Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

It is a critical problem if your entire record of life-saving drugs you've given them in the past 24 hours suddenly goes down. You have to start relying on people's memories, and it's made worse by shift turn-overs so the relevant information may not even be reachable once the previous shift has gone home.

There are plenty of drugs that can only be given in certain quantities over a certain period of time, and if you go beyond that, it makes the patient worse not better. Similarly there are plenty of bad drug interactions where whether you take a given course of action now is directly dependent on which drugs that patient has already been given. And of course you need to monitor the patient's progress over time to know if the treatments have been working and how to adjust them, so if you suddenly lose the record of all dosages given and all records of their vital signs, you've lost all the information you need to treat them well. Imagine being dropped off in the middle of nowhere, randomly, without a GPS.



That's why there's a sharpie in the first aid kit. If you're out of stuff to write on you can just write on the patient.

More seriously, we need better purpose build medical computing equipment, that runs on it's own OS, and only has outbound network connectivity for updating other systems.

I also think of things like the old school "check list boards" that used to be literally built into the yolk of the airplane they were made for.


I’m afraid the profitability calculation shifted it in favor of off-the-shelf OS a long time ago. I agree with you, though, that a general purpose OS has way too much crap that isn’t needed in a situation like this.


> That's why there's a sharpie in the first aid kit.

That doesn't help when the system goes down and you lose the record of all medications administered prior to having to switch over to the Sharpie.


> It is a critical problem if your entire record of life-saving drugs you've given them in the past 24 hours suddenly goes down.

Will outages like this motivate a backup paper process? The automated process should save enough information on paper so a switch over to paper process at any time is feasible. Similar to elections.


Maybe if all the profit seeking entities were removed from healthcare that money could instead go to the development of useful offline systems.

Maybe a handheld device for scanning in drugs or entering procedure information that stores the data locally which can then be synced with a larger device with more storage somewhere that is also 100% local and immutable which then can sync to online systems if that is needed.


And with their luck, those handheld devices will also be sent the OTA update that temporarily bricks them along with everything else.


no money for that

there are backup paper processes, but they start fresh when the systems go down

If it was printing paper in case of downtime 24/7, it would be massive wasteage for the 99% of time system is up


A good system is resilient. Paper process could take over when system is down. Form my understanding healthcare systems undergo recurrent outages for various reasons.


Many place did revert back to paper processes. But, it’s a disaster model that has to tested to make sure everyone can still function when your EMR goes down. Situations like this just reinforce that you can’t plan for if IT systems go down, it is when they go down.


My experience with internet outages affecting retail is the ability to rapidly and accurately calculate bill totals and change is not practiced much anymore. Not helped by things like 9.075 % tax rates to be sure.


How about an e-ink display for each patient that gets drug and administration info displayed on it?


Real paper is probably as much about breaking from the "IT culture" as it's about the physical properties. E-ink display would probably help with power outage, but happily display BSOD in an incident like this.


Honestly if you were designing a system to be resilient to events like this one, the focus would be on distributed data and local communication. The exact sort of things that have become basically dirty words in this SaaS future we are in. Every PC in the building, including the ones tethered to equipment, is presently basically a dumb terminal, dependent on cloud servers like Epic, meaning WAN connection is a single point of failure (I assume that a hospital hopefully has a credible backup ISP though?) and same for the Epic servers.

If medical data were synced to the cloud but also stored on the endpoint devices and local servers, you’d have more redundancy. Obviously much more complexity to it but that’s what it would take. Epic as single source of truth means everyone is screwed when it is down. This is the trade off that’s been made.


> synced to the cloud but also stored on the endpoint devices and local servers

That's a recipe for a different kind of disaster. I actually used Google Keep some years ago for medical data at home — counted pills nightly, so mom could either ask me or check on her phone if she forgot to take one. Most of the time it worked fine, but the failure modes were fascinating. When it suddenly showed data from half a year ago, I gave up and switched to paper.




Consider applying for YC's Summer 2026 batch! Applications are open till May 4

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

Search: