Intel Corp. really, really wants to drive the wearables bus. At the International Consumer Electronics Show (CES) earlier this month, the chip company unveiled Curie, a “system on a chip” or SoC that’s no bigger than a button. The tiny computer chip can be fitted into things like bracelets and bags and sunglasses and was designed specifically for wearable device developers.
Jewelry and sunglasses and even wearable cameras that can fly aside, Intel’s wearable interests are bigger — and much more interesting — than just retail. Last August, the company announced it was forming a new partnership with The Michael J. Fox Foundation for Parkinson’s Research. To help advance the understanding of the neurodegenerative disorder, Intel outfitted 16 patients and nine control volunteers with wearable devices that collected data 24 hours a day, seven days a week, to track things like sleep quality and tremors.
These smart watches are “generating 300 measurements per second — roughly a gigabyte of data per day, per patient,” Ron Kasabian, vice president of big data solutions at Intel, said at Strata + Hadoop World in New York City last fall. “We take that data, take it from the device through the patient’s handheld or cellphone and load it into a cloud-based infrastructure we built.”
The streams of data are ingested and cleansed. (“The accelerometers in some of the devices occasionally emit inaccurate measurements, so we clean those out,” he said.) And then the data is loaded into Hadoop, a distributed computing framework that can process large data sets, for analysis.
Compared to the information collected from the traditional dexterity tests given every three to six months to Parkinson’s patients, the streams of sensor data are a potential goldmine Kasabian said. In the future, researchers are hoping to profit from the power of big data by combining de-identified patient data to uncover correlations that can answer questions such as why the disease progresses faster in some than in others.