Hmmm... I'm skeptical. Not necessarily because I don't think that pneumonia could be detected by testing VOCs in breath, but because I'm currently working on a project that uses sensors to do breath analysis and my amateur research has informed me that it's fairly hard to get right (which is why my primary goal is to identify deltas rather than achieve numerical accuracy).
For one, VOCs can be present in breath for other reasons besides some sort of infection in the lung, and VOCs are incredibly hard to differentiate with just a sensor. The fact that they tend to be faint in human breath even at their highest (in contrast to O2 and CO2) doesn't help. Even the most expensive PID sensors for VOCs (they get up into the several hundreds a pop) can't really tell you whether the predominant gas is acetone or alcohol or acetaldehyde or hydrogen sulfide. So you've got to figure out whether the presence of VOCs is truly an anomaly and not just a part of ketosis. In which case you will also need to measure at least VeO2 to see whether the VOCs correspond with the Respiratory Quotient.
The "e-nose" project, as described on the MakeZine article, doesn't appear to do that. It does have an alcohol sensor. But these sensors aren't particularly sophisticated. They use semiconductors with heating elements to detect the presence of gases, and there is almost certainly some overlap between the alcohol and VOCs sensors.
If VOCs are produced by pneumonia, then yes, it's conceivable that even just the VOCs sensor alone would detect this. But can this group of sensors used in the e-nose differentiate pneumonia from catabolism?
Maybe? ¯\_(ツ)_/¯
After all, this thing uses AI. And maybe AI can recognize something that a human can't by simply looking at a line graph. I dunno... Such things should be tested against known inputs before being suggested to diagnose anything.
One good way to debunk this is to measure raw sensor output and compute Mutual Information (which incorporates sensor noise/variability). If the sensor only produces X bits of information, no algorithm will be able to extract more classes than that. In the SCIO case it was just under 8 bits total of information. So something like a poor color sensor. You could train on apples and oranges and maybe do an investor demo, but it's not actually going to do anything useful (as the Kickstarter crowd soon learned).
True, but there are things where AI can help. For example, in the domain of electronic gas sensors, AI can be used to disentangle confounding variables like gas, humidity and temperature. All three affect the sensor output in a nonlinear fashion, and an ANN can learn the transfer function that extracts the (almost) pure gas response.
Gas sensing is really tricky. Metal oxide gas sensors respond nonlinearly to all three of gas, temperature, and humidity. Plus they drift. AI can help with the nonlinear response. Drift hasn't been solved yet, as far as I know.
Is the limit: A) sensor resolution, B) NN architecture and/or algorithm, C) training sample size, D) training data (labeling, segmentation) quality, or E) it doesn't sufficiently predict the variance with low enough error?
New NN models are able to do more with the exact same sensor data.
You cannot conjure information out of thin air. Even with infinite data and a hypothetical wormhole CPU that runs everything in O(1) and solves the halting problem, you still couldn't do this. So to answer your question, the reason is effectively (A). Sensor resolution might be the wrong term but it's the general idea.
How do creatures know what to eat? Evolution solved that for most creatures, so their sensors don't have to work as hard at runtime. And in other cases, some number of members of a population of creatures will die before the population learns the food is poisonous. Our sensors, and the information processing systems that manage their outputs, are remarkably efficient data processing engines that do the equivalent of approximating and predicting, often well beyond what the most advanced deep learning systems are capable of doing now.
So, sensor resolution is higher, there are multiple fields being integrated, in a massively-parallel spreading-activation Biological Neural Network, and that's how blank-slate creatures just know?
Is there enough information content - per the Shannon entropy definition or otherwise - in DNA and/or RNA to code for the survival-selected traits that
I'm not sure that the (Shannon entropy, MIC, Kolmogorov,) information content of the samples is the limit of any given network trained therefrom? Is there anything to be gained from upsampling and adding e.g. gaussian blur (noise)? Maybe it's feature engineering, maybe it's expert methods bias, maybe it's just sensor fusion; that's the magic noise.
Perhaps this is moving the goalposts a bit, but e.g. depixelation does appear to defy such a presumed limit due to apparent information content? Perhaps it is that the network reading the sensor carries additional information associated with the lower-resolution or additional-fields' sensor data?
> Given a low-resolution input image, PULSE searches the outputs of a generative model (here, StyleGAN) for high-resolution images that are perceptually realistic and downscale correctly.
Maybe no amount of feature engineering can actually add information?
Because they receive additional information from the environment through highly sensitive sensors producing massive amounts of information. Whereas the information you get from a cheap sensor effectively discretizes to a few bits.
Having designed sensor systems, I've lost more than a few hours of my life having to explain "why do we need that big expensive sensor when you can do everything with machine learning?"
The idea that a magic math technique can replace expensive sensors predates NN's by a few decades. Dozens of start-ups have gone bankrupt trying to do non-invasive blood glucose with portable sensors.
This is a very crude but at least conceptually useful rule of thumb: It's all of the above, but ultimately the analysis result is a mathematical function of an array of values produced by the sensor. Very few math functions do not have the property, that variation in the output increases with the level of variation in the input.
AI can extract information from a sensor that is 'obvious' when you look at it by eye, yet no easy combination of frequency filters and a carefully tuned threshold can extract reliably.
AI can detect more information in the whole dataset, because it for example has the whole "breath in- breath out" cycle in view. Fungi residing in the mouth would be present as background noise even during breathing in and out. But fungi-products existing at the end of a breath out cycle, are most likely to originate from the lungs, due to the mouth contamination being "flushed" out by the breath itself.
Priors can make sensor information more useful maybe, but that is just knowledge that helps first limit possibilities before taking a measurement. Priors also work against you when you are trying to sense something novel that might indicate a thing you don't expect.
An aside on sparsity priors (which that article uses).. reality is actually a lot less sparse than the researcher models would have you believe. If most dimensions are not truly zero (e.g., have some small noise present) these sparsity methods fall apart. That's why you (never?) see the methods deployed in actual products.
Specifically, the support determination step usually breaks down in epsilon sparse and you also get "noise folding".
