with you is often coloured by the where they are. A few of my doubters are essential workers who are fearful of going out in it. Other people felt it scary and more real that they knew someone who had it. And in my family where I am the one who never gets sick and is often the caregiver, there was sort of an upending of the relationship. Accompanied by not knowing how to navigate this unfamiliar territory.

Admittedly, feverish and brain fogged, I might not have always responded well, and have some olive branches to send out. Still I learnt over the weeks to pause and consider where the person was in terms of the virus before sharing my news with them. (Seems counter-intuitive to me now broadcasting it, oh well)

I think the same is true with our engagements with customers, co-workers, employees, partners, and suppliers. Sometimes we need to acknowledge where people are, their capacity to hear what we have to say and what might be influencing the engagement. Engage with them, but in a way that we meet them where they are.

Your customer needs to be the hero of your story

In the middle of this, and at the clear part of my days, I listened to a series of LinkedIn Live sessions by one of my hero’s, Whitney Johnson. It is titled Calm Amidst the Chaos. It’s about applying the concepts of her personal disruption framework to help people navigate through these times. She launched into it March 16, as a positive pandemic response.

Knowing Whitney, I also reached out with my early messy tale of woe. We talked (via email) about how I might want to change my story in the future so it doesn’t overwhelm. I noted that Whitney often talks about a period of her life that was difficult in a way that is not dwelling into it but using it to highlight another message. Her response, and one I’ve also heard from Nancy Duarte, is to attempt in the telling to make the customer the hero of the story. I think this is where people who try to be authentic in their marketing often fail. They know that storytelling is a great tool, but they too often make the story about themselves. And then it doesn’t resonate with people.

I’m not sure that I’ve gotten there yet. It may still be a bit early and the experience too close. The intent of this post is not to just tell my experience of COVID-19, but to share insights I gained along the way that we can apply to our use of data in business. Hopefully, people will see the lessons and find them helpful. That is my intent.

In the spirit of Wear Sunscreen…

Any advice here has been based on my own meandering experience. Any real advice on the virus mirrors my doctors, with a little spice thrown in:

  • Practice social distancing and hand washing.
  • Stay home and only go out for essentials.
  • Act like you have it. Wear a mask when you go out to protect others.
  • Stay away from the vulnerable and people at high risk of a severe case.
  • Have a thermometer and oxygen meter at home, so you can monitor symptoms and head to ER if they escalate
  • Virtually connect with people you love. Meet people where they are.
  • Always be looking for the learning in your experiences

Also published on Medium on April 23, 2020