idealmedtech

Ideal Medtech | https://idealmedtech.com | Senior Systems Engineer | Senior Software Engineer | Asheville, NC | Full-time | ONSITE preferred, FULLY REMOTE with the right skill set (USA only)

Ideal Medtech is a clinical stage medical device startup looking to control every feasible physiological parameter in the hospital -- starting with glucose. This technology has the potential to save tens of thousands of lives a year, billions of healthcare dollars, and thousands of nursing hours.

We're looking to hire a senior engineer with experience developing medical devices in the US or European Union. You'd be working directly with me, the technical founder, owning product development and documentation required for three Class II 510(k) submissions (pumps, control software, and maybe helping a strategic as needed with sensors). Some (but not all) standards you should be familiar with: 60601, 62304, FHIR

You would be working directly with a team at a contract manufacturer (CM), and would be involved in high-level technical decision making, software development, and helping the CM with documentation.

As an early hire, you'd have the option to take equity as part of your payment.

If any of the following skills describe you, feel free to send me an email (see profile):

- Medical Device Cybersecurity

- Firmware Development

- Formal Verification

- Physiological Closed-Loop Control

- Knowledge-Based Systems

- Reinforcement Learning

Please include your CV, salary expectations, timeline for hiring, and any open-source code you're proud of and would want to talk about!

Ideal Medtech | https://idealmedtech.com | Senior Systems Engineer | Senior Software Engineer | Asheville, NC | Full-time | ONSITE preferred, FULLY REMOTE with the right skill set (USA only)

Ideal Medtech is a clinical stage medical device startup looking to control every feasible physiological parameter in the hospital -- starting with glucose. This technology has the potential to save tens of thousands of lives a year, billions of healthcare dollars, and thousands of nursing hours.

We're looking to hire a senior engineer with experience developing medical devices in the US or European Union. You'd be working directly with me, the technical founder, owning product development and documentation required for three Class II 510(k) submissions (pumps, control software, and maybe helping a strategic as needed with sensors). Some (but not all) standards you should be familiar with: 60601, 62304, FHIR

You would be working directly with a team at a contract manufacturer (CM), and would be involved in high-level technical decision making, software development, and helping the CM with documentation.

As an early hire, you'd have the option to take equity as part of your payment.

If any of the following skills describe you, feel free to send me an email (see profile):

- Medical Device Cybersecurity

- Firmware Development

- Formal Verification

- Physiological Closed-Loop Control

- Knowledge-Based Systems

- Reinforcement Learning

Please include your CV, salary expectations, timeline for hiring, and any open-source code you're proud of and would want to talk about!

Ideal Medtech | https://idealmedtech.com | Senior Systems Engineer | Senior Software Engineer | Asheville, NC | Full-time | ONSITE preferred, FULLY REMOTE with the right skill set (USA only)

Ideal Medtech is a clinical stage medical device startup looking to control every feasible physiological parameter in the hospital -- starting with glucose. This technology has the potential to save tens of thousands of lives a year, billions of healthcare dollars, and thousands of nursing hours.

We're looking to hire a senior engineer with experience developing medical devices in the US or European Union. You'd be working directly with me, the technical founder, owning product development and documentation required for three Class II 501(k) submissions (pumps, control software, and maybe helping a strategic as needed with sensors). Some (but not all) standards you should be familiar with: 60601, 62304, FHIR

You would be working directly with a team at a contract manufacturer (CM) we've done the initial product development with, and would be involved in high-level technical decision making, software development, and helping the CM with documentation.

As an early hire, you'd have the option to take equity as part of your payment.

If any of the following skills describe you, feel free to send me an email (see profile):

- Medical Device Cybersecurity

- Firmware Development

- Distributed Computing

- Formal Verification

- Repeatable Simulation

- V&V

- Physiological Closed-Loop Control

- Knowledge-Based Systems

- Reinforcement Learning

- Blood-based Sensing

Please include your CV, salary expectations, timeline for hiring, and any open-source code you're proud of and would want to talk about!

Roman Hovorka (last author on the paper) has been working on this problem at Cambridge for the better part of 20 years! His was one of the first to see hospital use (through a partnership with BBraun) and also to employ the more advanced MPC techniques from chemical engineering.

Finally a thing I'm qualified to talk about! This is my area of expertise. The term "artificial pancreas" is a historical artefact of the original systems from the 1980s, which called anything that controlled blood sugar effectively an "artificial pancreas". These systems do not cover any of the hepatic or hormonal functions of the pancreas, just the malfunctioning islet cell part.

That being said, commoditization of a fully closed loop insulin delivery system would be a godsend for the millions of people living with completely uncontrolled diabetes. Uncontrolled diabetes accounts for between 30-40% of raw healthcare spend every year, and it's growing. Primary care like this will keep patients adhering to it out of the hospital and improve their lives immeasurably.

So no, I wouldn't compare this to an insulin pump alone. It's like comparing the latest self driving cars out of Waymo to the cruise control on a 1994 Corolla.