It looks like the principle is that a machine learning model trained on the combined output of four different kinds of gas sensors can discover correlations between unintentional characteristics of the sensors. For example, the manufacturer of an ethanol or nitrogen dioxide sensor is not going to specify anything about how it responds to vanillin, but it seems plausible to me that the relationship between their responses contains some hidden information that could help to discriminate between vanillin and eugenol. With enough different sensors, there's quite a bit of information to be found in mining their undefined behavior.
That is to say, you can treat the sensor reading as being completely meaningless and skip interpreting it as indicating VOC levels. You're just using the sensors as black boxes that produce arbitrary values with the property that exposure to organic vapor changes the output "somehow", and letting model training find some meaning in it.
> With enough different sensors, there's quite a bit of information to be found in mining their undefined behavior.
It sounds like you would need to be exceptionally careful that your meta-process didn't "find" some signal in pure noise (via re-using test sets and so on).
> It sounds like you would need to be exceptionally careful that your meta-process didn't "find" some signal in pure noise (via re-using test sets and so on).
It sounds like you’re actually talking about ordinary levels of carefulness in this (ML) context.
You're intentionally depending on the "personality" of each gas sensor to get data measuring unknown features, so you can't expect consistency from sample to sample. Anything that was completely portable between different sensors would inherently be less powerful.
Most high-accuracy systems incorporate an onboard calibration target of some kind. Could be a gas cell (either sealed or consumable) or a special lamp etc. Or you buy an instrument that comes with calibration coefficients from the manufacturer. For example if you sell spectrometers, you put in the grating and manually adjust it for the desired range. This is the case for cheaper instruments (eg Ocean Optics) as well as expensive bespoke systems which are all hand built. Even if the grating and mirror mounts are fixed, the tolerance in manufacturing is rarely good enough that calibration isn't required. It's way cheaper to do some relatively low accuracy machining and then just epoxy all the screws down.
In this case you'd probably calibrate each sensor to a standard chemical sample and then use the calibration output. You could train on that, not the raw samples and then you have a model that works on all devices.
He was specifically looking to identify fungal pneumonia not just any old kind of pneumonia.
The linked Wikipedia article indicates mortality in immunocompromised patients can be as high as 90 percent. That sentence fits with my general impression that fungal pneumonia is both real serious shit and also typically found in people with advanced cases of other serious medical problems, like AIDS or cystic fibrosis.
It sounds reasonably plausible to me that it's feasible to detect fungal pneumonia in specific this way with some reasonable confidence level.
From working on the environmental sensor side of things, I'd concur. The VOCs will be able to be picked up, but the cross talk will be huge across other VOCs that don't themselves indicate pneumonia. There isn't one VOC, there's thousands. False positives are written all over this. This is the very same approach Theranos went. On a science level, sure, technically possible maybe. You'll even get boolean outputs. But on an engineering and regulatory level, you're in for a world of pain without the spectral tech that is still 2-5 years away before this is worth basing human lives on.
You would partner with doctors at a research institution that had lots of patients of this type, and the doctors would need to know how to run a clinical trial. But realistically, you would do this in any number of ways using existing samples before running a trial. Tissue samples are fairly easy to get.
This is what happens when you give children the tools to succeed by teaching them math and science in ways that directly relate to their world view. Obviously, this kid is very bright, but giving kids the tools to understand how the ideas they're learning can be applied in the real world is so satisfying. They aren't blinded by previous failures, or the current market, or what can and can't be done.
Sometimes you give the children these tools and they decide to do something they are more interested in. It is important to realize that not all children are interested in sciences.
Of course, no one should be forced to make things. But a lot of kids are naturally curious and can look at things in ways that are not obvious to adults.
But, by the law of nature, not all kids are prodigy. The normal distribution curve still holds, and the current society is putting too much pressure on them. Last time I visited India, I was shocked that parents were forcing their kids to learn IIT related thing in class 6. I encountered same thing in US, where kids were being prepared for competitive exams like SAT etc… Parents expect their children should go to prestigious universities like MIT and standford. The case is worst in China.
Do you need to be a prodigy to wire up sensors and controllers? More pressure to learn how modern technology works? That should be a great thing.
This exactly is the issue: not all kids are geniuses and not all kids are morons, but you'll NEVER know if you treat them like you think they deserve to be treated. If you believe that kids can't understand complex things, they of course won't because you won't explain it to them.
There's a difference between allowing people to learn what they're interested in, and forcing them to cram for standardised tests. The first one is the best thing that ever happened to me, and the second is among the worst. (I was the one forcing myself, but the point still applies.)
I was one and honestly I'm really happy my parents were neglectful (in the legal sense); my peers at award dinners might have had their university and career plans set out by 12, but none of them struck me as happy or emotionally healthy people, and the older I get the more I'm thinking that turning our smartest kids into robots or sociopaths is a bad idea.
I was a hyperlexic kid + a mental calculator who was a child programmer.
At the age of 6, I was reading at the level of a college graduate, and I started coding in elementary school; I taught my first intro to programming class when I was 11.
My middle-school standardized test scores put me in the top .03 percent and my IQ was tested to be in the mid-140s.
So I think I would count by most metrics, barely. (Nothing quite like going to dinners for the top 200 whatever and knowing you're number 199 or so ha).
I'm also a mental basketcase who developed MS in my 20s, so while I've done pretty well for myself so far, I don't have the personality, dedication, or temperament for great success. There are just a lot of things from my childhood that really make me raise an eyebrow now as an adult.
The amount of pressure is terrible, of course, but in addition to that, there's this weird push and pull where a lot of adults will say out of one side of their mouth how special you are and hold you to standards that are unreasonable for children and accept you + your contributions if they're helpful, but the minute you have your own opinion or disagree, you're a child and obviously don't know anything.
I'm a lost cause for society, intellectually speaking, but I feel concerned for the me born in 2015.
That said, every single kid who I knew growing up who had access to these sorts of opportunities but did something "more interesting" has ended up regretting it in adulthood. This probably isn't true across the board, but among the people I know it definitely is.
Maybe. It’s more like it’s what a kid’s parents do when kids can no longer be kids and are driven to start working on their college applications earlier and earlier.
To briefly clarify on why I believe it is the former: Caleb just happened to have caught a really bad fungal pneumonia when he was 9yo, and that was his inspiration to explore what could be done to diagnose things differently. And he just happened to do tons of research to try and re-use open source technology, etc.
