Viewing entries tagged
quant self

Wolfram Alpha and big data for healthcare, or why I really hate my alarm clock

  The quality of my sleep doesn't determine how long I sleep. You'd think it would, right? I mean, if I don't don't sleep well, shouldn't I sleep longer to compensate? On 68% of nights, I sleep between 5.7 hours and about 3 hours. I wake up about 9 times a night. 

There's a good reason quality doesn't affect duration. It is small, mostly black with a glowing red face. Every morning it jars me out of  my peaceful slumber with an auditory slap to the senses. I kinda hate my alarm clock.

I learned all of this, and more, about my sleep habits in about 30 seconds. First, I downloaded a month of sleep data in Excel format from Fitbit. Then I uploaded the Excel file into Wolfram Alpha's new Pro site. The site spun around, did a little animation and boom - graphs!

Unless you are a data geek or survived college calculus, you may not know about Wolfram (the makers of Mathmatica, often used in academic settings). But, if you are an Apple iPhone 4s user, then you likely use Wolfram Alpha every day. Just ask Siri to convert your 10k training route to miles. Wolfram Alpha is the backend for a lot of Siris mathematical and data based answers.

Last week, Stephen Wolfram and team announced a new Pro feature for $4.99 per month ($2.99/month for students). With Pro, you can upload your own data sets in a verity of formats, including:

  • tabular
  • CSV and Excel
  • audio files
  • pictures
  • PDF and CDF (for interactive graphs)
Once you throw some data at Wolfram Alpha, it quickly goes to work like a million little digital ninjas, slicing and dicing the results. In a basic query, Wolfram will attempt to develop graphs, statical analytics, and interpretive results. You can go further by specifically asking for variance (ANOVA) or regression tests, for example. There's a lot you can do which I don't claim to understand.
Why am I, a guy who bombed out of Calc 301 in college, so excited about math and data? There is tremendous potential in healthcare.
As the old expression goes, we are data rich and analysis poor. Healthcare providers create a lot of data. With the growth in adoption of electronic medical records, it is only getting deeper. We have piles of metrics on health, weight, clinical measures, time, money, outcomes, quality, infection rates, payments, demographics, and usage. Do people who get annual physicals have lower cholesterol than those who don't? I'm sure we can find a study in the New England Journal of Medicine. But now with tools like Wolfram Alpha Pro, we can throw our own data sets against the big metal in the cloud and see for ourselves.
From a patient's perspective, this has big implications for the quantified self movement. I have a Withings scale, which outputs weight data over time in Excel format. My Withings blood pressure cuff does the same for my BP. My Fitbit knows about my food consumption, activity levels, caloric burn and sleep patterns. My Digifit tracks and exports my cardio response to exertion, along with resting and recovery heart rates. If I can bundle all that data and have a site like Wolfram Alpha analise it, do I need an expensive sleep study? Is there a correlation between my diet and sleep quality? Maybe I'd learn that nights when I eat fewer carbs, I sleep better but preform less well in the gym the next day.
If you think all those sensors, gadgets and data files sound are probably right. But just wait. Personal health devices are popping up everywhere. It is only a matter of time before we are dripping in data about ourselves, our health and our bodies. Will we need an expensive health system to understand the data, or will sites like Wolfram Alpha put the understanding within our own reach? How does that change the role of the doctor, and provider organization?
One thing is already clear, I really do sleep longer on weekends:

By the way, on an average night, while peacefully dreaming of healthcare for all, I'm bombarded with 372 times the amount of radiation I'd get from eating one banana. Data!

Wolfram|Alpha Pro: Experience the Next Big Step in Computational Knowledge.