Hacker News Reader

Comment by chatmasta | Hacker News Reader

This is probably the coolest announcement from the event, at least from the perspective of societal benefits.

Medical research faces a real challenge of selection bias. Survey respondents and study participants largely come from two groups: 1) undergrads responding to a campus flyer, and 2) local university-town residents looking to make extra money. Obviously this means respondents are geographically homogenous, and in many cases socioeconomically homogenous as well. Enabling a wider reach for researchers will not only increase sample size, but also eliminate selection bias. This is a very, very good thing.

Also, now research isn't limited to only university researchers.

I wonder if the plan is to allow participants to earn credits on apple pay.

View on HN

Considering the underwhelming nature of healthkit and the rather rudimentary substantial use-case, I am rather reserved.

We've had similar stuff like ResearchKit in existence for quite long. One of the most popular of such sites is patientslikeme.com .It and others did a great job around this, but usage is relatively rare.

The hard thing about such sites is:how do you incentivize people to contribute really personal information and do so on a regular basis(even if said action takes a relatively long time) ?

So i wonder what can Apple do to make this activity common, that other companies couldn't do ? How will they achieve this ?

> So i wonder what can Apple do to make this activity common, that other companies couldn't do ? How will they achieve this ?

A large marketing budget helps, e.g. see the video they launched with. Making it easy to use and frictionless would also help (no need to log into the website, no need to figure out how to sign a PDF or download a consent form or whatever the normal user flow is).

What concerns me is people gaming the apps; downloading them and filling them with fake data just to be malicious. You fix one problem and introduce another...

There's a huge opening here if pervasive sensors ever become consistent, calibrated, and easy-to-use-properly enough that they can supplement in-clinic data streams with at-home ones. This would open up the research contribution opportunity to people who cannot participate in clinical trials today due to distance from trial sites---this is something like 99% of the people who might feasibly be interesting to a clinical trial.

The big step in clinical research is the PIII trial and these are almost universally held up in enrollment. If this stuff was pervasive and trusty enough to be used in even a fraction of trial therapeutic areas it would revolutionize research and medicine in those indications.

All that to suggest the opportunity. I'm not exactly bullish on Research Kit tbh though. The "if" in my first sentence still appears to be a ways out and the kind of "throw the kitchen sink at it and average everything out in post" analytics championed by big data/machine learning of today doesn't fly well with the FDA.

At least not yet.

I don't see the similarity with patientslikeme.com

Apple has created a UI toolkit to enable researchers to directly get data from participants via an app. All control is in the hands of the researcher creating the app.

And the app will be used for traditional research studies, the only difference is that participants enter data via their app.

You would move selection bias towards users who can afford the Apple Watch. This will not be inherently worthless data, but will be biased nonetheless.

Apple watch is not required.

I think having millions of participants in most cases is very worth having the bias that it will introduce. It'll be better than the middle class white kid bias that already exists in a lot of this research.

Selection bias will now be those who have Apple products.

APPL said research kit will be open source, so if that is indeed true, then it will depend on those non profits if they will support android. if health kit gets ported to android.

I wonder how easy it will be to bias the samples for profit. Make some bots to submit bogus data that shows that people that consume product XYZ are healthier. Or, just make the health apps bias the data. Like say company XYZ gives kickbacks to "RecordYourActivity" app in exchange for reporting something that suggests MySnakeOil is good for you.

Of course I guess that kind of thing happens already. Just wondering will this make it worse, better or about the same.

Responses will likely be correlated to iPhones and AppleIDs in some way, so the spammers won't be able to do much.

On the flip-side this would be great advertisement information.

spammers maybe not but what about just subtly reporting false/bias results in otherwise legit apps?

The Parkinsons app, for example, correlates symptoms and physical activity. I don't know who would have an incentive to bankroll data skewing.

The current crop of apps have nothing to do with rating products.

Reply on news.ycombinator.com