Cheap Smartphone Dongle Diagnoses HIV And Syphilis | Hacker News Reader

Cheap Smartphone Dongle Diagnoses HIV And Syphilis

iflscience.com

102 pointsjoewee32 comments

Was it really necessary to involve a smartphone at all?

> color change [is] picked up by a set of photocells in the dongle, and the results are then sent to an app

The logic to discriminate a color change could have been handled by a 75-cent microcontroller. Add a cheap display or a couple LEDs to show the test result, plus a cheap AA battery, and you have a self-contained gadget for less cost than creating the smartphone interface.

Pregnancy tests and all sorts of sophisticated chemical test strips don't involve a smartphone.

I'm thinking that the smartphone hookup is either:

(1) a way to add future functionality like maintaining a database of the tests or uploading test results to somewhere (with all the privacy implications and risks of those), or

(2) a marketing ruse since we all know that a smartphone app that does X is way cooler than a old-fashioned self-contained gadget that does X.

An an example of how easily people fall for coolness of having a smartphone do something that doesn't actually need a smartphone, check out this comment to the original article:

"I have been resisting the smartphone trend for years because they are expensive and there was no unique use for them: everything a smartphone did I ALREADY own a gadget for. This changes the game. Now I want a smartphone."

A couple thoughts:

1. The comment you're talking about is from, essentially, someone who doesn't know what they're talking about. This isn't really meaningfully a "game changer" in terms of "Should I buy an iPhone?"

2. Computers are already used to analyze many of these types of diagnostic tests. It may have been easier to port those systems to a smart phone than to develop a new, bespoke microcontroller. Expertise is a cost.

3. There are massive benefits to having this hooked up to a smart phone, and as you have mentioned, much of it goes to extra functionality. During the Ebola outbreak, case reports often had to be dealt with in paper forms processed by hand, which slowed both reporting and contact tracing. From a public health perspective, being able to link test results (deidentified or not) with a central repository at a ministry of health is a big deal.

Imagine the use case of this not as one person and a smart phone testing themselves. Imagine it as one person and a smart phone testing 4 villages over the course of the week, in conjunction with actually getting people treatment.

> 1. The comment you're talking about is from, essentially, someone who doesn't know what they're talking about.

Yes, of course, the person who made that comment is clueless about technology -- just like 90% of the population. The comment shows that most people think it's really cool that the diagnostic device uses a smartphone and don't realize that the smartphone wasn't needed at all.

> 2. ... than to develop a new, bespoke microcontroller.

Nobody needs to develop a new, bespoke microcontroller. You use a tiny cheap off-the-shelf microcontroller (like the PIC) and program it. Based on the information available in the original article, the program might be as simple as light the red LED (positive) if the color sensor indicates it's above a certain threshold and light the green LED (negative) if below a certain threshold. But even very complex algorithms can be ported to microcontrollers.

> 3. Imagine it as one person and a smart phone testing 4 villages

I did say databases and uploading were a possible reason for the smartphone, and do I agree that a smart phone does make sense for the use case of testing a whole village. This doesn't change my original point that a smartphone wasn't really necessary, not even to reduce costs, and might in fact be more expensive than a standalone device. The original article makes a big deal of the smartphone connection without explaining the real reason why the smartphone is there.

That reminds me of a parody of the lean start-up movement - a must-read in my opinion:

http://theanorexicstartup.com/wp-content/the-anorexic-startu...

They were featured on the Science Friday podcast:

http://www.sciencefriday.com/segment/02/06/2015/honey-i-shru...

This pages talks about the false positive rate of some existing tests:

http://mobile.aidsmap.com/Accuracy/page/1323395

False positive rate : 14% LOL !

No dongle jokes, please...

1) To be pedantic, it detects your immune response (not the microorganisms directly)

2) "14% false alarm rate" => That makes it basically useless.

False negatives are very bad. False positives not so much. Worst case is you send someone to a better test to find out if it's true. This is what they aim for with home pregnancy test kits, afaik, for pretty good reasons.

That very much depends on the disease in question. For these diseases, false negatives are bad. For diseases with intensive and dangerous treatments, false positives can be actively harmful to population health. This is commonly a problem in cancer screening.

Disease and plan. Almost every rapid HIV test out there is spoken about in "negative" or "maybe positive" terminology. If you get the two dots instead of one dot, you're referred to a physician who will administer a 4th gen test and probably a western bot, you aren't immediately started on ARVs.

> False positives not so much.

Really. No psychological effect then, you assume ?

1. Which, makes it potentially of limited use for HIV testing in high risk populations, since it can take months for immune response to be detectable by this sort of test. Not useless, certainly, as it is actually comparable to what most people get when they go to the doctor to get tested for HIV (not in accuracy, as your second point mentions, but in how long one would have to wait after exposure to know if you're infected). But, for populations not at significant risk, it would be a very good in-between test.

2. Early HIV tests had quite high false positive rates, and an imperfect false negative rate. That improved dramatically over the span of a few years.

This is a new research device. It is not expected to be perfect. My first version of anything is never perfect.

Technology futures can be hard to predict, but history indicates the following are consistently true: It'll get smaller, faster, cheaper, more precise.

Will getting smaller, faster, cheaper, and more precise make this device an effective diagnostic tool? I'd be willing to bet it will.

Yeah, false positives is what destroyed that company recently that had convenience store HIV tests. Lots of people who bought stock lost their shirts on that one.

Given the consequence of a false negative for both of those diseases (fatal if not treated), it makes perfect sense to optimize for a higher negative predictive value (if the test says you don't have it, you don't have it) versus a higher positive predictive value.

While it would be ideal to have a high rate for both, that's the one I'd choose.

