Tellspec: amazing IR spectrometry advance or scam?

[cosmicaug]

So on another web forum I found this product, Tellspec, with an IndieGogo crowd funding campaign. Their objective seems to be a keychain sized product which you wave over food like a magic wand and which will scan it to produce a nutrient breakdown as well as check for particular components.

Nothing that I know about instrumentation would suggest to me that this is currently possible but there is much I do not know (I'm so far from being an instrumentation guy it's not even funny) but, in principle, a lot of this stuff is doable. I repeat, in principle.

Apparently, light spectrometry on a chip exists at least as a product in development (the separation being done by some sort of resonance effect). However, they claim to be doing Raman spectrometry at http://tellspec.com/howitworks/ :

Quote:

The handheld scanner is a Raman spectrometer. The low-powered laser inside the scanner emits a beam through the front window. Light emitted from the sample is then collected through a filter in the same window. This light then passes through a diffraction grating that disperses the light onto an image sensor, which converts it into an electrical signal that is then digitized.

Maybe there are chips out there which can do this. I have no idea, quite frankly (though my spidey sense is definitely tickling on this product concept as a whole). The first hit Googling that talked about integrating this sort of analysis on silicon was from a laser manufacturer and talks about why a particular research team chose their laser. One of the stated reasons is "High power (> 2 W @ 785 nm & > 500 mW for entire bandwidth) that is required for Raman spectroscopy which is inherently a very weak process". And yet their product claims to do the same thing with a laser powered by a rechargeable battery in a device that you recharge from a USB port and which is smaller than most computer mice. And you are, supposedly using this device in uncontrolled environments.

And let's look at their team:

This is one founder, Dr. Isabel Hoffmann, who apparently co-founded anti-aging clinics in Beverly Hills. Can you say "potential red flag"? I knew you could!

The other founder is a mathematician which sounds legitimate. Strangely, his most prominent and recent experience seems to be in a multimedia company fighting intellectual property battles.

Then we have:

• Martin Merener - Mathematician (looks like a buddy from York University. Nothing wrong with that and mathematicians make sense as this would require heavy lifting on the analysis side. fMRI data analysis even seems potentially relevant if you need to deconvolute the messy data that this device would produce)
• Maureen Novak - food photographer
• Margaret McClintock - writer and editor
• Chao Gao - industrial designer (the guy who made the mock ups, maybe?)
• Gisele Waters -customer experience analyst (WTF? What customers? They don't even have a product yet!)
• Xue Feng -graphic designer (helped out with the videos and website, I'm guessing)
• Sarah Ong - food scientist (undergraduate degree --her graduate degree is an MBA)
• Pranay Gupta - (Indian fellow. Looks hopeful. Nope, another marketing drone )

So mostly marketing drones and not a single technologist??? Maybe that's what they need the money for, to hire the technologists who can make the product?

Besides, do you really mean to tell me that this magic wand you haphazardly wave in the general direction of food can isolate and tell apart individual proteins (their "demo" shows the device identifying gluten) in their analysis and yet the site doesn't list a single reference?

The level of signal resolution implied is beyond the ridiculous. Unless they are using technology secretly recovered from the Roswell crash site, I just can't buy it.

I call scam!

[catsmate]

I'd go with scam.

[Evilgiraffe]

Raman spectroscopy in a lit room? Scam.

Raman scatter is ridiculously weak. So weak that whenever I've done it, you need a fairly powerful laser in a darkened room and plenty of time to get a spectrum that looks halfway decent.

[Cylinder]

Originally Posted by Evilgiraffe

Raman spectroscopy...

I spent much of my college years wondering what was actually in those delicious mystery noodles. This is Nobel territory, in my layperson opinion.

[ben m]

A few other things. First, Raman spectra spread out information over a fairly small band of wavelengths---say 50nm. That means you need, not just any cheap spectrometer, but a very-high-dispersion spectrometer, to meaningfully separate these wavelengths. Now, the fiber-resonator "spectrometer on a chip" that Cosmicaug links to---neato!---is indeed high-resolution. But that's not what Tellspec says they're doing. They say they're sending light through a diffraction grating and onto a CCD. Sorry, kids, you're not getting sub-nm resolution with a spectrometer that small. Not even close.

If you're not working at high resolution, then you're going to be looking at overlapping spectra from everything in the system. That's fine if you have only a few components and need to distinguish them coarsely. I can maybe imagine a handheld Raman spectrometer if its only job were, say, distinguishing water and vodka. (And it wouldn't look like this.) But that's not what Tellspec says they're doing.