There's more to making electronics than getting into MIT and prestigious universities. I think a lot of the users on HN like myself enjoy creating things just to create them. I don't need prestige or accolades; just build something, break it, fix it, take it apart. Some people enjoy learning how things work.
Does anybody have a link to what the kid actually did? I skimmed the video, looked at the diagram, read the original MAKE article, and it seems like the kid did a science project and it doesn't actually diagnose pneumonia. Am I missing something important?
The reason I'm asking is that I see a lot of these (used to judge science fairs, worked with smart undergraduates who build their own equipment) but most of it overstates the technical advances made by the kid.
My guess,
1 He found/was given an off the shelf fungal sensor designed to detect pneumonia
2 He hooked it up to a raspi
3 He trained a small tensorflow model to give true/false signals based on the input
All in all, not that bad of a little hack.
What I'm most disappointed by these science fare projects is that its often found that the parents of the child are top engineers in the specific field of the projects. In this case, perhaps his mom is a Sr Engineer at a company producing artificial noses aimed at detecting pneumonia where she is in charge of developing dev-kits and SDKs that happen to include sample tensor flow models.
What annoys me is that the story is often one of a kid, against all odds, learning all of this tech out of their own gumption. Where in the same science fare, there probably was a kid who had no help from their parents, who hacked together a 'are the lights on' circuit, using hand-me-down tech components, who's getting no notice.
> there probably was a kid who had no help from their parents, who hacked together a 'are the lights on' circuit, using hand-me-down tech components, who's getting no notice.
I'd have struggled to articulate what annoys me about stories like this, but this absolutely hits the nail on the head. I went to a school in the City of London with very elite investment-banker-parents demographics, and I can't tell you the number of stories like this. One comes to mind where one kid won a contest for designing a stockpicking algorithm, and it turned out - of course - that his mum was a fund manager at Goldman, specialising in that exact same area. I don't know what the point of it is. Is there not more to life than gaming university applications?
What bothers me the most is there was another kid who didn't win, who did real scientific work, and will go on to be a great scientist, but will never get the attention, credit, or funding that the first kid did.
I cannot over-emphasize how utterly demoralizing things like this are to those children. The kids who are smart enough to do real work are also smart enough to figure out that nobody will care.
I was so discouraged by finding out the other children who liked web development and coding had outside help, and I had a really hard time understanding why things like Synapse could get a PC Mag review, but I'd be accused of being a liar if I talked about my own projects, because that's what happens when you're a kid working without an adult. Without an appropriately credentialed adult vouching for you, people accuse you of stealing your work, lying, being an arrogant snob, etc.
ESPECIALLY if you're self-taught or were taught by adults society doesn't think much of. It's believable that the Pages taught Larry to code when he was wee, but obviously I couldn't have learned anything from my parents since one was a high school dropout and the other was the son of a factory worker. How could THEY have known anything?
Note that Larry Page's parents did more than just to code. They embedded him from birth in the academic system so that he was prepared, at Stanford, to recognize the importance of, and capitalize upon the organic value of web content. He was also exposed to the Grateful Dead, which probably helped build a healthy dose of capitalism and libertarianism.
Similarly, I was raised in an academic environment exposed early to computers (4th grade) in a gifted and talented program that many other students didn't have access to. That same program was literally "grad school prep in elementary school" which meant that when I got to grad school, it wasn't like I was in some sort of foreign country; I was home. And when I finally got to google, eveyrthing there felt normal and natural since it was just an extension of grad school, but with more money and better sysadmins.
That said, I guess if you don't get a leg up as a kid and go on to be successful through grit you can always appreciate that your hard work and determination paid off?
I used Larry as an example because we're from the same city, ironically. My grandparents (the ones I knew anyway) are from the class that repaired electronics; we're the mechanics to their automotive engineers. My grandfather (born in 1919) was obsessed with televisions and repaired them as a side business, so the tools were all around for my dad when he started buying broken microcomputers and fixing them so he could play games. Which in turn taught him enough of the basics (ha) of BASIC and meant that when I was a kid and had basic coding questions he could help. But since we all learned on our own over 3 generations, our facility with tech doesn't 'count'.
I can't claim to be completely without privilege: I'm about as well-off as you can get and still be considered from a disadvantaged background: My mother dropped out of high school and ran away at 15 (ironically her family included engineers and at least one MSU professor), so while I never had the money, my socio never really matched my economic, class wise. It's more that if you were any lower on the ladder than I was, you just didn't have a computer or MAYBE you might have something like the 5 free hours of AOL in the late 90s. A computer with hours of internet access in 93 required some privilege.
There were a couple of G+T things I did, but I got the money we did scrounge for those from my dad, and once he married my stepmom that stopped because she believed very strongly in gender roles (she wanted me to like clothes shopping and makeup like my sisters and her female relatives) and she also wanted to live in the middle of nowhere.
> That said, I guess if you don't get a leg up as a kid and go on to be successful through grit you can always appreciate that your hard work and determination paid off?
Eh, I'm not going to be successful, because doing so would require neglecting either my own health or that of my family. I have MS, and that just explodes the ability to do any kind of career planning.
That has its own silver lining, though, because I can say whatever the hell I want. Which is its own kind of freedom: I have a 'get out of hustle free' card. It's almost like skipping from being 25 to 65: I had to do a lot of reckoning with my own mortality, what I was worth if I wasn't able to have a high powered career, what did I actually want out of life, etc.
Precisely. You can definitely see the downstream effects of this, too, with lots of academics who see great success by publishing total tripe in well-packaged books (see: Malcolm Gladwell, the entire field of social psychology, etc).
University applications have such a compounding effect on things the rest of your life that paid-gaming of university admissions might well be the highest ROI investment for many wealthy.
That kid a few years back who fed buckets of soil bacteria on plastic waste and selected for the buckets that ate the waste best was pretty cool. I really hope that wasn't fake.
Dig deeper into some of these stories and you realize they also knew someone at the newspaper. Or a very expensive college applications specialist (i.e., $20k+) orchestrated the entire thing from concept to connections to media feeding, etc. Throw in a back-story professionally written and you've got top college acceptances!
Obviously, also throw in hand-selected medical specialists who diagnose you and prescribe extra time on the exams, great photos of your child at the local soup kitchen, a clutch summer internship with the local congressperson's office.
The entire college game is comprehensively stacked against the poor. Throw in the abandoning of test-based systems towards "leadership evaluation" acceptance methods and you get even more invested into gaming this process by the wealthy.
The MC knows what the crowd wants to see. If there is some contest, they will make sure the most attractive person wins. This is what the crowd wants, and the losers have no reasonable basis to protest and if they do they'll be (falsely) accused of being poor sports. Most attractive people don't know what's really happening, and assume their win is real.
The thing is, I get it. There's a wholesome excitement around the idea of discovery and you want to do your part and not be a wet blanket. And it's a white lie that is good for society - if not for the ego of the hero. You want there to be a new discovery, that came out of nowhere, because that's the better story. It's the kind of Myth that a good society runs on, and needs, even if it's false, because the real out-of-nowhere discovery stories happen too infrequently to be of use.
The best thing to do, really, is to give the kid a medal, and shut up about it not being real, and hope to high heaven he isn't misled by the easy victory.
Best science fair I ever saw was at a remote construction site near Qinshan, China in 1999. Many Canadian engineers lived on a camp by the site, building two nuclear reactors [1]. The camp also had a school for the engineer's children, literally one room with a teacher and about twenty children from grade 1 to grade 8 [2]. It was a good school, the teacher excellent and the kids clearly loving it. The older kids got a lot out of helping the younger ones. There was excellent quality recent school work in evidence on the walls. Though I did occasionally pop by the school when I'd visit the site, I usually didn't.
On one of my site visits I was asked if I wanted a detour from the project site to check out the school science fair. I later figured out that the minor scheduling difficulties I had around that particular visit was so that I'd be there on the day of the science fair.
Every student had a project. There were a few of the usual suspects, like the baking soda volcano and potato battery. However, those were the exception. Most of the projects were astounding, well beyond what I'd seen as an engineering undergrad in university.
The kids, standing proudly in front of their project and the bristol board explanations, knew very well how to explain the project and had a deep understanding of how it had come together. They'd definitely done the work and were justifiably proud.
That said, the majority of the projects were such that they could only have been the product of many evenings and weekends over months of father[3]/child working together. I'll assume that work on the next science fair would have begun the day after the science fair I saw wrapped up.
2. High school was a boarding school back in Canada.
3. I'm pretty sure that all the engineers were male, for I'd be remembering a female engineer, but the school kids were a balance mix of boys and girls.
I have friends where the dad works, and the mom runs the household, but the mom is just as good an engineer (or better). Since you're acknowledging and explaining the issue with [3], perhaps just using "parent" would have been better.
While I understand the logic of using parent, as a general rule I'm uncomfortable deliberately substituting words with less information when a word with more information is available.
It'd be like seeing a flock of geese fly over and saying birds. If you weren't really sure they were geese, or thought maybe a few were not geese, then maybe you write 'birds'. However, if you saw geese and it would have been striking and obvious if one or more of the birds was not a goose, then more information is given saying geese rather than birds.
If there had been a female engineer at the site, working or at home, I'm pretty sure I'd have known. This was rural China in the late 90s. A live in nanny would have been available at very low cost. Plus, the hunger for engineers willing to live at a camp site in rural China for months at a time was such that had there been any engineer spouses, they'd have had to make a very deliberate decision NOT to work.
No that’s the point - there wasn’t evidence that any of 1, 2, or 3 was ever done. Some mime guy puts together a gas sensor and tinyml setup - the kid makes some report on the hypothetical ability to use to diagnose fungal pneumonia in reference to some papers in the literature, but I don’t see actual evidence of an actual experiment.
We ran into this in Cubscouts with pinewood derby...the solution we had was a build day where the kids could go from raw block to finished car with our help and tools (belt sander with used up belt, parent or leader running the scrollsaw for the younger kids...paint at the Cubmaster's house and the parent doesn't have to worry about spraypaint)
Then to get past the 'parents doing all the building' we ran an outlaw class where the siblings and parents could compete...but it's the same kind of dynamic.
I don't immediately see the issue with the parent helping a child with tech they're familiar with...helping my son 3d print and sell fidget spinners had lots of little life lessons wrapped up in it.
Heh... at "build day" only the son of the person with the tools was allowed to use any of them. I ended up building a real clunker and felt terrible for years when it lost every race to better-engineered systems. I didn't get any real parental help.
This time around (by which I mean, my son was in cub scouts and doing the derby) I helped my son by showing him some basics of woodworking and how to make something that looked right and rolled properly, but beyond that it was all him. He didn't win any races, but wasn't bummed about it at all.
Following that, I bought a bunch of pine blanks, read a few papers on how to make faster cars (those nail axles are REALLY DUMB), bought Fusion 360, designed a car, and flip-milled it on my personal CNC, over a period of a year (it's never raced). it amuses me to no end that imposter syndrome and OCD drove me to be a well-compensated software engineer with enough free time to build his own pinewood derby racecar in his own time on his own terms.
I was cubmaster for 3 years and felt a little bad that my two boys didn’t get near the attention the other kids got…til the last year I helped them with weight distribution, lubrication and axle alignment. The Wife and I ran in Outlaw and had the family been eligible would have taken 4 of the top 7 times. (I think the boys got 2nd and 4th)
I’m looking at 6 of the cars now, I really should mount them in a display or something.
One of the boys is learning chassis fab and welding and the other is learning Industrial Design…so I guess it was a good experience.
One contest I'm aware of accidentally had the organizers kid win the contest. Who would have thought it possible? I feel really bad for the one that actually deserved to win.
Does anyone else approach these "teen invents x" or "wiz kid middle schooler discovers y" articles with extreme skepticism? About half the time the invention turns out to be bogus or trivial, and in the other half it comes to light the parents were behind it.
I, for one, am one of those people. However, after watching a few minutes of the linked video, I'm convinced (and pleasantly surprised) that Caleb has firm grasp on his design, and honestly sounds like he was the driving force behind his own particular implementation.
