troyastorino

Herbert Hoover was also an engineer.

I don't think it was coincidental that both presidents who were engineers were disastrous for the economy (efforts to engineer the macro economy typically fail).

Definitely true. In the US, because fax and phone were carved out in HIPAA as not being ePHI, they have this special protected status that makes it totally cool for providers to fax records around, but sending emails something that risks jail time :/

I wouldn't underestimate the technological barriers to making interoperable health record systems actually useful. There are a lot of different kinds of medical information (SNOMED CT, the best ontology for healthcare, has >1M concepts!), and the best way to structure that information is an unsolved problem. There are lots of different ways out in the wild (complicated by there being lots of half-assed EHRs that were just made to grab incentive money), and the standards that are out there don't really help things (they are so broad that basically every EHR implements their own "flavor" of the standard).

Sadly, even countries with universal healthcare systems don't have universal health informatics systems (the NHS is a prime example — they spent £12B trying to build an integrated system [1]). Lots of countries attempt, including the US — HIPAA was actually originally about data portability [2], and we just spent another $40B [3]. Thus far only smaller countries have had success with integrated health IT systems [4].

[1] https://en.wikipedia.org/wiki/NHS_Connecting_for_Health [2] https://en.wikipedia.org/wiki/Health_Insurance_Portability_a... [3] https://en.wikipedia.org/wiki/Health_Information_Technology_... [4] https://en.wikipedia.org/wiki/Healthcare_in_Denmark#eHealth

Hey Brian, it's really great that you're doing this :)

If there's more to your medical history that you want to track down, or you want to get your data transformed into a structured format, you should reach out to us at PicnicHealth and we'll see what we can do.

For us — we send out requests for medical records to doctors' offices. When doctors send records back to us, they usually include the cover sheet we sent over. Since that cover sheet has a barcode on it, we know which record request the incoming records are for.

Anyone see anything about how they handle simultaneous incoming faxes to the same number? With other faxing APIs we've used, if we get a new inbound fax while another is being received, the second (new) inbound fax will get a busy signal and not get procewsed.

Phaxio has said they can handle any number of simultaneous inbound faxes, but we haven't had a chance to try them yet. It would be nice to know if Twilio is an option.

Awesome, I was just talking about how I'd want this kind of product!

I'd be more than willing to pay a monthly fee ($5 / month) for you to have someone confirm the details of my meals and label meals that your system doesn't recognize. I'd be happily paying you to build a higher quality training set because

1) I don't want to fill out the extra information (although your interface makes make that process less painful than it would be otherwise)

2) Paying would make me much more likely to be a consistent user

Anyway, excited to try it out and if you every try a paid upgrade I'll definitely be a guinea pig :)

HIPAA is actually pretty terrible in terms of protecting your privacy from government agencies (or from insurers). It can require covered entities (health care providers, plans, or clearinghouses and buisness associates) to hand over medical information to law enforment officials without patient consent, and can even ban covered entiities from informing patients about disclosures under some cirumstances.

The EFF has a lot more detail here: https://www.eff.org/issues/medical-privacy and more specifically: https://www.eff.org/issues/law-enforcement-access

Link to press release from SLAC: https://www6.slac.stanford.edu/news/2016-09-21-schroedingers...

Ah yes, you're right, I was thinking of it that way. Thanks a bunch for your clear and thorough explaination, it makes a lot of sense! So if I understand what you're saying, a 1x1 convolutional layer for collapsing 100 channels to 10 channels would take a 100x512x512 tensor and collapse it to a 10x512x512 tensor?

[Also, sorry for attempting to answer your quesiton incorrectly. I was thinking of putting a disclaimer saying I hadn't worked with CNNs and so might be misunderstanding what the convolutions are doing; probably should have haha]

Maybe when the author was saying 'one can think the 1x1 convolutions are against the original principles of LeNet', he was anticipating my kind of confusion? :)

As far as I understood from the description, in LeNet the convolutional layer lets you avoid training parameters that will effectively be doing the same thing as a convolution. Adjacent pixels are highly correlated, so convolutions can capture most of the information in groups of adjacent pixels without having to train a fully-connected layer of neurons. Effectively, you're kinda downsampling the image without losing information.

So, if you're using 1x1 convolutions, I think you're basically having a neuron per pixel, so you're forcing your fully-connected layers to learn the spacial correlations of pixels, instead of capturing that information in a convolutional layer. In other words, you're wasting training on capturing spacial correlations of adjacent pixels instead of other correlations.

Insurance policies for rocket launches typically expect 1 in 10 rockets to blow up on the pad. ULA's 100% success rate is unusual for the industry. Elon and SpaceX are obviously shooting for higher reliability than the industry standard, but I don't think you should chalk this up to 'move fast and break things' causing substandard reliability.

Awesome stuff! Why are you guys focusing on produce as opposed to say corn, which has so much more acreage in the US?

Uber doesn't actually have to win these legal battles. They just need to prolong the time until a definitive ruling on driver classification is made until they are ready to roll out a driverless fleet. And prolonging the outcomes of these legal cases 5 or 6 years (appeals, etc) seems pretty reasonable.

There would be a political route to changing driving classification, but Uber has an incredible lobbying machine.

PicnicHealth (YC S14) - San Francisco, CA - Full Time, Intern picnichealth.com, demo.picnichealth.com

PicnicHealth helps patients achieve better care by centralizing and streamlining all their medical information.