In their example images, they show the concept-device pointed at a truffle and "detecting" vanillin. I've made truffles, fellas. Looking up a recipe, the proportions are something like 350g chocolate, 100g cream, and 10g (2 tsp) vanilla extract. How much vanillin is in vanilla extract? According to Wikipedia, 1000g of extract might draw from 100g of beans, and beans are 2% vanillin. So, they're claiming that this handheld spectrometer can detect 20mg of vanillin on a background of 450g of mixed other stuff! That's 40ppm detection!

I'll be pretty blunt about this. This is impossible. I'd be cautious in interpreting a 40ppm trace-component detection from a high-end laboratory spectrometer. From a cheap one? Casually aimed at a sample? No way. No way at all. Even if the laser, the dark field, the background subtraction, etc. were chosen optimally, the sort of camera you can put on a chip like this only has an 8-12 bit ADC to begin with.

And ALL of their product concepts look like this. Tartrazine (a dye!) can't be more than parts-per-ten-thousand. "Essential vitamins and minerals"---geez, Vitamin C in orange juice is already 0.05%, other vitamins are below that. They mention monitoring mercury intake---good lord, meaningful foodborne mercury detection starts at part-per-million and goes down from there.

This is unambiguous, people. These people have no clue whatsoever whether this product is possible. Someone on this team read about Raman spectroscopy, read that it can serve as a chemical fingerprint, and guessed that (a) it could be miniaturized---because, hey, everything can be miniaturized these days, right? and (b) "algorithms" could process complex spectra into ingredients-lists---because, hey, big data, algorithms, that's how it works, right? They drew pictures of a concept product based on this guesswork, not based on any actual physics, spectra, spectrometer technology, or knowledge of food.

If they get enough funding to attempt to build the product, they will find that it's impossible.

This is like what happens when an entrepreneur watches CSI for product ideas. "Hey, they have software that can enhance grainy surveillance-camera footage to 10-megapixel quality! I could sell that, lemme raise some money and I'll hire some developers. Hey, they have a portable MRI machine! That's a great idea---an MRI machine is, like, just a magnet and a radio transmitter? That'd totally fit in an iPhone these days, that'd be so useful. I'll start raising money. Hey, on Num3ers they created an algorithm that can identify anyone by listening for their unique footstep frequency in the background of other people's phone calls. That's marketable. I'll schedule an IPO for tomorrow and hire a podiatrist and a mathematician the next day."

[jsfisher]

I'm still curious how they can get so much information about the apple by just scanning its skin. Must be the same technology that lets the device deduce the content of a hamburger from a scan of the onion and tomato on top.

Its all so sciency.

[pgwenthold]

Originally Posted by ben m

In their example images, they show the concept-device pointed at a truffle and "detecting" vanillin. I've made truffles, fellas. Looking up a recipe, the proportions are something like 350g chocolate, 100g cream, and 10g (2 tsp) vanilla extract. How much vanillin is in vanilla extract? According to Wikipedia, 1000g of extract might draw from 100g of beans, and beans are 2% vanillin. So, they're claiming that this handheld spectrometer can detect 20mg of vanillin on a background of 450g of mixed other stuff! That's 40ppm detection!

It's not the concentration that is the question here. The problem is the specificity. Condensed phase vibrational spectroscopy is extremely low resolution due to intermolecular forces. It's great for functionality, but not for detailed structure. You measure something that has vanillin in it, and you will see OH stretch at 3500 and you will see carbonyl stretch at 1700 and a phenolic peak at 1600. And that's great. But you look at a cookie, you are going to see a huge OH stretches and carbonyl stretch all over the place due to the carbs, the fats, etc. Yeah, you might be able to pull out a peak for a nitrile or alkyne, but other than that, you aren't going to have any specific peaks that you can assign.

And that is just if you are looking vanillin, much less "assess the nutritional content."

Lastly, this has nothing to do with the instrument. You couldn't do this measurement with an actual Raman instrument. You aren't going to do it miniature, either.

[ben m]

Wait a minute, here's something from their Indiegogo campaign:

"Over the past nine months, we've come a long way in the development of TellSpec. We've developed the analysis engine, coded it, and tested it; we've successfully tested three prototype scanners, including one using a nanochip; we've finished the industrial design for the scanner; we're well along with the user interface design; and we've been doing an independent validation of the underlying algorithm. "

Wow, seriously? This moves my opinion from "these people are incompetent" to "this is a scam". They claim they've tested a "nanochip" already---with what manpower, exactly? A spectrometer-on-a-chip is not, AFAIK, commercially available anywhere on Earth---and developing one is generally cutting edge, multimillion dollar R&D job. How did YOU, a not-really-funded Indiegogo startup, secretly develop a high-resolution spectrometer-on-a-chip, in a world where well-funded university groups (like these people: http://ieeexplore.ieee.org/stamp/sta...umber=06472720 ) are still inching towards low resolution? And how did you do so with not a single experimentalist on your team?

