Global Sources
EE Times AsiaWebsite
Stay in touch with EE Times Asia
eeBlogs-Article content Home?/?eeBlog?/?eeBlogs
Nickname:?By Design???? Articles(17)???? Visits(32106)???? Comments(0)???? Votes(7)???? RSS
Find out what EE Times Asia's community thinks about the latest products that made it to market by design or by accident. No holds barred--we rave and we rant. Want to join us? Email bydesignoraccident [at] gmail dot com with your eeForum ID.
Blog Archive
2015?-? Dec.,?? Sep.,?? Jul.,?? Apr.,?? Mar.??
2014?-? Dec.,?? Nov.,?? Oct.,?? Sep.,?? May.??
2013?-? Dec.,?? Nov.,?? Sep.??
View All
Comment?|?Add to Favorites

Posted: 06:42:13 PM, 20/03/2015

Apple's ResearchKit: Is open source good for you and Apple?

? ?

The recent Apple Spring Forward event was notable for the hardware it unveiled. And indeed, here is a quote from my previous comment:

As is often the case in situations like this, the "rest of the story" will only appear in the coming hours, days, and weeks, as Apple spokespersons accidently (or not) let more details slip during the post-presentation product showcase, as developers leak more details in the process of finalizing their applications and associated services, and certainly as initial review units show up at doorsteps.

To wit, I've already learned some interesting additional facts about the Apple Watch and MacBook, which I'll share in a coming-soon follow-up writeup. But Spring Forward's various services announcements were, to me, the highlight of the event. Partner (at least for now) HBO's "Now" unbundled offering was particularly notable, for example; a coming-soon dedicated-topic post will delve into it in greater detail. And the other new-service release that was front-and-center on my radar screen was ResearchKit.

Apple's website defines it as "an open source software framework that makes it easy for researchers and developers to create apps that could revolutionize medical studies, potentially transforming medicine forever." As I heard the pitch, I was reminded of SETI@home, a distributed-computing volunteer project that those of you old enough (like me) and geeky enough (ditto) might also remember.

SETI@home is still around, actually; you can download a copy of the client utility from the website hosted by the University of California, Berkeley (SETI@home is funded by grants from the National Science Foundation, NASA, and donations from volunteers). About those volunteers, here's how the overview page describes the service: "SETI@home is a scientific experiment that uses Internet-connected computers in the Search for Extraterrestrial Intelligence (SETI). You can participate by running a free program that downloads and analyzes radio telescope data."

That "free program" is, I suspect, at the core of why SETI@home is no longer as popular as it initially was. In part, this is because the service has plenty of competition nowadays, ironically including distributed computing projects focused on medicine. And in part, if my particular experience is indicative of the norm, it's because the client didn't do a particularly good job of discerning when the computer was idle and only tapping into available processing resources at those times. Instead, I frequently found my computers slowing to a crawl when I needed them to be responsive for even modest word processing, email, and other similar tasks, unless I manually paused the client beforehand (and remembered to un-pause it afterward).

Implementation hiccups aside, the concept of SETI@home was sound; it relied on the altruism of its participants, something that I'd wager will be even easier to harness when the possible outcomes are improvements in the quality and longevity of fellow human beings' lives, versus the potential for finding signals broadcast by aliens in space. Think about it for a minute, and I think you'll agree that a smartphone or tablet, soon to be (in the Apple world) or already (in the Android ecosystem) paired with a smart watch, is an ideal platform for logging, pre-processing, and uploading to the "cloud" (for further processing, and archiving) a diversity of medical data.

First and foremost, consider the platform's mobility (in and of itself a key enabler) and ubiquity, the latter driven both by our desire to communicate and (in some geographies) the equally important fiscal allure of service provider subsidies. Consider the formidable (and faster all the time) available host CPU processing facilities, along with the additional general-purpose compute capabilities of co-processors such as the GPU, DSP, ISP, etc. Consider the plentiful allocations of both local volatile and nonvolatile memory. And consider the (mostly) persistent and (increasingly) bandwidth-robust cloud connectivity.

Next, let's look at system specifics. Focusing first on the smartphone and tablet, an abundance of integrated resources are available to tap into a diversity of medically interesting attributes:

  • High-precision touchscreen
  • High-resolution front and rear cameras
  • Microphone(s) and speaker(s) (plus external headset support)
  • GPS (for current-location and movement discernment, along with associated current-temperature assessment)
  • Compass (additional movement details)
  • Accelerometer and gyroscope (ditto)
  • Barometer (for both weather and elevation discrimination)
  • Unique identification capabilities, and
  • Encrypted communications support

Now consider that the Apple Watch not only can harness the hardware of its wirelessly tethered iOS partner, but also contains a touchscreen, microphone, speaker, unique ID, encryption support and accelerometer, the latter helpful for additional (and wrist-centric) movement measurement purposes. Plus, it integrates all-important heart rate monitoring facilities. The first five announced ResearchKit applications use this potential in interesting ways, foreshadowing even more creativity to come:

With the Asthma Health app, developed by Mount Sinai, Weill Cornell Medical College, and LifeMap, "data from the GPS in your iPhone is combined with information about the air quality in your city to help you avoid areas where symptoms could be triggered."

mPower, from the University of Rochester and Sage Bionetworks, focuses on Parkinson's Disease and leverages the gyroscope in an iPhone, along with its touchscreen, to measure dexterity, gait stability and other factors.

