Comments on Changes in Laboratory Informatics’ Analytics Usage


I happen to have a side interest in predictive analytics, which is a way of using numbers to make predictions. Those of you who regularly read my blog and newsletter probably already know that I am not keen on statistics, as they’re so often misused or used as ways to lie to us to try to force us believe things that aren’t actually true. As such, this idea that we can use numbers to predict something fascinates me, but I should add that the study of predictive analytics does use some statistics, as well as other tools.

I’m fascinated partly because it’s something I should know about for my business, but probably more because I wonder if it will end up being used to mislead us just as statistics currently are. Are  we sensible practical people or will be be pulled-in by those using it as a simple crystal ball?

Who Knows and Uses This Information?
It is often the small companies like mine that figure out these new types of tools and strategies, first. We read articles and we tell each other about these things. We don’t have to ask permission to try them out. Once the large companies discover some of these things, they have more resources to create departments around them and explore them in more depth. So, when I speak with large companies about these types of topics, they tend to react in various ways. Some don’t want to speak with me on the topic because, “Gloria, your company is so small that you just don’t know anything about anything?!” or because they’re just not doing this type of thing, yet.

When I’m talking about the larger companies that I sometimes discuss this type of thing with, I’m partly doing it just to see how many companies are getting into using predictive analytics. I ask partly just because I’m curious, more than anything else. The types of companies I speak with includes software vendors, services vendors and periodicals that are related to our own industry, mainly, although I occasionally speak with companies truly outside our industry. The responses I get from these companies tend to be similar regardless which category they fall into.

Recent Responses
Recently, though, I’m starting to see that the larger companies are probably starting to use some of this. But there are still those that don’t seem to have realized there is anything like this around. First of all, while I’m interested in this, I will admit right here that I’m no expert and what I do for my own business in this respect is rather limited. With that said, here are the two most recent related conversations I had with some relatively large companies in two different categories (software, service, periodical):

For one example, I happened to be in-contact with one of these large companies about an unrelated topic and we started talking about predicting behavior which was a slightly related tangent to the main conversation. It wasn’t terribly in-depth because we were both in a hurry, but we had a conversation that left me positive that this company was working this type of thing because they knew what I was talking about and actually had a discussion with me on the topic.

The flip side, and my second example, is that I was in-contact with someone from an entirely different large company. In it, the person made a statement that both included a number and was just obviously wrong . I responded to the person that he should be more careful about writing to the public (I was referring to myself) with statements like this because so many of us in laboratory informatics are analytical types who would immediately pick this apart. I gave an example of how the numbers wouldn’t support the statement. In return, he was quite angry with with me. However, while the person wasn’t in Marketing, I made the mistake of thinking that I was in-contact with another numbers-type person, which turned out not to be the case. In this instance, I misjudged my audience. However, I still did learn something.

My point is that some, but clearly not all, of the large companies are now starting to understand that this isn’t all just some voodoo, and that it’s probably just a matter of time before they all are doing it and it’s a more common practice.

Bottom Line
Change always happens. Small companies like mine gain an advantage by being nimble and holding onto whatever the latest useful tools are but the large ones always catch-up, forcing the little companies to find yet another advantage. With regard to predictive analytics, that curve is called the “Adoption Curve” and I like this one partly because it’s in color, partly because it shows the chasm of innovation from the classic book “Crossing the Chasm”
http://www.amazon.com/Crossing-Chasm-Marketing-Disruptive-Mainstream/dp/0060517123/ref=sr_1_1?ie=UTF8&qid=1366983700&sr=8-1&keywords=crossing+the+chasm
While this latest edition was published in 2006, this classic book was originally published in the 1990’s. In any case, here the curve:
http://www.google.com/imgres?imgurl=http://suewaters.wikispaces.com/file/view/Slide12B.JPG/31092781/Slide12B.JPG&imgrefurl=http://suewaters.wikispaces.com/Rogers&h=336&w=448&sz=21&tbnid=SdGxMznuRI-EgM:&tbnh=90&tbnw=120&zoom=1&usg=__vutIpybCLyc9FCJmFDz51WkMClo=&docid=d2n_l93O8kimeM&sa=X&ei=cYN6UcTlLsmVqQGx9oFA&ved=0CDkQ9QEwAg&dur=415
However, the fact that it comes from Rogers or has numbers assigned to it can be ignored for the purposes of this discussion.

Gloria Metrick
GeoMetrick Enterprises
http://www.GeoMetrick.com/


3 responses on “Comments on Changes in Laboratory Informatics’ Analytics Usage

  1. Being a statistics grad, I agree with the idea that the numbers can be made to say what one wants. On predictive analysis, one needs to make sure that they age the data. One business used older data to predict behavior. Years and circumstances can change how people behave. One has to ask how old is the data.

    • One example I like to give is this one that I learned years ago about the danger in eating bread:
      Did you know that 100% of people who eat bread, die?

      Sometimes, people think this is a ridiculous and extreme example, and that’s true, but people do sometimes wander pretty far from the truth with their numbers and we should all keep this in-mind.

  2. I think foretelling the future from numbers/statistics is fundamentally flawed. You can’t do it. You can give a probability and when you act on it you are still taking risk. Folks called day traders (read gamblers) do it all the time.

    One thing is for certain… the rich and mighty (read big company) have always wanted to know the future since the days of Pharaoh and they pay big money for those who wil tell it to them. The good thing is you can’t lose your head today for getting it wrong. All you have to do is give some plausible explanation and wait for the next sucker to pay for your next “scientific” prediction based upon high brow math. Things really don’t ever change. They were once called wizards, now they are called scientists from MIT and Harvard who are laughing all the way to the bank.

    Don’t get me wrong, there is value in predictive analytics but don’t take it to the bank. Take it with a bit of healthy skepticism and hedge your bets. This is how Insurance companies and casinos assure their profits. Those companies do not gamble. Its their customers who are gambling.

Leave a Reply

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