[ben m]

Originally Posted by pgwenthold

Lastly, this has nothing to do with the instrument. You couldn't do this measurement with an actual Raman instrument. You aren't going to do it miniature, either.

With an actual Raman instrument, if you had two identically-prepared truffles, one with 1% vanillin and one with none, I bet you could subtract the spectra and tell which was which. That's about as far as I would take it.

[ben m]

Originally Posted by ben m

Wait a minute, here's something from their Indiegogo campaign:

"Over the past nine months, we've come a long way in the development of TellSpec. We've developed the analysis engine, coded it, and tested it; we've successfully tested three prototype scanners, including one using a nanochip; we've finished the industrial design for the scanner; we're well along with the user interface design; and we've been doing an independent validation of the underlying algorithm. "

Wow, seriously? This moves my opinion from "these people are incompetent" to "this is a scam". They claim they've tested a "nanochip" already---with what manpower, exactly? A spectrometer-on-a-chip is not, AFAIK, commercially available anywhere on Earth---and developing one is generally cutting edge, multimillion dollar R&D job. How did YOU, a not-really-funded Indiegogo startup, secretly develop a high-resolution spectrometer-on-a-chip, in a world where well-funded university groups (like these people: http://ieeexplore.ieee.org/stamp/sta...umber=06472720 ) are still inching towards low resolution? And how did you do so with not a single experimentalist on your team?

It's possible that when they say "we tested a scanner", they mean "we pointed a laser at some food". Hey, the laser hit the food! They don't say that they acquired Raman spectra with a prototype scanner.

[Hellbound]

Originally Posted by ben m

How did YOU, a not-really-funded Indiegogo startup, secretly develop a high-resolution spectrometer-on-a-chip, in a world where well-funded university groups (like these people: http://ieeexplore.ieee.org/stamp/sta...umber=06472720 ) are still inching towards low resolution? And how did you do so with not a single experimentalist on your team?

Homeopathic Engineering.

They taped a printed-out picture of a real chip to the back of an iphone.

[pgwenthold]

Originally Posted by ben m

With an actual Raman instrument, if you had two identically-prepared truffles, one with 1% vanillin and one with none, I bet you could subtract the spectra and tell which was which.

Possibly. However, the problem with that approach is that you have to know that one has vanillin in it and the other doesn't. Assuming that they were otherwise identical, of course.

Unfortunately, that's not a very useful analysis approach, to help you examine unknowns.

[marting]

Originally Posted by ben m

Wait a minute, here's something from their Indiegogo campaign:

"Over the past nine months, we've come a long way in the development of TellSpec. We've developed the analysis engine, coded it, and tested it; we've successfully tested three prototype scanners, including one using a nanochip; we've finished the industrial design for the scanner; we're well along with the user interface design; and we've been doing an independent validation of the underlying algorithm. "

Wow, seriously? This moves my opinion from "these people are incompetent" to "this is a scam". They claim they've tested a "nanochip" already---with what manpower, exactly? A spectrometer-on-a-chip is not, AFAIK, commercially available anywhere on Earth---and developing one is generally cutting edge, multimillion dollar R&D job. How did YOU, a not-really-funded Indiegogo startup, secretly develop a high-resolution spectrometer-on-a-chip, in a world where well-funded university groups (like these people: http://ieeexplore.ieee.org/stamp/sta...umber=06472720 ) are still inching towards low resolution? And how did you do so with not a single experimentalist on your team?

Yep. Seems like a reasonable take.

[ben m]

Good lord, I watched the Indiegogo video. jsfisher, I thought you were kidding about the apple. Nope, they really do that---scan an apple ("this apple has no pesticides", says the salesman, another feat of sub-part-per-million detection!)---so now they wave the device over a mixed pasta plate, point it at a piece of sushi. They're operating anywhere between 1-10cm away from the food. Where's the focal point of your spectrometer, then?

There's a neat way you could fake the product demo, if you cared to. Instead of a laser spectrometer, make a small camera. The user "scans" their food, but they're actually just photographing it. The "scanner" uploads the photo to a server, which guesses what the scanned substance is and replies with a vaguely-plausible "nutrition label" for that type of foodstuff. You could put on a pretty convincing stage show that way---instead of a server, you have a human confederate who can type fast. "Try it, young man, try scanning that candy bar you're holding. OK, the data is uploading, wait a moment ... just another moment ... <meanwhile the confederate has seen the photo of a Mr. Goodbar, and is retrieving its nutrition facts from a database previously queued up.> ... here we are, the sensor detects sugar and chocolate and peanuts."

