The Intern's Metaphor on: Siren Data Analytics

Thursday, December 3, 2009

When I was so new to the company that the new car smell hadn't worn off me yet, I asked one of the people working on it, "how would you explain a data cube for Data Analytics?"

She smiled and said, "That's what we've been trying to figure out for the past few months."

I'm doomed, I thought. So I got down to studying it - what is it for, what does it do, who would use it?

Siren Data Anaytics is developed for agencies that want an easy tool to identify trends within the patient care reports made by their paramedics, in order to answer important questions, such as "how efficient are our procedures?" and "what changes could be made for better patient care results?"

To put it as simply as possible, for a person who has little to no programming experience: if a data cube wasn't a packet of data, it would be a room full of patient care reports, each lovingly printed out and piled up in neat stacks. Depending on how many incidents your company had, the number of such reports could range from just a few stacks on a table to thousands of reports piled up, taking up a little office. Piles and piles of patient care reports, waiting to be sorted through so your agency or administration could identify trends and figure out what they could do better.

Just imagine you were the office intern set to do all this sorting work. One day, your boss came in and said, "Okay, sort all the reports by gender." And you could spend a few happy hours sorting the reports into two piles - male and female patients, recording the numbers. Then when you were done, your boss would say, "Now, out of the pile of female patient records, get me those who had cardiac incidents." So you'd go through that pile, searching for cardiac incidents. After this, your boss would say, "record down who were transported and who weren't." You'd have to count who were and who weren't, and you'd write it down somewhere. Say your boss continued on with the instructions: "separate the pile by location of patients", and "split them into age groups," and down to "search patient records for a Ms. Brown. She might be in that pile" and you'd probably be able to tell by now whether there's a Ms. Brown or not in the few remaining reports you had left.

At the end of it, you'd give your boss a patient care report of Ms. Brown on this date, with such and such results, from such and such an incident. You might also hand him the numbers he'd asked for. Your eyes would be watery from looking at so much text.

That is what we would call: drilling down to a specific patient care report.

Say your boss wanted to know other statistics and said, "okay, put all the piles back together. Start by separating the patient care records into age groups instead." So you had to go through all the records and sort them by age groups. (You might try to give each age group different piles for gender too, just in case.) After this, your boss wanted you to sort them based on medication the patients received.

After you were done recording the numbers of patient care reports by age group, and then medication results per age group and then whatever else your boss wanted you to do, your eyes would be watering over some more, your bladder full from the caffeinated drink of your choice and your system wired from said caffeine.

In Siren Data Analytics, this is what we would call slicing and dicing.

The beautiful thing about Data Analytics is that, all those painful hours I just described? Could be done within seconds. Minutes if your computer database was a bit slower than par. It would instantly give you the numbers you need. All those tedious hours sorting through that paper could be spent actually studying the numbers and figuring out trends within your patient care reports. Of course, I'm only simplifying, and there're a ton more complex things you could do with Siren Data Analytics.

Just imagine that. A room full of patient care records on paper. All that could be a data cube in your computer database.

Jaymee
Intern