If you watch the video it's very apparent that the kid deeply knows how everything works. If he did not make this entire (very impressive) project himself, he certainly could have. He understands how chemicals interact with the various gas sensors (used limonene, pine and seed oils as test substances), how the model works, how to grow fungus in a sterile environment to get training data, how to get an API together to service the device...
I agree, these clickbait articles are rarely honest and most of them reek of parents trying to turn their children into (internet) celebreties.
Also I noticed that they typically employ a very common pattern in which the headline makes a truly big claim (e.g. "10 year old invents cheap way to purify any water source"), then afterwards it turns out that this claim is far from accurate (e.g. the child did not invent it himself, had massive help and while the solution technically works it is in not feasible at all in the way the headline suggests).
Once it is noticed that the claim is false in the way it is presented the article then gets defended by pointing out that a child that young coming into contact with such a project is still impressive, which is again technically true but ultimately comes off as a dishonest deflection.
The international science fair is a good example. I grew up with a couple of people who would participate every year, and every year it was obvious from their academics and being in the same classrooms and social circles as them that they didn't come up with this stuff. Their projects were gigantic, expertly researched, and featured technology that no high school student in the pre-internet age would have been able to source or use independently. One of them had a single mother with a PhD in the field, the other had a science teacher mother and a father with an MS in biomedical engineering... in the same field.
Yeah it's usually "adult who knows how to do a thing presents kid with all of the pieces and guidance to make it happen". Which is great, you should do that for kids, but the articles about it usually make one cringe.
Yup. There's very little drawback to just disregarding any article which highlights a person's youth and their amazing accomplishments. Same reason to avoid "30 under 30" type articles as well. You're playing the game those people want you to play by ingesting those articles, and I see no need to engage myself in other people's "grinding".
Yes, seems like people really want to believe stories of teenagers suddenly making breakthrough discoveries. In this case it's probably having some sensors with data, throwing them into a neural network and getting good results on a training set. The question is, does it work at all on real world data...
I think the important part is to understand that the primary goal here is not to advance science. We have lots of adult scientists with better education, time, equipment doing that.
It's to improve the pipeline of kids excited to go into science by making science accessible, rewarding, and prestigious.
I mean this is a cool application, but not original idea. Gas/particulate sensors are used in ag to detect fungal spore concentration zones for spot treatment by looking for a specific particle size using the laser in the sensor. The worst part about these type articles is that they don't even cover the design usually so you can't tell if it's a novel idea or not. This one features hand drawn diagram with some incubator system, I guess it's for training purposes but idk. Mostly it seems like this is an advertorial for Microsoft AI.
Cool. And, I have a 13-year and she is still, literally, crying over spilt milk, hacked Roblox merchandize, how done the steak is, why her monitor is tilted wrong, and why I din't warned her before rebooting the primary router.
> and why I din't warned her before rebooting the primary router.
Basic sysops rule: either create redundancy (which is hard to do in a consumer space outside of Mac Pro machines as 99.99% of laptops carry only a single LAN port and in towers, about 3/4) or warn your users before doing maintenance.
Well, I do have backups. This is India, so I even have a backup for the backup. I have three ISPs load-balanced, and not experienced any downtime since the beginning of the Pandemic (early 2020). I do realize them going down but we never realize until I looked them up.
It runs almost all the time, but sometimes I need to update settings, etc. which needs reboots the load balancer that distributes everything from.
And you cant do this durring a) the school day when the primary end user is not home or b) wake up at 2am and reboot it while the primary end user is asleep?
Honestly, I think your kid should break out the SLA and check what compensation they get paid for prime time outages.
I mean, another solution is to give warning at regular intervals beforehand (day, hour, ten minutes, 1 minute perhaps). Basically planned outages with ample warning. Of course, avoiding prime time planned outages is always good in any scenario and I'm sure goes a long way to keep all clients happy.
With that said, I doubt your 13 year old daughter client is actually paying for that level of service, so... ;-)
I believe the SLA for a 13 year old is 110% uptime, with consequences being The end of the world.
I am not a lawyer so feel free to go to court claiming that 100% uptime is the limits of mathematics or that the world will not end if <insert social media platform of choice> is not accessible for 5 minutes on a tuesday afternoon. Its a losing case every time, best just to settle up the case quietly with an extra scoop of icecream or some robux and cut your losses.
My son, now a sophomore at UCSC, definitely gave me a few moments of "Uhh... I really hope my kid isn't an idiot" at that age. Living in Silicon Valley, he had friends creating crazy Gary's Mod levels using Python (this was a decade ago) that they collaborated on using GitHub. I was shocked at how sophisticated junior high coders could get! My son, however, is not a techie and like his father, has always been a little immature for his age. I was like, "Why is my kid the only one who isn't a genius!?!"
It all turned out well and now he's happily studying economics (yeah, my apple didn't land anywhere near the tree). Everyone matures at their own pace, and computers, as I'm sure all of us know from our own history as geeks, are easy to impress people with. If you're really into biology, animals, astronomy, etc. what can you show people to wow them? Not much that hasn't been seen before. But any 13 yo can download and learn how to use the latest professional CAD software, the same IDEs pros use to make AAA games, or the same backend AI services used by major companies. And they are encouraged to do so! I can't imagine there's a lot of "Learn CRISPR at home!" tutorials out there. That makes a big difference.
a high end high school will have a biology teacher who can teach high school students to do crispr (crispr is quite easy).
My path into biology was looking into a microscope and seeing a world of non-computer state machines that behaved like cellular automata. and then my path led back to computers because real biology is much harder than CS.
And I think some kids get into astronomy if I demo my telescope showing them planets and stars. But I don't really push kids too hard in one or another direction; my parents thought I should learn german so I could excel at organic chemistry
Ha! This is HN, I should totally know better than to throw stuff like that out there. I admit, as I wrote the crispr line, in the back of my head I wondered if it was true or not and just went with it because I thought it sounded witty. I stand corrected!
Is the headline being a straight-up lie a valid reason for flagging? The kid did not diagnose pneumonia. He came up with an idea for a design that might hypothetically detect pneumonia.
Unfortunately there is no other way to signal trash submissions. "Not upvoting" is not really a sufficient signal when enough people get baited by the headline.
