[palish]

What breaks my heart is that this kind of tragedy is avoidable. He was only 51; died of colon cancer. All we need is to design a wearable device which continuously monitors your body for changes that correlate with "might be developing cancer". (An undershirt, for example.)

This is a solvable problem. Cancer causes measurable (but typically unnoticed) changes to the rest of your body. If it can be measured, it can be monitored; and when that change is detected, a quick trip to the doctor's office would likely save the next 30 years of your life (if it was indeed the very early stages of cancer).

Cancer is beatable; we simply need to beat it before it has the chance to incubate.

This is a problem that I would like to personally work on, but unfortunately it would take at least a year to arrive at a feasible design. It'd need only a small investment (~$1M) but I doubt any investor would be interested in "zero profitability for at least 12 months".

[koevet]

Unfortunately I don't think that a "continuous body monitoring device" will help. I'm no doctor but I have read enough on cancer to safely affirm that the number of "markers" associated to the plethora of cancers will trigger so many false positive to make the device useless.

Recently there was a very interesting article on HN about the mindblowing complexity of cancer: http://www.sciencebasedmedicine.org/index.php/personalized-m...

[LinaLauneBaer]

Another alternative is to go to a doctor of nuclear medicine that has a MRT and tell him to make a full body scan. As long you don't have some metal implanted in your body this is harmless. Then ask him to see if anything is wrong with your body. Usually they will give you a DVD with the data they recorded for free. You can then take this DVD and give it to another doctor.

With a full body scan you can find an incredible wide range of things that could be wrong.

As far I know no insurance on the world pays for this without indication but its not really that expensive. Depending on the MRT device that you prefer and the country you are living in it costs about 3000$.

If you are afraid of a "closed MRT" there are also open MRTs:

http://www.radiologie-muc.de/uploads/pics/Radiologie_offene-...

[leoedin]

Hardware and R&D investments are regularly considerably greater than a million dollars. Lots of investors are interested in zero profitability for 12 months (assuming you actually have a business case and a viable product). Outside of the silicon valley software tech bubble, hardware design and manufacture is still an expensive business. I know of a number of companies building machinery for renewable energy generation that have received collectively tens or perhaps hundreds of millions of pounds in funding with no short term profits likely. There's also a lot of medical engineering research going on in universities, which doesn't generally have the same profit requirements attached to it.

However, your concepts sound like vague ideas rather than anything based in reason or science. One dimensional sensing (eg a heartbeat monitor) is fairly easy,but good and reliable multi-variable embedded sensing is hard.

In a later post, you mentioned mapping teeth with a toothbrush. That's not a trivial thing to do - without a fixed frame of reference it would not only be very complex mathematically, but also pretty unreliable. Similarly, chemically analysing anything is not something you can easily do in an embedded and reliable fashion.

A million dollars would not get you very far with embedded, reliable chemical testing. You clearly don't even have a concept of how to do it.

[palish]

Hey, thank you for the critique. It's a rare gift to find someone who's willing to disagree + solid reasoning + references to existing work. Thank you.

In a later post, you mentioned mapping teeth with a toothbrush. That's not a trivial thing to do - without a fixed frame of reference it would not only be very complex mathematically, but also pretty unreliable.

Of course. That is why I'd incorporate a frame of reference into the design of the toothbrush. The point of the post was to quickly paint a picture of what "the future" could be like if I were given time and freedom to pursue these designs. I assumed that if anyone was interested in the details, then they'd simply ask me to provide them.

The toothbrush contains two technologies that enable it to build a map of the surfaces of the teeth you brush. #1: a three-axis gyroscope + accelerometer (e.g. same as an iPhone's). This gets me the rotation of the toothbrush relative to the direction of gravity. Now, of course, this would not be enough by itself; we also need some way of determining the position of the bristles within the mouth. This is accomplished with sensor #2: along the length of the toothbrush near the mouthpiece, there are dozens of tiny light sensors. The more the toothbrush is inserted into the mouth, the less light that reaches the light sensors. This gets you a reading of "how deeply is the toothbrush into the mouth?" which you can easily combine with "what are the current angles of rotation?" to conclude e.g. "the bristles are therefore currently brushing the front surface of your bottom-left-rearmost molar."

Light sensors are easy to work with, but probably not the most reliable way of doing this. I would look into using some kind of sonar sensor, possibly. Or you could do something as simple as "detect which part of the toothbrush that your lips are currently touching"; that might work too.

The goal is straightforward: "to determine an accurate-enough approximation of how deeply the toothbrush is currently inserted into the mouth", because that can be combined with the rotation angles to give you a solid frame of reference.

I of course don't know which technique will prove to be the most pragmatic, since I haven't tested any them yet. =) This is a solvable problem, however.

I've got to run for now, so unfortunately I ran out of time to address your other [excellent] points. But I'm not sure anyone is even reading this discussion anymore anyway. Feel free to post a reply if you'd like to continue, or shoot me an email (see profile). Thanks again!

