A story about money, survival, and data

In my previous post I talked about how collecting and analyzing data can provide opportunities for positive change in our personal lives and in our work.

This time, I have a story to tell.

This story is about a girl named Juniper. Before I go on, I want to say that this story has a happy ending. Juniper is real. She was born premature, far too early. Her parents ended up writing a book about the experience, titled “Juniper”. Here is a brief excerpt from the book’s description:

Juniper French was born four months early, at 23 weeks’ gestation. She weighed 1 pound, 4 ounces, and her twiggy body was the length of a Barbie doll. Her head was smaller than a tennis ball, her skin was nearly translucent, and through her chest you could see her flickering heart. Babies like Juniper, born at the edge of viability, trigger the question: Which is the greater act of love – to save her, or to let her go?

To Juniper’s parents, the answer to that question was obvious. They wanted to save her life.

To doctors, hospital administrators, and insurance providers - the answer was less certain. Survival rates of babies born at 23 weeks is estimated at roughly 25%. Only one in four make it. They are also at a higher risk for developmental disabilities.

There was also the issue of cost. A single baby born earlier than 28 weeks gestation might need upwards of $200,000 in medical care by age seven. Juniper’s care cost more than $6,000 a day.

This is where Juniper’s story of survival met the reality of money. Fortunately for Juniper, it all worked out. But what if the circumstances were different, what if money was an insurmountable obstacle - would it be right to give up on a nascent life because of money?

Money is a very real practical consideration for many.

What if there was something that could be done to lower the costs of caring for premature babies? Lowering costs is not only a cost-saving measure for the hospital. It also means that we can change where we draw the line for which babies are worth saving and which ones aren’t. If cost is a significant factor in that calculation, lower costs means more babies may be given a fighting chance of surviving.

This is where data enters the picture.

As it turns out, premature babies are connected to all sorts of monitoring devices and equipment that produce a ton of data over time. Here is how one medical center put that data to good use:

Having collected ten years of [premature infants’] data, Universitair Medisch Centrum (UMC) Utrecht […] developed a smart algorithm […] that can support or deny the suspicion of an infection in premature babies, such as sepsis (blood poisoning).

Why use a smart algorithm?

The model that was developed has an accuracy of ninety percent in forecasting the presence of the bacteria that causes sepsis. This is significantly higher than when doctors make predictions [in which case] the accuracy is forty percent.

Wow! More than twice the accuracy of doctors.

And why does that matter? Because it comes back to money. When doctors are not sure if a baby has an infection, they tend to err on the side of caution and administer antibiotics. The treatment has a cost associated with it. Further, babies with an active infection have to stay in the hospital longer.

Sixty percent of babies were treated unnecessarily with antibiotics

We’re talking about costs of antibiotics plus thousands of dollars per day for every extra day that a baby is at the hospital. It’s a case of data and algorithms used to save lives by reducing costs.

Now, we don’t all have stakes as high as this where lives are on the line. Yet we can still all benefit from meaningful improvements brought about through smart application of data retention and analysis. Better products, more satisfied customers, fewer defects, higher quality services, …

Talking about data retention and analysis, notice one important aspect of this story.

Having collected ten years of data …

They had collected, stored, and analyzed 10 years of data. That’s likely billions or trillions of data points. How do you even store that much data, let alone analyze it all?

This is what event-sourcing is designed to help with. I’ll talk more about how we can achieve this in future posts.

Have any questions or comments about this post? Email me at sasha@persistr.com

Get Daily Event-Sourcing Tips

I send out a short email each weekday with code snippets, tools, techniques, and interesting stuff about event-sourcing.