I agree, but I highly suspect that if you flag a lot of submissions that aren't also flagged by others you end up on some kind of list where your flags are ignored or deprioritized.
I've mentioned this elsewhere but I think Hacker News needs people who are vetted users who can flag these articles with specific bits, such as "this is clickbait that oversells what happened here", or "this is an unnecessarily partisan analysis of COVID policy outcomes". That signal is much more important to me than "this post was flagged because people don't like emacs"
The main invention is in the electronic nose. The kid just did the plumbing of connecting it to some ML library.
Of course, the electronic nose itself is a work of plumbing too, where some existing gas sensors are put on a pcb.
In short, nothing seems really new here, but the application is interesting. I guess it's always interesting when people start looking for correlations in data and get some positive results, so from that point of view it is noteworthy.
The sensor (the four gas sensors on the board) was created by a third person.
The artificial nose is the TinyML model which trains on the sensor data (CO, NO2, ethanol, VOCs) to detect arbitrary scents by their signatures in those four categories.
The fungal pneumonia detector wires up a whole API with azure etc. and trains the model specifically to recognize pneumonia, based on an actual science experiment which grew and measured the fungus in artificial lungs.
As far as I'm concerned, both Caleb and Benjamin had brilliant ideas, executed them fantastically, and created something that may be truly useful. A $40 sensor that can detect disease just by breathing on it is more of a tangible contribution to humanity than many software engineers make in their life and almost certainly more than 99% of us did before the age of 14.
It is great that the kid is involved and interested in these technologies. Whether it works is another issue. You are going to need metrics such a LODs, LOQs, sensitivity and specificity to determine if this beats the gold standard tests.
You are wrong. The kid GREW FUNGUS IN ARTIFICIAL LUNGS using a sterile field made out of a plastic bin with dish gloves cut into it. He didn't need medical data. Anyone could have done this- anyone with the intelligence and creativity that this kid has.
is it possible you're overstating what the kid did a bit here? The history of sterile fields shows that even great scientists take decades to debug contamination that causes false positives and negatives. ASn experiment like this can be easily thrown off by any number of variables that weren't carefully controlled for.
can't you kind of say this about all ML? That the main driver in ML is 99% the quality of the training data and 1% the specific details of the neural networks used?
For most real-world applications of ML, yes. Of course, what happens in ML-research is different (where e.g. new networks for new modalities are invented).
But back to the topic, I bet the kid didn't even invent their own neural network topology, but just pulled a predefined network from a library, perhaps without even knowing it. Which is ok, because that is how most people use ML.
I think the news is less "13 year old revolutionized medicine" and more "13 year old used ingenuity to create cool thing". Or at least, that's how I think it should be read. Focusing too much on the end result is likely not the best outcome because, while its a working prototype, many interesting prototypes never make it to fully end user capable system and there are a lot of hurdles to overcome to get it there.
But that doesn't take away from the fact that its a cool project and the kid did a great job in coming up with it and executing on it! Its definitely far beyond what most people achieve, nevermind 13 year olds.
This is a story about inspiration and achievement. Objective facts are less important than the message, IMO, especially after decades of "try-hard" being ridiculed in the U.S.
No, the author designed the ML training project also. The last section of the Make article is how to send the data to Edge Impulse and configure the ML training [0].
Mentioning Microsoft Azure IoT Central in the article and the video is odd, because you definitely don't need that to complete this project. It seems to be a feature that the Microsoft employee added to the GitHub project their self [1].
Caleb mentioned in the video that a co-worker of his aunt authored the research paper about detecting bacterial pneumonia from VOC levels. Everything else feels like a Microsoft marketing hype train that went off the rails.
The vast majority of work is plumbing, but there are still new things.
What matters isn't making new things, it's making new things that work, and bringing them all the way to completion.
The vast majority of innovation isn't exciting discovery, it's coming up with tests to convince yourself you actually did what you thought you did and sharing those results with regulators. That's the difference between Theranos and GRAIL.
I just built a tardigrade detector neural network; nobody has done that exact thing before, and now i'm talking to the world's leading tardigrade researcher because my tool might be helpful in answering an unanswered important tardigrade question.
But thousands if not millions of individuals worked over hundreds to thousands of years to bring us that tardigrade detector; all I did was take advantage of that to label 100 images! I've tried being an ML researcher; my conclusion is the best models are distilled from postdoc tears, and we merely retrain those models without pain.
I am not skeptical of the kid being able to do this. Good for them! I'm sure they will grow up to be an inquisitive and brilliant member of society.
However, like many, I feel like this article could be papering over... something.
OK I'll just say it: Privilege.
And hey! Not every kid with privilege ends up being brilliant! And he may not be privileged! But it is a lot easier to succeed when you have it.
And my problem with this article is this: We are constantly papering over how much of a difference a good education can make, and how little opportunity to get that quality of education there is in the United States.
You often see people bemoaning their lot in life: "Ugh. When Mark Zuckerburg/Bill Gates/insert CEO was my age, they already started Microsoft!"
And my reply to this sentiment is this: How many hundreds of thousands of dollars did your parents spend on your pre-university education? I'm willing to bet it wasn't in the hundreds of thousands.
I see this kid is from LA. Sometimes all it takes is being in the right zip code to have access to... science fairs? My school didn't even have AP classes! I thought science fairs were something that only happened on TV.
I realize this is a bit petty, and it 100% comes from my childhood where I went to a poor rural school where I was a poor student, and so-fucking-desperately wanted more, and then moved to the city, and succeeded, flourished once I got access to a better quality of education. But pretending it isn't there feels dishonest.
It feels like an onion article headline: "Kid with everything going for him, despite all odds, tremendously succeeds"
I say good on the parents for putting their resources to good use. A lot of them squander it away on spoiled kids, waiting for the kids to show interest in anything useful. Yes, it would be great for society if all kids have these benefits, but we will only get there when people understand that this is something worthwhile.
It is, and very American-centric. Plenty of innovative people are born in poorer countries with much less resources at their disposal. Indeed quite a lot of American top scientists are immigrants from not exactly rich places.
Of course, even they are sort-of privileged by the fact that they weren't born blind, on in a period of outright war, or didn't get cancer at the age of three. But this is already stretching the meaning of "privilege".
