Hacker Newsnew | past | comments | ask | show | jobs | submit | ttandon's commentslogin

One thing that I'd point out is that this comes entirely from the press rather than the startup - we see ourselves in a totally different sub-industry from Theranos (Athelas is a single type of blood test, at-home device, specifically for oncology patients, and a totally different technology). Really the only common word is "blood", but unfortunately I think that often gets drowned out in the media


agreed! checkout our data (including 3rd party run studies): https://athelas.com/publications.

we also ran a live demo with sarah from TC here: https://techcrunch.com/2017/08/28/athelas-launches-its-amazo...


one of the founders (Tanay) here - happy to answer any questions. we also put out a new trove of data here: https://athelas.com/publications


Hi, pathologist here.

* How do you account for interstitial fluid expressed into the droplet?

* Do you use the same flags as a CBC analyzer?

* Have you tested blasts?

* Have you tested leukostatic samples (very high white count: over 100k)?

* Have you validated against platelet clumping?

* Have you validated for basic sources of interference (lipemia, hemolysis, icterus).

* Have you engaged Cerner? Epic? DrChrono? AllScripts? Other EMR providers?

* Have you engaged a pathologist?


Great questions

-The first line of defense against interstitial fluid is wiping away the first drop (commonly used protocol in most drop test). The inevitable remaining interstitial fluid chunk is then classified out by our image processing models, which have been trained to recognize differences in images based on debris/cell concentration.

-We currently report the WBC, WBC Differential, Platelets, and RBC indices. The flags we're missing are secondary RBC metrics like MCH and RDW estimates but we have plans to roll these out soon as well.

-Yes, we've run both bench and clinical testing on patients with north of 100k white blood cells - the image processing can segment out cell chunks in these crowded samples better than traditional flow cytometry. Not in the currently published studies, but some good data coming out about this soon.

-We've run basic initial tests on platelet clumping (artificially induced in certain diluted samples), and present in clinical samples. Since our model looks at the boundaries of these platelet cells, we have a lot less trouble with this than flow cytometry/beckman and even human pathologists. Still, we're working to find more samples with clumping to define the limits of detection on this front.

-We've done interference studies with Hemolysis, EDTA, not with lipemia and icetrus. These are on our list of clinical samples to source as well.

-We've begun working with some of these EMR providers, the good thing is many of them have sandboxes to get setup in relatively quickly. The larger ones do have long, long-term engagement timelines.

-We work with dozens of pathologists to help review results, train the system further, and in general go over good morphology practice. Has definitely been a core part of our strategy.


Wow great to have a Pathologist in the Room.

I think this device is super awesome and I was trying to come up with something similar a few years back.

The primary concerns is that one of the first things I learnt in Med School is that capillary blood is way less reliable in compare to venous blood for something as important as CBC.

Not sure how they are accounting for the issue that the sample they have isn't the best one to get started with.

ps: killjoywashere how do I get in touch with you ? Am at Asingh [at] healthenclave.com


Isn't "athelas" Tolkien-inspired? The Tolkien estate isn't super friendly with regards to this kind of reference. Try to find out how much Palantîr paid for an agreement with them (my understanding is that it's a very large number), and ask if that really makes sense...


The Tolkien estate is full of shit, they have no leg to stand on. Names aren't copyrightable and the book certainly doesn't give them a trademark over these industries.


Can one device be shared by multiple people?

What are the details behind $20/month? E.g. is the machine "rented"? Or is the $20 for consumables?

Is it connected to the cloud? (In other words, is the machine-learning algorithm's "counting" happening in the device or are you uploading image data to a server for analysis?)


The $20/mo is for the device and the test strips per month, as such the device is rented. Each subscription comes with 10 test strips a month usable between different users!


You guys definitely seem to be on the right track - especially with being open about your results in light of the whole Theranos debacle. I remember an article a few months ago describing your CV/ML based approach and think it's a great use case. Good luck.


Did you submit a 513(g) to get classified as (apparently) Class I Non-Exempt?


We're working under the 510k Class 1 and Class 2 framework


What are false-positive and false negative rates for Circulating Tumor Cells with your tech?

Do you control for age and other bio-factors and how?


CTCs are a very early application in our tech and one that we're excited to continue working on. We have some strong sample share relationships with Stanford and sensitivity/stage/specificity testing is in its early stages right now. Will publish on the data page when more information


Look at CU-Anschutz as well, CCTSI is very active in CTCs.


As a former anemia patient and deep learning practitioner, I'm really rooting for you guys. This is a great approach.


Thank you!


did you purposely pick a name that sounds kind of like Theranos just to mess with people?


Does anyone know why a side-view of the actual touchdown (like they usually release after) hasn't been posted by SpaceX yet?


Dhruv's planning on putting out a series of posts, where he'll probably be including the model training/set up components! Look out for the next bits this coming week.


Athelas (YCS16) | Full-Stack Engineers | Mountain View, CA | onsite | http://athelas.com | fulltime

Athelas is hiring full stack engineers (react/mobile, python, js, backend). We're making blood diagnostics low-cost and decentralized. Send resumes and projects to tanay[at]getathelas.com


Interesting - thanks for the feedback! have been internally chatting about making the video beginning/in general more technically focused as opposed to overly emotional (something we noticed was coming about), helpful to hear that here as well. I think the direction we were taking was based on clinical fact (lower cost, earlier testing esp for baseline metrics strongly associated with better clinical outcomes), but I can see how it comes across with that over-hype vibe.

will work on fixing


Thanks!


PS. "See the demo here" links to https://www.youtube.com/watch?v=fEcx3hx-398 that says "private video."


We have two versions of our device - one that's fully automated (the one requiring a 510k), and another where results are manually confirmed post computer vision in-house before being tabulated (this is classified as a "manual count" or hemacytometer). To start off distributing, we're shipping with the Class 1 version to beta users and early locations. Once the FDA clears our algorithms to be fully automated (under class 2), that won't be needed anymore. Let me know if that makes sense


I see, GHO exemption. Yes that makes complete sense.

So Malaria test is off table for now right? That can't fall under class 2.


Hi HN! Athelas cofounder Tanay here. Happy to share any more stories/answer questions about the post.


From the website, it looks like you're taking a direct-to-consumer approach by selling a $250 device and a $15 monthly subscription service (Uber for blood testing!?).

Can you explain why you chose that go-to-market strategy? I don't know much about this market, but it seems non-traditional, to put it gently.


There are a lot of people we've started working with (elderly populations, old age homes) for whom a monthly test is really useful (at-risk for a lot of WBC screenable conditions), but who don't want to buy an entire device/learn to use with app and related software. The monthly screen came as an offshoot of this population's use case.

While the front-facing website is a direct to consumer device, the majority of our deployed devices will be partnerships formed with clinics - this just enables anyone who wants to use the tech to be able to without the long email, custom quote, phone call, partnership process.


Just like in engineering, the initial investment into a go-to-market strategy is only the tip of the iceberg; it's the maintenance and support required to make it successful that kills you.

At both my former and current company, I've experienced firsthand how distracting splitting your focus on multiple customer segments can be.

I would advise you to recognize upfront that the consumer and clinic partnership channels entail vastly different motions for acquiring and supporting customers. If you're investing resources into pursuing both channels simultaneously, make sure that you are measuring ROI for each one.

Hopefully you're getting good go-to-market advice from YC and your investors. If not, consider applying to StartX, where there's a solid community of AI-for-med startups (caveat: I'm an alumnus).


If I buy a device now, can I upgrade it as you add more tests?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: