3 ms·

Well, yeah. If the purpose of your biomedical device is to measure x or y or z so the patient can make a certain decision, you need to be able to prove that you

Well, yeah. If the purpose of your biomedical device is to measure x or y or z so the patient can make a certain decision, you need to be able to prove that your device can provide accurate enough data for the patient to make that decision. If the purpose of your device is to provide alerts, you need to prove that it your device actually alerts the patient at the right time. To do this you need accurate continuous measurements, and a proven system that alerts the user based on that data. This is extremely difficult. I work for Dexcom, the only company that makes an FDA approved glucose monitor that can both fire alerts, and is approved for use by children under 12. We started with a very small team and it took a solid decade before we got FDA approval, but FDA approval is basically the highest bar for a medical device. So much so that many nations automatically greenlight any device with FDA approval. If you've built something, and you can't prove with documentation that the specific device you're selling a customer will work exactly as advertised, before they take it out of the box, you can always sell your probably-works device to another nation (or nations) first until you get your ducks in a row for the FDA.

bsder 5y ago
True. But in the case I cited (the doctor was actually asking our company to do the backend transport for the device), the device works. It will bring quality of life improvements. However, as you well know, nothing is ever 100%. Even if the detector was 100% (and nothing ever will be), the network carrying those alerts is most certainly not 100% (all the cellular carriers in the US have specific language barring medical usage because they know they're terrible). And when, not if, some notification doesn't go through you're on the receiving end of a lawsuit. So, the problem is that if the device doesn't obviously make enough profit to cover the legal settlement cost from the case where it fails, it's dead in the water. Indemnification tips that balance a bit to allow devices that might be a net win but it will be difficult to prove that a priori. It's such a problem that I have a black-humor question to everybody who thinks they want to do a biotechnology device: "That's a great idea! So, how many people are you allowed to kill?" If they get over the shock of the question, I start doing back of the envelope calculations for how many people the device will kill. If they're still with me after that, we can have a real discussion about whether their device might be feasible.

bsder 5y ago

True. But in the case I cited (the doctor was actually asking our company to do the backend transport for the device), the device works. It will bring quality of life improvements. However, as you well know, nothing is ever 100%. Even if the detector was 100% (and nothing ever will be), the network carrying those alerts is most certainly not 100% (all the cellular carriers in the US have specific language barring medical usage because they know they're terrible). And when, not if, some notification doesn't go through you're on the receiving end of a lawsuit. So, the problem is that if the device doesn't obviously make enough profit to cover the legal settlement cost from the case where it fails, it's dead in the water. Indemnification tips that balance a bit to allow devices that might be a net win but it will be difficult to prove that a priori. It's such a problem that I have a black-humor question to everybody who thinks they want to do a biotechnology device: "That's a great idea! So, how many people are you allowed to kill?" If they get over the shock of the question, I start doing back of the envelope calculations for how many people the device will kill. If they're still with me after that, we can have a real discussion about whether their device might be feasible.