Katalin Karikó, one of the main brains behind mRNA, grew up in shabby Communist Hungary and her lab equipment at her home university was likely worse than what a median high school in the U.S. has at its disposal. (There wasn't much convertible currency east of the Iron Curtain to buy top stuff, and not enough capacity to manufacture it locally.)
Sure, there are geniuses that can spawn out of anywhere that occasionally rise out of bad situations. They are notable because they are EXCEPTIONS to the rule.
But there is massive inequality and poverty in the United States. Here is an example: In my home town, the poverty rate is 12%-13%. In the US state of Georgia it is 17%. In the Czech Republic, which has about the same population as Georgia, it is 10%.
So your assertion that "her lab equipment at her home university was likely worse than what a median high school in the U.S." is questionable. There are plenty of people living in horrible conditions in the US. Our scores in mathematics are 30th amongst developed nations.
Our cities are full-to-the-brim with a homeless population that we have abandoned to the streets that our cities and citizens do not have the wealth to address due to all of the money going to 1% of the population. In fact, our homeless population is almost to 0.2% of the total population, coming in at around 500k people.
Also, the assertion "other people have it worse" is not useful. I can be critical of our current society and also realize I have a privilege living where I do. I can see the impoverished system I grew up in, compare it to the opportunities afforded other people, and say: "Hmm. Maybe we can improve society somewhat."
I am from the Czech Republic. Our inequality is better than American, for sure, but we seem to be on the same track. Especially the gypsy ghettos are fairly bad. Not as bad as the worst places in the U.S., but shockingly bad for the middle of the EU. In neighboring Slovakia, some of the mud hut villages look like a mixture of the Middle Ages with rural Nepal. So I am not really sure if the better numbers aren't result of some creative statistics.
I still believe that Karikó's equipment was fairly bad, because I was alive in the 1980s and I remember the shortages of everything. Late stage Communism was all about stuff shortage. Either something was really plentiful (but usually of questionable quality), or very hard to acquire. This included school equipment. The first home computers spread some 10 years later than in the West and we regarded them almost as godly objects.
I remember having an assignment like this my second year of college. It was basically an array of various smoke detector sensors wired up to a parallel port. With diverse enough sensors, you could stick a cup into a box with a fan and the sensor, and be able to tell if the cup had OJ, coke, or coffee in it using basic PCA.
Given the tools available now, I'm not surprised a smart 13yo could build something like this. Especially if the sensor itself is an off-the-shelf component with device drivers and such already available for it.
Pretty awesome project really! As a sidenote: A bunch of the ML training and edge deployment magic is done via https://edgeimpulse.com which seems to make it much more accessible to build such a thing.
> No door is ever closed. You can do anything! I am a thirteen-year-old kid, and I can do this—if I can do it, anyone can!
If only a goal of education was to get kids to feel like this! Of course, not everyone is equally bright, but without the above attitude, they will not even try, not even try to be interested, because they think they are too dumb, which is a tragedy.
In 9th grade I took first in a Science State championship competition. I didn't just win, I blew the competition away, and it changed how people approached the competition for the next 20 years. My device was ugly, large, obtuse, poor. People laughed at it. The favorite in the competition was this slick device that rumor had it was designed and manufactured by their father's engineering company. Everyone laughed, that is, until my ugly device performed like a Lambo in comparison to their Ford Pinto.
My father? A smart, but not affluent, guy who thought hard on things. He's the one that actually invented the device...but I learned a few things along the way.
Parents absolutely help children become high achievers, but it doesn't always mean it was attached to their day job. Having attentive parents is a privilege.
note: As for the device, it was just a little car powered by a weight.
Does it matter? Watching everything through the myopic lens of "privilege" is wrong.
A kid can have an idea and parents can help. For now, the set of {kids, parents} that can do that is limited. But technology changes and becomes more accessible. What matters is the new things that become not just possible but easy and cheap.
For a previously "costly" problem that in 2012 would involve a 5 MP digital pictures + geotagging + OCR then sending the raws for GPU processing, any random smartphone from 2022 will do.
In 2012 you could have screamed "privilege!". Not in 2022.
As a kid, I'd have loved to try to hack together a app that recognizes mushrooms (or flowers, or fruits which I all found so super interesting, especially bugs and OMG they fly if I blew on them!!)
It would have been hard. I would have benefited from some help. But I would have had a lot of fun, after which I would have used the app to fill in the name for my leaf-book collection effort (I wanted to have a specimen of EVERYTHING from the garden, then from the street, then...)
I only had books and some websites and a bad camera. So I drew :) A kid now could have picture search engines like yandex to do better with a much better camera too (MACRO MODE!) and some generic photo processing software. A rich kid then could have had something similar, with an expansive Nikon camera, and photoshop (crop, filter...) and maybe some parental connections to biologists and botanists.
Is it privilege if they did? Yes. And it's wonderful because every kid has this privilege now! And they can have more fun!
Pause a second and look at what you wrote and what you're responding to. You're arguing very passionately against a straw man, nobody mentioned privilege until you did.
I was just curious and very aware of the possible blowback. I even doubted for a while to delete the comment ( no shallow dismissals ).
But it isn't and it is a question that comes along with the subject of 13yo geniuses.
I was curious and now I have an answer ( not from you ).
edit: In a broader, more contemporary frame : It is now we need to be ever more critical of news we get served and rejecting that under the guise of 'think of the children' is just lame.
It does when the title says "13yo kid builds e-nose". It's not about privilege, it's about being honest. Maybe the title is honest and the kid is just very bright, that's cool too!
Yet most people seem to be jumping to the conclusion, making assumptions, and letting their views taint their judgment, without even knowing all the facts (see a comment below asking if the parents were already working in the field)
> Does it matter? Watching everything through the myopic lens of "privilege" is wrong.
May I ask why?
You see it all the time, unironically: "How a 23 year old couple bought their dream home!" and it ends up in the article their parents literally paid for it.
Or Bill Gates. There is the classic: "How to become as rich as Bill Gates: Choose your grandparents carefully."
I see it elsewhere in this thread: "Ugh. And MY 13 year old just wants to play video games!" And that is unfair to themselves, and unfair to their kids.