[jsfisher]

By the way, Indiegogo and Kickstarter behave completely differently if the funding goal is not met. Care to guess which returns all monies to the backers in the case of falling short?

[Horatius]

This is the sort of thing that might cross my desk at some point....Bookmarked!

[uvar]

Originally Posted by jsfisher

By the way, Indiegogo and Kickstarter behave completely differently if the funding goal is not met. Care to guess which returns all monies to the backers in the case of falling short?

You can have fixed funding campaigns on IG (where, like Kickstarter, money only goes if they reach the goal), it's a choice when you make the campaign. There are still benefits if you get to the goal on flexible funding, for instance I think IG takes a smaller share of the funds.



Anyway, they do claim to have tested something (from the FAQ):

Quote:

In our tests, the analysis engine identifies the food correctly about 97.7% of the time, and detects gluten correctly 99.8% of the time. We expect further improvements.

And, from the IG updates:

Quote:

Our U.S. patent application 13/958,909 entitled "Analyzing And Correlating Spectra, Identifying Samples And Their Ingredients, And Displaying Related Personalized Information" was filed on August 5, 2013.

I haven't been able to find that one (though maybe I messed up or the shutdown is to blame?). I did find this though, which sounds broadly similar to somebody like me who knows nothing about Raman spectroscopy, is designed to analyse drugs in solid/powdered/liquid form in 10 to 15 seconds, and (see paragraph [0010]) costs about $5000.

I'm optimistic, so here's my guess: It doesn't analyse the food directly; it just 'scans', receives something (that may or may not resemble the real spectrum), looks in the 'learning algorithm' database and returns information for the best match. It might even work okay in the lab with consistent conditions, specific foods and a small number of unique results to return. It's not technically a scam, but it's not a proper spectrometer either and will probably be ridiculously unreliable anyway.

[ben m]

Apparently handheld Raman analyzers are common-ish---there are a dozen on the market---for the police to identify drugs. Here's a performance evaluation:

http://www.nfstc.org/?dl_id=214

Some features of these devices: 100-300+ mW lasers (compare claim of 5mW for the Tellspec) , thermoelectrically-cooled CCDs to keep the noise levels down (no room or power for that in the Tellspec), sample pressed against an input window or at known focal length---one product I looked at had an adjustment knob to alter the focus past thicker or thinner plastic bags (unlike the Tellspec claim of waving a wand a random distance past food), frequent calibration checks against a standard (Tellspec shows no such thing) and decent performance on pure substances but mediocre/poor performance on mixtures (pasta salad, for example) and no trace-component detection whatsoever (unlike half of the Tellspec claims).

[jsfisher]

Originally Posted by ben m

Apparently handheld Raman analyzers are common-ish---there are a dozen on the market---for the police to identify drugs. Here's a performance evaluation:

http://www.nfstc.org/?dl_id=214

Some features of these devices: 100-300+ mW lasers (compare claim of 5mW for the Tellspec) , thermoelectrically-cooled CCDs to keep the noise levels down (no room or power for that in the Tellspec), sample pressed against an input window or at known focal length---one product I looked at had an adjustment knob to alter the focus past thicker or thinner plastic bags (unlike the Tellspec claim of waving a wand a random distance past food), frequent calibration checks against a standard (Tellspec shows no such thing) and decent performance on pure substances but mediocre/poor performance on mixtures (pasta salad, for example) and no trace-component detection whatsoever (unlike half of the Tellspec claims).

100+ mW lasers, eh? Yeah, would be safe in the hands of your typical nutrition-conscious consumer.

Luckly for Tellspec, they've perfected using 5 mW red lasers at random distance from the target without the need for calibration, so safety concerns are a non-issue. Heck, they can even scan through cellophane and other food packaging without compromising accuracy.

[catsmate]

Originally Posted by jsfisher

100+ mW lasers, eh? Yeah, would be safe in the hands of your typical nutrition-conscious consumer.

Luckly for Tellspec, they've perfected using 5 mW red lasers at random distance from the target without the need for calibration, so safety concerns are a non-issue. Heck, they can even scan through cellophane and other food packaging without compromising accuracy.

I assume the 5mW power output is due to the US restrictions on higher powered lasers?

[jsfisher]

Originally Posted by catsmate1

I assume the 5mW power output is due to the US restrictions on higher powered lasers?

