Here's something to ponder: “[Healthcare industry providers and payers are spending an enormous amount oftime and resources normalizing data and layering on analytics, but no one really knows what they need to know.” Healthcare isn't the only industry looking to analytics for insight and innovation, and the conundrum about not knowing what you need to know affects everyone.
But I think people are beginning to appreciate that great insights are not just going to emerge by themselves from even the biggest, baddest collection of data. We need to be asking questions of our data – even if they're not the right questions from the get-go.
This is the objective, a scientific approach. Astronomers don't just measure a load of star light and then expect the data to mold itself into truths about the universe. They come up with theories, and test them against the data.
That's what businesses need to do. If you're working with, say, health data, then first you need a data standard which ensures you're analyzing the data properly – comparing apples with apples. Then you need to frame some theories. Off the top of my head: If the Emergency Room is busy on a Friday night, it will be quiet on a Monday morning. This is based on the assumption that some percentage of accidents is related to the onset of the weekend. We could chew through the data and find the theory isn't true in one hospital – perhaps because Monday morning is short-staffed or people have realized you can't get seen at the weekend, so they wait for Monday.
That's not an exciting example – but it could be part of a chain of investigation and process redesign that eventually saves lives. It could easily be a dumb question, or a question that doesn't give as high a return as some other question. But how long does it take to frame and process such questions? Literally, seconds. Let's not wait around for lightning to strike. Let's standardize the data, and start the queries. HDM