Idea 1

A project log for Non-invasive continuous glucose estimator

I hope to create an arduino based CGM to track trends between meals

BethanyBethany 01/12/2018 at 01:564 Comments

I stumbled on an article titled " The Non-Invasive Glucose Monitor No One Wanted (?)" back in late August 2016 (my senior year and almost cried thinking that exactly what I want .

I was even more impressed by the patent  

I sent Larry Ishler a linked in Friend request which he accepted in September 2016, and I was super excited only to have no reaction to any of my messages to date.

Thus to avoid violating patent laws I can't "make", "use", or "sell" and may have to delete this blog eventually but I am currently doing my best to be working under the research exemption (given it is a medical product idea that was not actualized and marketed) .

I begin with this as a first foundation because it seems simple, beautiful, and inexpensive.

     simple concept: temperature differential in the ear--> bg.  

beautiful:  it could resemble earbud headphones wired to a medium sized hip mounted device

inexpensive:  (assumes arduino Uno kit ) needs 2 thermistors and a way to hold them in the ear

Future alternatives and upgrading options will likely come from other patents discussed within "

The Pursuit of Noninvasive Glucose: “Hunting the Deceitful Turkey”,%20Fourth%20Edition.pdf

Because it is a great compilation of great ideas.

I hope to potentially contribute to that field and yet remain unique in not striving for replacing but merely supplementing glucose data for those kids (like younger me) who will be able to see how implementing tedious endocrinologist recommendations (doing diabetes right) makes better graphs, encouraging from the beginning better health for better life, and adults like me to continue managing the chronic condition to the best of our abilities daily too.


falk wrote 01/12/2018 at 08:25 point

Just so i can get it out of my head and back to work: in your schematic you use the two thermistors with 2 adc channels - since you are interested in the temperature differential you can get more accuracy by using the 2 thermistors combined in a voltage divider configuration with a single adc tuned with resistors so your interesting temperature range is amplified a bit (I assume 10 to 40°).


Actually since this will never go to 0V(unless you have a source of relative negative voltage) you can reduce it to this:


the 10-bit ADC of the Atmel chip resolution should be plenty with this and you only have to worry about thermistor consistency.

anyway back to work for me ;)

  Are you sure? yes | no

falk wrote 01/12/2018 at 07:10 point

hmm, this is interesting if you can get temperature sensors of the required accuracy. I assume this works by measuring the relative metabolic rate via the temperature from the skin from a place relative "inside" the body to a place more outside(thus loosing more temperature to the air depending on metabolic rate).

if that is right I have a few suggestions. The first is - does it have to be that particular place?

In your diagram i see an earring - wouldn't it be just  as accurate to measure the temperature inside the earring hole? This would only be practical if you could make it smaller luckily this is were my second suggestion comes in.

Well you see it should be entirely possible to even with "at home" tool to make this far smaller ;)

A few days earlier and you could have entered this into the coin-cell challenge. I made led-greeting cards this year for friends and family that used basically the same setup electrically.

You take a small micro say an attiny48. Its like a smaller version of the Atmeg328 in your UNO and can be Programmed with the same UI. The Picture on the linked page shows the *biggest* package you can get it in. You directly attach to that your Thermistors (it has 4 x 10 bit ADC channels) and a Coincell for power. The whole package should be below 10gram or so of weight (I calculate well below 5gramm for this based on the datasheets i looked at)

Size-wise it would be maybe 3x the diameter of the earring in the picture. Log capacity on the chip would be a couple of days with moderate compression of the data. Run time would depend on coin cell size but with a 3gramm cr2032 it would be months to years :P. cost if mass produced would be well below 1(USD/EUR).

You would read the data with a clothespin like device.

  Are you sure? yes | no

Bethany wrote 01/13/2018 at 04:16 point

Yes, I think it is an oversimplification of the concept of "relative metabolic rate" based on heat transfer (Q=m*Cp*ΔT )

"does it have to be that particular place? " I don't know, but I assume dozens of pairs of locations could be calibrated (with varying degrees of accuracy).

Actually I thought about that, making a DIY platinum wire RTD earrings before realizing how complex that could become, in addition to remembering how my sensitive skin + certain metals can get easily irritated and inflamed around piercings that would give bad data.

I honestly don't (yet) have the electrical skills to design that myself and not the skill capacity to focus simultaneously on the system's accuracy and miniaturization. So cool as that is, the bulky plug and play arrangement  to a desktop device with wires to a simple ear sensor holding device would be not unlike earbud to my computer right now.

 When it works a tiny duplication would be amazing, so I will keep that in mind

  Are you sure? yes | no

falk wrote 01/13/2018 at 07:28 point

Ah, I had not considered the skin irritation factor. I have been thinking about putting the through-hole thermistor inside a tube for stability- some thermistor types already come in that form factor actually.

Any thoughts on the differential thermistor idea in my other comments?

  Are you sure? yes | no