Well, the restrictions relate to an eye safety concern. An appropriate warning label for most lasers would be "Don't stare directly into the beam with your remaining good eye." The blink reaction will protect you from accidental eye exposure up to around 100 mW for nominal beam diameters. Above that, though, permanent damage becomes increasingly likely from the briefest flash.

Plus 5 mW laser diodes are dirt cheap and give you reasonable battery life.

[cosmicaug]

Originally Posted by ben m

In their example images, they show the concept-device pointed at a truffle and "detecting" vanillin. I've made truffles, fellas. Looking up a recipe, the proportions are something like 350g chocolate, 100g cream, and 10g (2 tsp) vanilla extract. How much vanillin is in vanilla extract? According to Wikipedia, 1000g of extract might draw from 100g of beans, and beans are 2% vanillin. So, they're claiming that this handheld spectrometer can detect 20mg of vanillin on a background of 450g of mixed other stuff! That's 40ppm detection!

Except they don't even claim to be doing that in the promo video. Instead they claim to be able to detect artificial vanilla. Given that the vanillin molecule is the same regardless of whether it is from a vanilla pod or from a chemical manufacturer, that is impossible. If we want to be charitable, we could conjecture that there might be other ways of making this determination and that though they are saying one thing in the video it is really a shortcut for a longer explanation. So if we want to give them the benefit of the doubt one could speculate that the signature of natural vanillin is that it always appears with a set of other compounds so when they say they have detected artificial vanillin what they really mean is that they have found vanillin not accompanied by other compounds characteristic of the natural extracts.

Of course, most such accompanying compounds would also likely be expected to be found many other foods both as expected components parts of natural ingredients as well as as additives in their own right. Figuring out which compounds to focus on and what would be the expected ratios that they would be found in would require extensive validation and we can be pretty sure that they haven't done this (this is without even going to the even more basic validation which would be involved to characterize the different sources of natural vanilla extracts).

This is not to say that, assuming the analytical capabilities they claim (which I do not believe), they couldn't train their system recognize a known piece of chocolate with known artificial vanillin content. But I would not believe that they'd be able to go from this to a complete unknown food item (whether it be a piece of chocolate or otherwise). And besides, if what they were going to be doing was a database of known foods, they'd be better of making their device into a UPC & QR code scanner (like you suggest later) than pretending that it does some bulldoodoo spectroscopy that wouldn't work even in a laboratory setting with full powered instruments.

Originally Posted by ben m

This is unambiguous, people. These people have no clue whatsoever whether this product is possible. Someone on this team read about Raman spectroscopy, read that it can serve as a chemical fingerprint, and guessed that (a) it could be miniaturized---because, hey, everything can be miniaturized these days, right? and (b) "algorithms" could process complex spectra into ingredients-lists---because, hey, big data, algorithms, that's how it works, right? They drew pictures of a concept product based on this guesswork, not based on any actual physics, spectra, spectrometer technology, or knowledge of food.

I like this interpretation. I think I should have made room in my mind for self-delusion instead of immediately jumping to the conclusion of scam. I have run into the mention of someone online which made me realize that the notion of being incredibly fantasy prone must be considered at times. However, the fact that they chose a flexible funding model (all pledged funds are collected even if they do not reach their funding goal) might be seen by some as evidence pointing against the incompetence & self-delusion hypothesis (as well as the involvement of someone with fMRI data analysis experience --since I'd expect such a person to be too well informed to self-delude themselves in this matter).

[cosmicaug]

New video at http://youtu.be/oLrX4J3gXZk .

[jsfisher]

Apparently, we may soon be visited by Dr. Stephen Watson. In anticipation, I predict the following:

• The Tellspec hand-held device is envisioned to be little more than a laser, defraction grating, CCD array, and a blue-tooth transceiver.
• The company founders have high-hopes without foundation that their simple design will give them enough food "fingerprint" information to identify the product from a cross-match with their extensive database.
• The promotional video is a testament to their naivete or their intent to defraud.
• The device, if it is able to perform anything at all resembling a Raman scan, will be very noisy and unreliable.
• The back-end server guesses much of the time, and not very well.
• Assuming there is eventually a product, the company will be sued out of existence by the first person who has an adverse reaction to food incorrectly identified by Tellspec.

[John Jones]

Originally Posted by cosmicaug

Except they don't even claim to be doing that in the promo video. Instead they claim to be able to detect artificial vanilla. Given that the vanillin molecule is the same regardless of whether it is from a vanilla pod or from a chemical manufacturer, that is impossible. [...]