You see it everywhere. So many musicians, artists, writers, PEOPLE succeed in part because they just don't need to make money. Because they already have it.
And then they act like it was all done themselves and while they are not bad people for having money, and they are sincerely talented, that isn't the whole story. The whole story is that they didn't ever really even need to succeed to live a comfortable life, and that is a HUGE advantage over other people.
---------
Allow me to tie this to my own experience:
First, I want to acknowledge that I am extremely privileged in my own ways. I grew up in a wonderful home with a great family. We were educated, and kind, and loved reading and were encouraging. The rural place I grew up in wasn't a fancy high-tech metropolis, but I did not experience any violence in my community which counts for a lot.
I went to a private university on a big scholarship. It was cheaper than any state school I could have gone to. My first two years where I was in the dorms and was paying with student loans, I was very active in the student volunteer community, and the computer science club.
But once I left the dorms, that was IT. I needed to 100% support myself, rent/food/etc. I spent my days working manual QA for a software company, 40/hr week, while attending school at night.
And my life changed. I didn't make any new friends and lost the ones I had. I was never around for the "college stuff." Every waking moment became toil, either through work or through school.
While my peers were doing research studies for natural language processing, or participating in CS contests, or building relationships, or falling in love, or actually doing well on homework and tests, or any number of productive things, I was plugging 3g WIFI dongles into and out of laptops for $15/hr. I was running test cases for 8 hours a day, then going to class from 6-9, then doing homework until I went to sleep, then getting up in the morning and doing it again.
And the deficit I had in my education followed me for a LONG time. Still follows me today.
So yes, I do think we should interrogate situation people were in when they achieved something. Because sometimes people don't really achieve anything other than spending their parent's money.
And there are circumstances where something that seems like a big achievement was really just an inevitability.
>No door is ever closed. You can do anything! I am a thirteen-year-old kid, and I can do this—if I can do it, anyone can!
Although he may have similar opportunities to his peers, he (understandably) doesn't yet realize that he has a "spark" that not everyone is fortunate enough to cultivate. Age is not strictly the limiting factor to being capable of reaching a high potential.
I wish I could read the paper. I’m interested in what kind of fungal pneumonia they were looking at. My googling got me to find the title on the Middle School Facebook page but not sure if it’s available ti read.
I would be interested in what they are testing, it probably wouldn’t be spores but likely something along the lines of biochemical compounds specific to certain species. For example we use blood tests for the presence of https://en.m.wikipedia.org/wiki/Galactomannan to help diagnose invasive aspergillosis.
I had no idea there was an e-nose project out there. Is e-nose a field with active research or is it something that's seen as being just too hard/cross disciplinary?
I see robust e-nose as being a huge development that would change how we see the world.
Did you ever measure a smell? Can you tell whether one smell is just twice strong as another? Can you measure the difference between two kinds of smell and another? It is very obvious that we have very many different kinds of smells, all the way from the odour of violets and roses up to asafetida. But until you can measure their likeness and differences, you can have no science of odour. If you are ambitious to find a new science, measure a smell.
we need to identify the factors that enabled this awesome kid to have the background knowledge, tools, environment etc and see how we can replicate them to find other such diamonds in the wild...
We seem to have a strange anti-intellectual bias in our culture when it comes to children.
I mean I was not pushed hard as a kid to play guitar or play football/wrestling but I was pushed. It took quite a bit of time before I really fell in love with playing guitar. It is hard to fall in love with something when you suck at it. I needed to be pushed. Sports were fun at the time but we completely overstate the value of organized sports IMO.
I don't know why we see playing a musical instrument as something different than learning scientific instruments. We expect the kid to just naturally be driven in science from day one and pushing them is seen as something morally off.
It reminds me of the crazy hockey parent that pushes their kid way too hard in hockey. While some kids might end up hating hockey, a huge % just end up being really good hockey players compared to the average person. People tend to fall in love with things they are good at.
I am childfree but if I did have a kid I would be a science version of the hockey parent and let the chips fall as they may. The risk/reward is just massive for the kid.
To what end? Our world is not short of this kind of talent, at any age. Those that have these abilities will get there, in time. So maybe it would be better at that age to teach them to paint, throw a baseball, fly fish, travel, etc.
Why should they be taught to paint or to throw a baseball but not to use a tool (like a program, etc)
I was merely saying that the knowledge and facilities to paint, to play baseball or program should be accessible, and kids must be exposed to the fact that they exist... then, the kids pick per their inclination?
knowledge does not motivate people to do great things.
what children need to learn first and foremost is to be good people, create the desire to help others and contribute to society.
once they have that, they drive themselves to learn what they need in order to achieve that.
this kid here had the drive to solve a problem because they had experienced it themselves. it doesn't matter how they solved it and how much help they got, what matters is what drove them to solve the problem in the first place.
if we can create that drive then any of the above, whether it is art, sports or programming will happen based on the kids motivation.
For one, VOCs can be present in breath for other reasons besides some sort of infection in the lung, and VOCs are incredibly hard to differentiate with just a sensor. The fact that they tend to be faint in human breath even at their highest (in contrast to O2 and CO2) doesn't help. Even the most expensive PID sensors for VOCs (they get up into the several hundreds a pop) can't really tell you whether the predominant gas is acetone or alcohol or acetaldehyde or hydrogen sulfide. So you've got to figure out whether the presence of VOCs is truly an anomaly and not just a part of ketosis. In which case you will also need to measure at least VeO2 to see whether the VOCs correspond with the Respiratory Quotient.
The "e-nose" project, as described on the MakeZine article, doesn't appear to do that. It does have an alcohol sensor. But these sensors aren't particularly sophisticated. They use semiconductors with heating elements to detect the presence of gases, and there is almost certainly some overlap between the alcohol and VOCs sensors.
If VOCs are produced by pneumonia, then yes, it's conceivable that even just the VOCs sensor alone would detect this. But can this group of sensors used in the e-nose differentiate pneumonia from catabolism?
Maybe? ¯\_(ツ)_/¯
After all, this thing uses AI. And maybe AI can recognize something that a human can't by simply looking at a line graph. I dunno... Such things should be tested against known inputs before being suggested to diagnose anything.