Ask HN: Seeking advice for the next steps of a health care side project

[JPLeRouzic]

Dear HNers, I am asking your advice for the next strategic steps about a half finished side project in health care.

Once a company asked for ideas about a COPD and Heart Failure early detector. I submitted my idea (based on Physionet 2016 competition), but the company then asked for a proof of concept. I had none, so I decided to make my own and make it public. I would be happy if it would be copied widely.

https://hackaday.io/project/19685-early-and-low-cost-detection-of-heart-failure

The PoC was developped in Java, uses ML and works well enough for me. I have a clear idea how to extend it to detect fibrous tissues in the heart and possibly in lungs. I have confidence that I can develop this, all by myself in the C language, embed it in a self contained $150 device and market it.

However several things worry me:

* Having a health detector is nice but is the information really useful?

* How to get a market authorization with little resources?

* What strategies about next steps, in particular I am thinking about a much more ambitious detector. Some guys, inspired by the Star Trek Tricorder and Qualcomm prize, already had wonderful ideas.

* A comfortable possibility would be to integrate an existing organization (I worked for 35 years in a large company).

Thanks for any suggestion.

[jwilliams]

I've known a few people go through this. One product was something that detected when someone hasn't moved in bed for a period of time. This can be used to prevent bedsores.

They found the regulatory environment absolutely crippling. After banging their head against the wall too long, they eventually gave up.

I'd say going with an experienced incubator/mentor/etc in this space would be a great idea. https://rockhealth.com/ is one good example. They can give you access to health care companies, networks, mentors, etc.

[JPLeRouzic → jwilliams]

Thank you very much for the feedback and the Rockhealth pointer.

[jwilliams → JPLeRouzic]

No problem! Good luck with the product.

[tixocloud]

I have a friend who's been working on a healthcare-related startup (smart baby mat). When I used to help out, I remember it being quite complex as he's gone to hospitals to help him perform some clinical trials before going to market. If you'd like, I can put you in touch with him.

[JPLeRouzic → tixocloud]

That is very nice, you can drop a message at:

https://padiracinnovation.org/feedback/

or connect through LinkedIn:

https://www.linkedin.com/in/jean-pierre-le-rouzic-39a72295/

Many thanks

[somid3]

Hi, I have a lot of commercialization background in this space - this is me - http://linkedin.com/in/somid3/ - shoot me a msg at somid3 at google's email service dot com

[JPLeRouzic → somid3]

Thanks for this reply, I really appreciate it.

[Jugurtha]

I've worked a bit on ECG signals (Physionet, too). You may want to take a look at:

- "Baseline Correction with Asymmetric Least Squares Smoothing" by Paul H. C. Eilers and Hans F.M. Boelens[0]. They used it for chromatography and spectroscopy but I've had great results applying that to detect and remove the baseline from ECG. The article provides 11 lines of MATLAB code to illustrate.

- Savitzky-Golay filter: For smoothing. I liked it for some of the noisy signals.

- Poincaré plot for RR intervals.

[0]: https://zanran_storage.s3.amazonaws.com/www.science.uva.nl/C...

[JPLeRouzic → Jugurtha]

Thanks for your detailled answer and those pointers, I appreciate this.

I am using phonocardiograms (PCG) not ECG but the processing is quite similar. In my case I record PCG with ultrasound Doppler but a mere microphone could do the job as well.

I think I went further than detecting the heart's rate and variability, as I try to recognise each heart sound (S1 ... S4) inside a beat and I developped a technique for having a beat signature in order to apply ML on the sequence of beats with HMM.

This beat signature technique could certainly be applied to ECG. It could probably even be used for any series of events.

However my problem is not technical, it is about raising awareness, getting help (including resources) and get strategic advising.

[mchannon]

Did the company that inspired you suddenly stop being interested in you, even if you had a proof of concept?

The profit motive means technology works as follows, particularly in health care: if you can build it and corner a healthy market, shut up and take my money. If you want to give it away or sell it as a loss, please leave a message at the beep.

Don't worry about building in all kinds of bells and whistles. Sounds like what little you have is more than good enough- call that company back and stop acting so profligate with the IP if you want them to throw the marketing and compliance resources at a project that desperately needs them.

[JPLeRouzic → mchannon]

Thanks for the nice feedback. I am more searching help than trying to make a profit.

[mchannon → JPLeRouzic]

You're welcome, but I don't think you got my point. The help you seek is one and the same with someone making a profit.

[JPLeRouzic → mchannon]

Ah good point, I will think about it!

[itamarst]

Keep in mind that selling something like this would quite likely involve an expensive FDA regulatory process. So probably not something you can do as a side project.

[JPLeRouzic → itamarst]

Thanks for the feedback! I guess it would be nice if I got some help from a large organization. Another possibility seems to not claim to be a medical device, or to be similar to an existing one which is well known. For example it seems stethoscopes are no more regulated.