DoD Wants Help with “Predictive Health” (Scroll Down) | Hacker News Reader

DoD Wants Help with “Predictive Health” (Scroll Down)

diu.mil

I’ve worked on a project with DIU. I was really impressed with the process and how they short circuited the often long and bureaucratic process of getting new technology into the gov.

Thanks for the insight! Any details of the project or process you can share?

if you scroll down you'll find a funded opportunity to work on a healthcare ML project, incidentally backed by the DoD, which as you know has a pretty awkward relationship with SV (including non-unique URLs). But in this case, they're trying to pay people to hunt cancer, other healthcare challenges.

This bit is fairly unique to the OTA contract vehicle that Defense Innovation Unit uses, and I suspect of significant interest to start-ups:

"Companies are advised that any Prototype Other Transaction Agreement (OTA) awarded in response to this Area of Interest may result in the award of a follow-on production contract or transaction without the use of further competitive procedures."

If you have a look here (1), you'll see the follow-on production contracts can be quite substantial.

(1) https://aaf.dau.edu/contracting-cone/ot/prototype/

I mean, it's incidentally backed by the DoD, but it makes sense. Between Tricare and the VA, the U.S. Military is one of the largest single payer healthcare systems in the world.

The interesting project that seems only fitting to google or amazon is the collation of ALL data in the world regarding threat analysis. Seems like the bellingcat scenario writ large.

Google tried this with the original variant of Chronicle and failed.

Threat analysis does not play well with common types of AI/ML applications because the problem space changes so frequently with the design of avoiding predictability. You can gain "awareness" of events using AI/ML (e.g., this road suddenly has tanks moving across it), but it's tough to call that a "threat" (maybe other intel told the analyst those tanks were supposed to be there). Compound that problem to "the world" or even "an entire country" and you run out of compute real quick.

Most solutions err on the side of just popping up alerts but don't identify "threats" until an analyst looks at them.

I know of this project (scroll down to #7) and I've met (online / video conference) the person coordinating it. He is damn impressive.

What is the DIU? Is that DARPA by another name?

DARPA explores the art of the possible.

DIU is more like YC with deeper pockets. Replace Paul Graham with Ash Carter. They have government customers with requirements and their job is to find companies who are capable of building them. They are more interested in the people than the product. They don't care about the prototype MMPRG that pits cats vs dogs on your cell phone, they care about the distributed systems engineers who built it. So DIU is staffed by scientists, engineers and military operators who have proven their ability to execute at an exquisitely high level.

Ash Carter was Obama's last SECDEF. He got a Rhodes scholarship and instead of going to Oxford to get a master's, he went to Oxford and got a PhD in theoretical physics (in 3 years). He was instrumental in the MX missile basing decision, the Nunn-Luger denuclearization deal, correcting the mistakes that bogged down the F35 deal, and much of the weapons developments, nuclear and conventional, in the DoD for the last 40 years. On his way out the door he created the Defense Innovation Board, which led to the development of the JAIC and continues to guide both DIU and the JAIC.

Reply on news.ycombinator.com