The results (66% TIR 70-180mg/dL) aren't incredible, but believe it or not are far better than most diabetics are able to get. For reference, a person without diabetes will typically have numbers in the 95% or higher range.

We're currently finishing a study in the hospital to improve on these results in a big way!

This would put companies like Audm out of business, but it seems like they already only employ one voice actor for most gigs (ya gotta respect how much she gets done though!). I wish there was more work for professional voice actors, audiobooks done by the likes of Roy Dotrice are an absolutely fantastic ride

> At the point in time when you slice the domains you might not know all the product requirements. Probably a feature will arise which forces you to tangle two services together - now you have one domain. But distributed. Urgs.

Are there any architectures which would allow you to monolithize services like this? Changing product requirements sometimes mean you've just arrived at a better abstraction, which may need you to combine services directly.

Type 2 in many ways is a disease that comes from struggling with diet and exercise. There are many medically valid reasons for that to be so, but it still is necessary to try and address them as much as you are able if you're struggling with Type 2. It may save your life.

That's so interesting! Giving me an idea for a side project that I'm sure has been done many times before :)

About to take a whack at reading your paper, but in plain programmer speak, can you explain a few ways this might be exploited in the wild?

A post like this without Twitter in the title isn't nearly as provocative, and would probably have died in /new. It's because of the authors decision to scope out a Twitter clone that we're having this discussion!

Yep, not to mention many groups will list new members higher up so they get cred faster

My home server is connected to my TV, so I can have a fully functional desktop environment when watching media. Probably the most useful thing on it is a little Arduino that's hooked up to a Django REST API that I can use to control my TV (lost the remote years ago).

Also the server itself is my ancient T530, which is still quite snappy on Arch!

Here's the remote code, for anyone interested: https://github.com/ijustlovemath/arduino-remote

When you get a certain level of academic inertia (tons of postdocs under you etc), you get author credit even when you don't actively participate in the research or writing itself. What's more interesting is how many first author papers are published per year, per author. First author is typically reserved for someone who did a significant amount of the work on the paper.

As a diabetes researcher, results like this don't surprise me one bit, but I'm glad they're getting national coverage! A publication in Nature is an incredible achievement, congrats to the teams involved.

We think good glucose control is at the center of a whole host of medical issues, inside and outside the hospital. Bringing critically ill patients under control has been found to reduce mortality rates by up to 30%, results which have been repeated several times in thousands of patients. These drugs, which are less intense than insulin therapy, confer similar glucoregulatory effects, which has all sorts of cascading hormone benefits etc.

Exciting stuff, thanks for sharing!

Whenever I want a lamba that's multiline, I usually just declare a nested function. Usually pretty clear, and reusable, and same scope rules as a lamba

And here I was thinking posting something similar would be read as too pessimistic!

I'm sure any released in the last 10 years would, definitely. But don't forget that there are probably tons of underfunded rural outfits that are using old tech/out of support software etc. Meeting in the middle is going to make sure they don't get left behind.

This is just a guess, but it's possible that surveyors are so used to measurements in ft that the existing tools mostly use ft (like the GPS in the video), so completely deprecating it might leave surveyors without the tools they need to do their jobs.

If you check out models/openai_model.py, you'll see it's using text-curie-001, among others, which is an OpenAI dataset. This model is imported in main.py under the name 'Model', which makes me believes the openAI API is serving as the main model used for inference

I recently converted some generation code to use a plan, so you can make a plan and then execute the plan. Architecting your code this way gives you things like dry runs and rollbacks for free; I feel like diff storage is in a similar realm of usefulness.

In my experience, the limitations imposed by the borrow checker actually make you express your ideas more clearly. In this way, you might be fighting the compiler, but when your code does compile, it usually does what you wanted it to (and is also designed in a much better way, since you're forced not to take the easy way to the finish line).

> This allows any sudoers user to obtain full root privileges

The way most sudoers files are set up, if you're in the wheel or sudo group, you're only a "sudo -i" from a root command prompt, so I'm not sure I see why this is a vulnerability. Can anyone elaborate?

It would be great if the kernel itself provided a header only definition of such a format, so you could focus on the data and not the parsing. Would also be able to integrate into their extensive testing infrastructure.

That's exactly how it should be done! Also subtle but important that you find the last closing parentheses, as an attacker could just include a paren in their process name to terminate your parse early.

There are certainly critics, but their opinions, if consistently opposed to the hype cycle, get either sidelined or lambasted. Just sort any post on some new ML application by controversial comments and you'll find the critics aplenty.

But how do you protect the shared secrets? Do you see my point? I wish there was some sort of industry best practice tool or book for modelling your threats and suggesting an architecture.

This is my problem with key management; at some point you have to have some sort of bespoke key management strategy for the key manager itself. I've yet to find a satisfactorily clear and modern way to accomplish this securely.

Using which key?

Organic Maps is a very good and low resource offline mapping app that includes trails, point to point elevation mapping, and very low storage footprint. All built on top of OpenStreetMap. Definitely recommend for camping/travelling etc where you might be out of service for days.