Ethyl vanillin is synthetic, and has a stronger vanilla aroma/flavor than vanillin. Perhaps they are alluding to adulteration. There is a lot of money being made/stolen in adulterated natural products.

[Olowkow]

The Tellspec device does appear likely bogus to me, mainly from way it is presented in the video. I agree that Raman spectroscopy is a tad more involved than their magic wand would suggest. I was reminded of the "Amega Wand" when I saw them waving the thing. The ultimate device is still the "tricorder", and I suppose that's what some will be thinking this thing is purporting to be.

The diode array spec was a big step towards miniaturization and increased scan speed, though the first ones were not exactly small.

But a miniaturized spectrometer is far from fiction. A colleague of mine told me about spectrometer, IIRC from Ocean Optics, which uses fiber optics and just plugs into a USB port. I saw something similar a few years ago. And as far as a spectrometer on a chip goes, there are many articles about various products under development with very clever monochromators on a nano scale. There is even a mass spec with single molecule sensitivity under development. I recently attended a lecture about research which involves unzipping a single DNA molecule with laser tweezers, and measuring the force as each bond separates.

Incredible research going on, but the device in question just rings that old familiar bell of bogosity.

[ben m]

Originally Posted by Olowkow

But a miniaturized spectrometer is far from fiction. A colleague of mine told me about spectrometer, IIRC from Ocean Optics, which uses fiber optics and just plugs into a USB port.

Yep, I've used that spectrometer. Very nice device. Fairly ordinary CCD-grade sensitivity.

Quote:

I saw something similar a few years ago. And as far as a spectrometer on a chip goes, there are many articles about various products under development with very clever monochromators on a nano scale. There is even a mass spec with single molecule sensitivity under development. I recently attended a lecture about research which involves unzipping a single DNA molecule with laser tweezers, and measuring the force as each bond separates.

Yes, absolutely. I would not be so skeptical of, say, the announcement of a micro-mass-spec (or ion-mobility-spec) "nose on a chip" that you wave over your food which detects single odor molecules.

The question to ask is what do you gain by minuaturization? What was going on in all the extra space/mass/area of the non-miniature device you're replacing? In the case of (say) atom trap mass spec, the extra space in a big spectrometer is "indifferent"; there's a lot of empty vacuum chamber between your ions and your electrodes, or whatever, that you could get rid of if you had the right design/fabrication. In the case of an ion-mobility spectrometer or a GC, extra space might be a source of noise---forcing you to dilute your analyte over a larger-area detector or something. In the case of a Raman spectrometer, the volume of a large instrument isn't wasted: the size of an optical signal (photon collecting power) is proportional to the size of the collecting lens. A big lens, a big spectrometer aperture, a big grating, are how you collect photons. By miniaturizing your device, you're straight-up miniaturizing your signal. Why would you do that? There are two possible reasons:

1) Maybe you're also miniaturizing your noise---like, by replacing a large-area detector with a small-area detector? Nope. Good full-sized CCDs are very nearly single-photon counting devices already. Cheap CCDs (or CMOS detectors) are noisier than big expensive ones.

2) Maybe you're concentrating the power of your tiny detectors on a smaller sample, and maybe that's good? Not in this case: the size of the sample probed is just the focal spot size of the illuminating optic---in this spectrometer, with a tiny lens and long focus, they've made the focal spot HUGE.

3) Maybe you get some miraculous parallelization---instead of running a process once in a big instrument, you run it 1000 times on 1000 nanochips, which all together are still smaller than the big thing and whose results average together somehow. Not relevant here.

Miniaturization is great, but it doesn't erase the laws of physics. Some things are better when big, and "spectrometer light-collecting optics" are one of them.

So yeah. Miniaturization is great and often makes for better, more sensitive instruments ... but not in this case. They're building, at best, an ultra-low-sensitivity instrument. If they were claiming to tell pasta from chocolate, I'd give them the benefit of the doubt. If they claimed that their fancy fingerprint algorithms had made a large gain in specificity on complex mixtures, I'd want to see the paper. It's really the ultra-trace claims (mercury, pesticides, flavors, preservatives) that makes it clear that marketers, not spectroscopists, are running this particular clown show.

[Olowkow]

Originally Posted by ben m

Yep, I've used that spectrometer. Very nice device. Fairly ordinary CCD-grade sensitivity.

Yes, absolutely. I would not be so skeptical of, say, the announcement of a micro-mass-spec (or ion-mobility-spec) "nose on a chip" that you wave over your food which detects single odor molecules.

Or the "optoelectronic nose." Already done.

Good points about miniaturization. Unlikely we will see a backyard telescope that exceeds the 200 inch Palomar any time soon, or ever.