We're looking for a full-stack and front-end web hackers. Ideally, we want to find an engineer who also wants to take ownership of our product's UX.

For interns, we want CS students (or equivalent) who are entering their 3rd or 4th year at school, and are able to quickly build out web applications.

We run Node and postgres on AWS, and Backbone, Browserify, and React on the client.

Checkout picnichealth.com/jobs for more info, or email jobs@picnichealth.com.

Hey there, PicnicHealth CTO here. Happy to answer any questions you might have about what we do or how we do it! (Or really anything else I guess)

We get records as far back as we can. Unfortunately, records that are more than 10 or 15 years old are often destroyed. Offices are only legally required to keep records for a certain period of time (e.g., for 7 years in California), and so after a patient leaves a practice, the practice will typically clear out that patient's records at some point.

Long story short: we get records any records that still exist.

Thanks for pointing that out! Just fixed the outstanding SSL config issues for the marketing site, and brought it up to 'A'

The original paper that Rognlie wrote:

http://www.mit.edu/~mrognlie/piketty_diminishing_returns.pdf

PicnicHealth (YC S14) San Francisco, CA Software Engineer: Full Time, Intern

PicnicHealth helps patients with serious medical conditions manage their medical records. We continuously collect all of our customers' medical records, digitize and normalize their information, and organize it into a single chart that makes it easy to manage their care.

Our tech stack includes Node.js and PostgreSQL on the backend, and Backbone.js / React on the front end. We face interesting challenges around building a friendly and useful interfaces to complex medical data and around applying machine learning to provide medical insights.

To apply, email jobs [at] picnichealth.com

It could depend on whether healthcare.gov signed Business Associate Agreements (BAAs) with the insurers that it's connecting to. If it did have to sign BAAs, then heathcare.gov would be covered under the scope of those BAAs, and would likely have to be complying with the security rule and the privacy rule.

Ah, didn't realize that there was this dimension to the debate. Sorry for the hasty/inaccurate comment, and thanks for correcting me.

The implication is that people who used to have a full time position would now be forced to take a part time position. The ACA employer mandate only applies to companies with more than 50 full-time employees, incentivizing employers to have less than 50 full-time employees. The thought is that some companies around this limit will avoid the mandate by making a greater percentage of their employees be part-time.

That said, the employer mandate has been delayed until 2015 for companies with >100 employees, and until 2016 for companies with 50-99 employees.

Hi, one of the PicnicHealth founders here. It's great that you were motivated enough to upload your data to Google Health! Unfortunately for Google Health, most people weren't as motivated as you. We believe that Personal Health Records haven't yet been successful because they require users to manually upload their data, and most people just aren't willing to do this. That's one of the reasons why we take care of collecting and organizing records for our users.

About fees, you're right: one of our users would have had to pay over $700 in fees if he had tried to get the records from just one of his hospitalizations on his own! We have relationships set up so that we avoid fees almost 100% of the time, and when we can't avoid a fee we cover that ourselves. Our job is also to sort through the mess and pull out the salient information so that you don't have to.

Labeling it luck also seems somewhat disingenuous as it assumes there is some distinction between people and the nature+nurture that makes them who they are. As someone commented elsewhere, the claim that people are exclusively a result of a combination of their environment and genetics is simply a statement of determinism. If that is the case, there is no luck involved. There is nothing one could be other than the person that he is.

The article is referring to automated electronic access to your medical data (lab test results, diagnosis lists, procedure histories, etc). If you were able to get automated, electronic access to all of you medical information, it would be a very big deal.

Until very recently, medical data was only available from your doctors in the form of copies of paper records, and it's a hassle to pester all your doctors until they give you your records. Even once you got it, the data wasn't digital.

Because of Meaningful Use requirements coming online this year, patients now have to be given electronic access to their data (Meaningful Use requirements determine how doctors have to show that they use Electronic Medical Records in order to receive subsidy money/not get penalized). Electronic access takes the form of online 'portals' where patients can see their data. And although this is a big step forward, the information isn't given to patients in a form that is very useful--the data isn't easily exportable (for yourself or to send to other doctors), and if you've seen doctors in multiple systems your data isn't aggregated.

I'm also one of the PicnicHealth founders, and we're taking advantage of these portals in order to give patients automated access to their data.

Although the relation to HealthKit was somewhat tangential, this post did an excellent job of relaying the status of and problems with digitized healthcare. HL7 is absolutely horrendous, and Meaningful Use has no interoperability requirements and weak prospects for getting good ones.

I think the Continuity of Care Document (CCD), though, is a bigger deal than he realizes. The View Download Transmit (VDT) requirement that requires the CCD is, in practice, making a HL7 3.x XML document available electronically. This opens a great opportunity to build real interoperability between EHRs.

Shameless plug: PicnicHealth (picnichealth.com) is working on taking advantage of precisely this opportunity. We're rolling out our alpha product. If you're interesting in what we're doing please reach out.

Are you serious? What part of integrating computer vision as central to user experience is cloned from the iPhone?

I just started using browserify after using RequireJS for a while, and also got annoyed at how slow builds were. Then I searched around and found watchify, and it's great! Build time during development isn't painful at all anymore.

Legal concepts that are derivative from the natural right to liberty, e.g. freedom of speech, are protected by the government, but that is not the same as them being created by the government. An individual certainly has liberty, and thus freedom of speech, in the context of a state of nature. He does not, however, have a guarantee that an organization with a monopoly on the use of force, i.e. a government, will protect that liberty.