[aantix]

Your message sounds feasible but lacks enough to details for me to figure out whether you're being serious or not..?

[nopinsight]

I'd say this is a good project to get funding via kickstarter.com. Have you checked it out?

It will be risky for the funders, but the risk is small compared to potential benefits for each individual. Several much less useful projects garnered higher funding than the amount you mentioned.

[dsrguru]

palish, if you have enough knowledge and are willing to invest the time given ~$1M, seriously make a kickstarter project. If this is actually both possible and doable like you say (I don't know anything about oncology), assuming it's practical enough for everyone to wear and doesn't have harmful side effects, this has got to be just about the biggest contribution to humanity that you have the power to make. (I imagine it would also be rather lucrative.) You have my word that I'd advertise the kickstarter project at the very least by word of mouth, and I'm sure many others would do the same.

[swalsh]

Anything medical has extra risk. What if the device doesn't work perfectly? When my phone doesn't work perfectly, the call drops... but when my cancer device doesn't work perfectly I become over confident, and maybe miss other signs. Its a potential liability issue.

[palish]

Using Kickstarter is a good idea, but I'm struggling with the moral implications of it. The chance that I'll succeed is inherently small, and I'd feel absolutely awful if that $1M wound up not resulting in some useful contribution.

Still, I have ideas, along with the skill necessary to test them. And if I were independently wealthy, then this is how I'd spend my time...

If I were to Kickstart this, then I'd keep a public daily journal of my efforts and short-term goals. There would be a webpage showing exactly how the money was being spent. I'd make everything as transparent as possible.

The thing is, most of my ideas will turn out to be impractical, but there's no way to know which until I test each of them. That process would require money from people who are okay with gambling on the off-chance that one of my ideas turn out to work. (Sidenote: if I happen to succeed, then I want to repay the people who funded me, at the very least.)

Here is a glimpse into the future I envision:

- A sensor in your bed scans for tumors as you sleep.

- If you have a heart attack, a sensor in your clothing immediately detects it and pushes an alert to your phone, which dispatches emergency services to your exact GPS location. (Most people keep their phone in their pocket or purse, so this seems doable.) A speaker embedded in your belt yells out step-by-step CPR instructions to the people around you.

- Your toothbrush takes your temperature and absorbs a sample of your saliva, which it analyzes for deviations from your long-term norm. It uses wifi to upload this data to your computer, along with a map of the surfaces of the teeth you brushed. The next time you use your computer, you'll see a visualization of your teeth; any surfaces which you missed when brushing are highlighted in red. (e.g. if you aren't brushing behind your molars, then those areas are highlighted.)

- As you're driving, a sensor on your rearview mirror will detect if you fall asleep at the wheel (your eyelids close for more than one second) which triggers an audible alert to wake you.

I'd love to spend my life coming up with pragmatic ways to use technology to benefit your health / monitor for emergency conditions. I'm just not sure it'd be ethical to use Kickstarter to fund possibly-crazy endeavors like this. I don't know.

[JoshTriplett]

It sounds to me like you have a pretty good idea of the technical details and challenges in such a project. You'd need to find someone with medical experience that you could work with to figure out the details, but you could find quite a bit of that expertise just by chatting with your local medical school. And at least in the short-term you won't have to worry about medical device certification, since most of the ideas you've mentioned would not directly get used for medical treatment; any actual diagnosis would still get done by a physician.

If you're serious about being willing to work on this proportionally to how much funding you get, then please by all means start a project. Pick a couple of appropriate ambitious-but-likely-possible goals, set expectations very clearly, and see if you get any takers.

[waterlesscloud]

Would this project need FDA approval? If so, the timeline would be quite long and a million bucks would come nowhere near covering it...

[JoshTriplett]

Long-term, some of these devices would need FDA approval, but many of them wouldn't.

Speaking as a non-expert (only familiar with FDA approval processes through the experiences of friends and colleagues): none of these devices would constitute a class II or III medical device, since they'd only act as an early-warning system rather than as a primary diagnostic tool or direct treatment system. Some of these devices might fall under class I (which also includes things like tongue depressors), but meeting class I wouldn't necessarily prove insanely onerous, and some care would likely allow avoiding class I as well in some cases.

[possibilistic]

This is literally the field I'm preparing to go into after I get my PhD in systems bio. The absolute best way to develop models for markers is to study metabolomics data for a huge sample size.

Take, for example, urine samples from a large number of at-risk persons and see how a specific type of cancer manifests within the group. Run the samples on GCxGC-MS and do high-order statistical comparisons or train an ML model to pick up on the markers. Given the proper treatment, these may correlate to breakdown products attributable to the development of cancerous cells.

You can model any metabolome too: blood, breath, sweat, biopsy... Urine is probably the least invasive. You should learn a thing or two about how elimination works: glucuronidation and other reactions that modify substrates to be polar enough.