I was involved in several projects, DARPA and others, that aimed to shrink various detection devices for military purposes, mainly nerve gas and thiol detection. Thinking outside the box is pretty much standard procedure.

[cosmicaug]

Originally Posted by cosmicaug

New video at http://youtu.be/oLrX4J3gXZk .

My reading on that video was "but, but,... we've got magic algorithms, I swear!". This makes Ben m's earlier comment apropos:

Originally Posted by ben m

This is like what happens when an entrepreneur watches CSI for product ideas. "Hey, they have software that can enhance grainy surveillance-camera footage to 10-megapixel quality! I could sell that, lemme raise some money and I'll hire some developers. Hey, they have a portable MRI machine! That's a great idea---an MRI machine is, like, just a magnet and a radio transmitter? That'd totally fit in an iPhone these days, that'd be so useful. I'll start raising money. Hey, on Num3ers they created an algorithm that can identify anyone by listening for their unique footstep frequency in the background of other people's phone calls. That's marketable. I'll schedule an IPO for tomorrow and hire a podiatrist and a mathematician the next day."

[CapelDodger]

Originally Posted by ben m

This is unambiguous, people. These people have no clue whatsoever whether this product is possible. Someone on this team read about Raman spectroscopy, read that it can serve as a chemical fingerprint, and guessed that (a) it could be miniaturized---because, hey, everything can be miniaturized these days, right? and (b) "algorithms" could process complex spectra into ingredients-lists---because, hey, big data, algorithms, that's how it works, right? They drew pictures of a concept product based on this guesswork, not based on any actual physics, spectra, spectrometer technology, or knowledge of food.

On behalf of the marketing people involved, might I ask what your point is?

[Horatius]

Originally Posted by uvar

And, from the IG updates:

Quote:

Our U.S. patent application 13/958,909 entitled "Analyzing And Correlating Spectra, Identifying Samples And Their Ingredients, And Displaying Related Personalized Information" was filed on August 5, 2013.

I haven't been able to find that one (though maybe I messed up or the shutdown is to blame?).

If they only filed in August of 2013, the application likely won't be published for another 16 months or so (usually published at 18 months after the filing date). Unless they filed an equivalent application somewhere else over a year ago*, we'll have to wait and see what they have to say in the application.



*Applicant searches on "Tellspec", "Isabel Hoffmann" and "Stephen Watson" all come up empty, so I don't think there are any earlier applications out there.

[Horatius]

Originally Posted by cosmicaug

New video at http://youtu.be/oLrX4J3gXZk .

Just watching that video, he seems to imply that their patent application is directed to the analysis engine, and not the optical device per se (see 1:53 and subsequent). In fact, he explicitly says the scanner device doesn't actually exist yet.

[Olowkow]

Looking at the website for Tellspec, I noticed a disturbing pattern in some of the commenters "What people are saying". Apart from the extreme feel of sciency claims, there is that subtle underlying tone of encouraging faith in the impossible.

Quote:

“I can’t even begin to imagine the complexity of what you report to have accomplished. If you can reliably and affordably accomplish what you have described, your product would disrupt multiple markets like no other device in history. … I would like to stay in touch and see if we can help you accomplish your mission so that we can eventually bring your technology to the task of improving the quality of the food that people use to fuel their bodies.”
—Jim Meehan, MD, founder of CatalystMD and the Chief Medical Informatics Officer for Prevently.com (MA, USA)

Here is just one of Dr. Meehan's posted videos on his YouTube channel. This one is very anti evolution, but there are lots of anti just about everything videos. His channel is pretty out there. Typical right wing neatsy tricks to try to keep the flock convinced.

YouTube Video This video is not hosted by the ISF. The ISF can not be held responsible for the suitability or legality of this material. By clicking the link below you agree to view content from an external website.

I AGREE

Then selecting another of the commenters.

Quote:

TellSpec will revolutionize the way we all eat and the food we shop. Knowing what is in our food from sugars, trans fats, toxic chemicals to allergens is an essential and important information for anyone that wants to stay healthy. Congratulations!”
—A. Belzuzarri, M.D. (Marbella, Spain)

This M.D was in charge of the now defunct(?) "Clínica la Fuente de la Juventud" (Fountain of Youth Clinic).

Quote:

He was born in 1940 on the shores of the Amazon River in the peruvian Amazonía. His father an M.D. teached him traditional medicine and he became a Shaman. In 1957 he came to Europe to study medicine.

Quote:

“Finally a device that can help those that are concerned about how to avoid toxic chemicals in their food. This disruptive technology is going to help so many of us to eat better and have a better health. Well done!”
—Gary Osborn, President, Texas Institute of Functional Medicine (Dallas, TX, USA)

I'll let the reader look Osborn up for himself. The Google hits are shall we say, numerous.