GlucoSuccess, created by Massachusetts General Hospital, combines user-entered glucose levels with data generated by iPhone sensors like the accelerometer and gyroscope, to enable you to compare your activity and glucose levels.

Share the Journey, provided by the Dana-Farber Cancer Institute, UCLA Fielding School of Public Health, Penn Medicine, and Sage Bionetworks, focuses on breast cancer research. It relies on you inputting data on a regular and frequent basis in order to discern how your activity levels correlate to symptoms.

Finally, MyHeart Counts from Stanford Medicine and University of Oxford places its attention on cardiovascular disease. "The activity data that can be measured with your iPhone can be assessed against your diet and lifestyle information to give you a much more objective Lifetime Risk Estimate," says the description.

Researchers sure seem enthused; check out, for example, this recent quote from Alan Yeung, medical director of Stanford Cardiovascular Health, after having received 11,000 signups for MyHeart Counts in the first 24 hours of the application's availability:

To get 10,000 people enrolled in a medical study normally, it would take a year and 50 medical centers around the country.

And in general, I'm also quite enthused about ResearchKit's potential, not only for Apple's customers but also for the computing and communications world more generally. To that point, and to the best of my knowledge, Apple hasn't yet been particularly forthcoming about what it means by "open source" in the context of ResearchKit. I'm guessing it doesn't mean that, for example, the iOS applications' source code will be freely provided to Android developers ... even if it were, the reliance on operating system-specific APIs wouldn't make the code very portable, anyway. Instead, I suspect what will be openly documented will be the types, formats, and precisions of the data required by a particular service, and the protocols necessary to get that data up to the associated server for analysis and archive.

With that all said, however, I do have a few concerns:

iPhones, iPads, Apple Watches, and the like appeal to a demographic that skews younger and more affluent than the norm. As such, it's not clear to me to what degree the data generated by iOS-based ResearchKit apps will be indicative of the population at large.

Speaking of data, its associated demographics are highly dependent on the honesty (and more general willingness to divulge personal information) of its users. Age, gender, ethnicity, income level, height, weight, and other similar attributes are both important to researchers and feasible to be fudged by participants.

Speaking of personal information, Apple and its ResearchKit partners will need to maintain absolute scrupulousness regarding how the data is handled in order to obtain and retain users' trust. Can a particular data set be traced to a particular person, for example? And will it be retained only for research purposes, or subsequently transferred to commercial enterprises? Apple has subsequently clarified at least some of these potential issues by adding entries to its developer guidelines, but it'll need to remain vigilant, including rapidly yanking apps that inevitably stray into misbehaving territory.

"Sandboxing" is another area that'll require ongoing vigilance. Apps that access information on a user's device beyond the strict data set necessary for whatever medical condition is being monitored must also be quickly dismissed from the App Store.

Speaking of age, and the recently augmented developer guidelines, one of them insists that apps must require minors to obtain consent from a parent or guardian before the program is allowed to run. The means by which this consent is assured is not yet clear, nor is the ease by which a minor can falsify his or her age (corrupting the integrity of the logged data set in the process).

Finally, government organizations such as the FDA and FTC haven't yet (again to the best of my knowledge) been explicitly mentioned by Apple in the context of ResearchKit. Recently, the FTC fined several companies who claimed that their apps accurately diagnosed skin cancer. Such mole-growth and other assessment capabilities are in fact possible, especially when a series of images consecutively captured over a period of time are compared. Where these companies seemed to go astray, however, was in going beyond marketing their products as diagnosis guides and instead promoting them as definitive diagnosis tools.

Share your thoughts on the latest from Apple in the comments below.


Brian Dipert is Editor-in-Chief of the Embedded Vision Alliance. He is also a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company's online newsletter. And he's an off-hours freelancer as the Principal at Sierra Media, where he contributes to (among other things) the Brian's Brain blog at EDN Magazine. Brian has a BSEE from Purdue University in West Lafayette, IN. His professional career began at Magnavox Electronics Systems in Fort Wayne, IN, where he worked for an aggregate 2.5 years as a co-op engineer.


Views(758) Comments(0)
Total [0] users voted ????
[Last update: 06:43:10 PM, 20/03/2015]
Have Your Say!

Bloggers Say

Got something to say? Why not share it with other engineers?

Just introduce yourself to us, we'll contact you and set you up. Yes, it's that simple!

See what engineers like you are posting on our pages.

Interviews & Viewpoints


Learn how senior executives are seeing the industry from interviews and contributed opinions.

Back to Top