And that false alarm rate is actually a dynamic function of the prevalence of the disease in the population being tested, which would make me hesitate to call it "useless", especially this early in its development.

Those values, for the record, aren't that much worse than rapid diagnostic tests for influenza, which have a False Positive rate of about 6% and a false negative rate of about 18%, assuming 30% prevalence.

How does a 14% false alarm rate make it useless? If it fails to turn positive for actual infection, I agree. But my understanding is its 114% positive, meaning if you have an infection its 100% positive and 14% positive when you aren't infected. Subsequent tests weed out the false positives.

Out of 1000 people, say 100 are actually infected with syphilis. With 14% positive error rate, 140 would test positive. It's much better to only need the expensive 2nd test for 40 people rather than 1000.

You are thinking about this wrong, what it actually means is that 14% of the time, it will detect positive. So say out of 1000 people, 1 is infected with syphilis. That person will tests positive, but so will 140 other people that don't have syphilis. So the test just referred 141 people to consultations and further testing, only 1 of which actually has the disease.

So, assuming the medical industry in the developed world gets out of the way (which I don't think is a foregone conclusion...it will take years, and possibly too many resources, to bring this to market in the US for a variety of good and bad reasons related to our regulatory environment), this is potentially one of the biggest revolutionary changes in healthcare in my lifetime.

This makes detection of both viral and bacterial infections possible with very low-cost tools; the fact that it can detect both makes me believe that detecting practically any infectious disease (or at least any infectious disease that has identified biomarkers in blood, or other easily obtainable bodily fluid) is on the near horizon.

One could buy this dongle, or one like it, and new diseases could be added via software upgrades (or by in-app purchases for the diseases you want to check for). This would potentially make doctors office visits obsolete for all but serious disease and chronic conditions. No need to go get checked for strep when you have a sore throat (of course, getting the antibiotics to treat it is another regulatory hurdle people in the US would face), lower need for cities and universities to provide STD testing, though treatment and education would still likely be within their perview. It allows better self-management of HIV, hepatitis, Lyme disease, and other chronic viral illnesses through more rapid and frequent testing.

Even if this doesn't evolve to include other types of detection, I would still happily pay several hundred dollars for a re-usable STD testing tool. I like to get tested every 6 months, but the hassle and cost of going to the doctor or Planned Parenthood or online testing lab is high (free or low cost testing is available, but you trade much more hassle for the lower cost...among my friends who get tested at the city health center, they usually spend an hour or two waiting, which is not a cost I'm willing to pay to save a couple hundred bucks). If I could test for HIV for free before every new partner (and, hell, make it a practice to test with every partner), that'd be fantastic.

Another article about this was posted a week or so ago to HN, but it didn't get any comments or many upvotes, which was disappointing, as this seems like entirely revolutionary technology, to me. And, it makes me want to go into medical tech because of how huge the potential is (but it would be an entirely new field for me, so my ability to be productive in it would take a while to develop).

You cannot possibly software update a device like this. The actual detection mechanics are chemical. The advance is in miniaturizing and automating tasks which used to take a laboratory and staff to do.

Is it? I don't know the mechanics. How about genetic testing, is that also a chemical process that is unique for each marker? Seems like a genetic sequencing tool could be programmable once the DNA/RNA has been analyzed. The best HIV test available is an RNA test, which can provide results as early as six days after exposure. This is obviously not my area of expertise, so I don't know how realistic it would be to downsize such a thing.

Also, how does this device perform a test for a viral infection and for a bacterial infection, if the tests are unique and chemical? Does it use a particular substance for each test, which has to be refilled per use? (Again, I have no idea how this stuff works.)

I know that 23andme continues to produce new information about my genes as new studies are done to find genetic markers for various diseases and traits, which is pretty cool. If there were a device that detected viral and bacterial RNA/DNA, would that be software extensible?

Did you read the article?

Quote from the article:

"Blood from a patient is drawn through microscopic channels where it then meets chemicals that react with HIV and syphilis antibodies to produce a color change. This can then be picked up by a set of photocells in the dongle, and the results are then sent to an app on the attached smartphone or computer."

Huh...I'd read the earlier article and didn't remember seeing that explanation. I didn't read this variant of the article, as I assumed the one I saw earlier was the more authoritative source, but, perhaps this is just a better article.

Yes, the tests are unique and chemical. You either need bespoke antigens that will react to the presence of specific pathogens, or primers and other components for DNA/RNA based tests.

Beyond that, these testing systems probably can't manage the sensitivity and specificity of a properly run diagnostic lab. That's fine for the setting they're being considered in, because those labs don't exist, are hard to reach or can be easily overwhelmed in an outbreak, but for a developed world setting? I don't really see it. Home diagnostics for infectious diseases over a broad range of potential pathogens is a really hard technical challenge with no real payoff - you still need to go to the doctor.

If you have accurate test results that show you need medicine and can obtain it without a gatekeeper then you don't "need" to go to a doctor. That's one hell of a payoff.

Yeah, there's no way you're getting prescription antibiotics or antivirals without seeing a doctor because "My iPhone says I have pertussis".

Lab results alone do not a medically responsible treatment plan make.

This is the exact reason that free STD testing clinics DO NOT discuss symptoms with you, and with a positive test immediately refer you to an MD. A testing environment is not treatment or management.

Its unclear who "you" means, most people on earth can actually buy antibiotics without this. Also unsure as to how you define what is "responsible" for them.

The post at the top of this chain makes it abundantly clear we're talking about the United States.

And, as an epidemiologist, the ready availability of antibiotics without any interaction with a doctor is something I'd be pretty comfortable classifying as inherently irresponsible.

Reply on news.ycombinator.com