Something does not pass the smell test with Tellspec. In my opinion of course. Couldn't be these folks are in on the scam, could it?

[jaydeehess]

This reminds me of the fake conspiracy I dreamt up.
In that a low power laser retina reader was being developed for use by banks to ensure security of your accounts.
Turns out though, that if one examines numerous scans of the retina one eventually can also map a complete genetic profile and the banks plan on selling that info to health insurance companies because it identifies risks of future health problems.

[uvar]

Originally Posted by Horatius

If they only filed in August of 2013, the application likely won't be published for another 16 months or so (usually published at 18 months after the filing date). Unless they filed an equivalent application somewhere else over a year ago*, we'll have to wait and see what they have to say in the application.

Ah - not having ever dealt with patents directly, let alone US patents, I saw "Data current through October 3, 2013." on the application search page and figured that meant all applications filed before then were visible. Thanks for the correction.

[Olowkow]

Originally Posted by jaydeehess

This reminds me of the fake conspiracy I dreamt up.
In that a low power laser retina reader was being developed for use by banks to ensure security of your accounts.
Turns out though, that if one examines numerous scans of the retina one eventually can also map a complete genetic profile and the banks plan on selling that info to health insurance companies because it identifies risks of future health problems.

Moot point now that insurance companies cannot legally discriminate against those with pre-existing conditions.

Plus, merely getting cameras with sufficient resolution and decent recording medium into banks has taken decades.

[jaydeehess]

Originally Posted by Olowkow

Moot point now that insurance companies cannot legally discriminate against those with pre-existing conditions.

Plus, merely getting cameras with sufficient resolution and decent recording medium into banks has taken decades.

Now there you go trying to introduce mere facts, logic and reason into a perfectly good bogus conspiracy theory.

BTW, correct me if I am wrong but insur-cos cannot deny coverage due to preexisting conditions, but can set rates taking them into account.

[CapelDodger]

Originally Posted by jaydeehess

Now there you go trying to introduce mere facts, logic and reason into a perfectly good bogus conspiracy theory.

Perhaps you could drop "insurance companies" and insert "gumment death-panels" and/or "FEMA death-camps". That should attract the target demographic.

[Olowkow]

Originally Posted by jaydeehess

Now there you go trying to introduce mere facts, logic and reason into a perfectly good bogus conspiracy theory.

BTW, correct me if I am wrong but insur-cos cannot deny coverage due to preexisting conditions, but can set rates taking them into account.

I've heard so many lies lately, I'm not sure of anything anymore. I don't know.

[Hellbound]

Originally Posted by jaydeehess

Now there you go trying to introduce mere facts, logic and reason into a perfectly good bogus conspiracy theory.

BTW, correct me if I am wrong but insur-cos cannot deny coverage due to preexisting conditions, but can set rates taking them into account.

This is correct. Everyone is mandated to get coverage under Obamacare, thus they can't be denied. But rates have to take that into account for an insurance company to remian solvent; either that or everyone pays a higher rate and the healthy subsidize the ill (i.e-those that follow doctors instructions, eat right, exercise, and care about their health end up paying more for insurance than they would just paying doctors bills...while the 3-pack-a-day smoking, chronic couch potato that consistently ignores medical advice concerning his diabetes gets a hell of a deal).

The idea behind setting rates, and what the actuarials and underwriters and such try to do, is group people into similar groups based on their risk factors: lifestyles, medical histories, and pre-existing conditions. The idea is that you share costs with people in a similar risk category to you.

It sucks for those with pre-existing conditions through no fault of their own, but trying to do otherwise ends up leaving the system so open to abuse you might as well chunk it. For example, buying a low-tier insurance until something bad does happen, then immediately updating your policy. In the aggregate, such types of practices would either make the higher-end insurance products prhibitively expensive or lead to the cost imbalance mentioned earlier. A national system, where all coverages are the same, avoids this type of abuse by (essentially) mandating a certain coverage level.

On the up-side, many insurance companies are doing a bit more, so even those with a pre-existing condition can lower their rates. Some are offering programs to help with management of long-term conditions, participation in which can give a discounted rate, as one example. Or health screening that can be participated in, where a few basic things are checked (height, weight, blood pressure, and a questionaire on drinking, smoking, and similar), and participating in the screening gives you a discount (the insurance comapany gains by having better statistical data to base it's actuarial tables on; the results aren't going to raise of lower your rates, but you get a discount each month just because you did the annual screening).