Things every Medical Device Startup needs to know

atdc-theclubhouseMy visit to (more on the space soon) coincided with an ATDC talk on how to get a medical devices startup going. There’s an ongoing gold rush in the industry that has been tempered by harsh realities, and the overall effect of the Affordable Care Act, if anything, is a rush toward cheaper, then better.

It’s not easy for startups these days. The FDA, which used to look over clinical data and rubberstamp new devices for free, now has hefty filing fees – a new device from a major company pays over a quarter-million dollars just for initial approval, a process that takes up to three years. The small business “discount” lowers this to $65k.

Good luck raising money for a new idea. Probably the best way is with economic development or research grants, which are available from the public and private sector if you can keep your ear to the ground. If you can’t get a grant, try selling the concept directly to doctors, who will provide a ready market for your device. VCs and Angel Investors will take a cut of the company, without giving much in return. Loans are worse, if you feel bad about going bankrupt (which you will, because faceless banks aren’t going to make a startup loan).

Of course, there’s the nothing-new option: If you can make a convincing case that what you make is just a modern version of something that was in the market in 1976, of all years, then your product can receive an approval at the lightning speed of 6 months from now! Just pay $1600 a year for your federal registration.

Generally, to get FDA approval, you need clinical data. But to get clinical data, your device needs to see broad enough use somewhere. If you have something brand new, you aren’t going to be able to afford a private research study for your device in a broad context. So where are device manufacturers getting their launch data from? Europe, of all places. CE Rating is an easier, faster process, and Europe has a well-developed distribution network for medical devices. After a few years of sales abroad, you’ll get the track record you need to press forward in the States.

And then you’re really on your own! Once you’ve got your Rolodex out and accepted the fact that you’re a glorified telemarker now, you’ve got to convince hospitals and private practices to buy your device. And here, you’ve got three or four separate battles. Are insurers going to pay for your treatment? If you can convince them to file it under an existing treatment code, rather than make a new one, you’ll be fine.

There’s the issue of having separate markets in each state, often with unique licence requirements, and differing accounting standards. Among the worst for sales? New York and New Jersey, whose hospitals notoriously pay bills 6 months after the fact, when they actually pay them. Better states: Michigan, Florida, California. The device market tracks heavily with the number of older people in a state, go figure.

Can you get your thingamabob into a standard kit for the treatment you’re enhancing? This is where you really start thinking about selling out – every needle and scalpel in those things might very well be from a vertically integrated conglomerate.

Say you’re a tech entrepreneur aiming to make a million dollars. The exact words the lecturer used was “plan your exit” – the idea that, as soon as someone else has a major equity interest in your company, you basically aren’t going to run it forever anyway, so you may as well sell out, move on to the next idea. How to get there? Well, pharmaceutical patents are a headline item, those are still worth something. A device patent? Big companies have been known to buy others just for patents, kind of an external R&D thing, but they’re way more likely to buy out one with a working business model.

As someone who has mainly viewed the device industry as an endless march of ever-pricier doodads threatening to crush the national budget with nonogenarian cyborgs, it’s heartening to hear that the Affordable Care Act is finally getting more patients to feel the sticker shock. Instead of just going for “the best,” as doctors always recommended when money was no object – people are starting to settle for “good enough.” The real growth in the industry might just be finding ways to perform the same old treatments for less. It’s not the glitzy cutting edge engineers often want to live on, but inventing a cheaper wheel is not all dull retread work, either.

Growing pains

I think I’ve gotten over most of the growing pains with the new set.  The Coolpad Rogue ships with two ways to get online, WiFi hotspot or USB tethering.  WiFi is working fine, at least in an isolated spot, but I hate crowding the 2.4GHz spectrum with a high-power signal.  USB tethering also pushes bits just fine, but I basically need to do an electronics project to keep it from draining my laptop and overfilling my handset battery.  What I really miss is Bluetooth tethering, maybe there’s an app for that.

As far as the throttling goes, I changed a few settings and found a better spot for coverage, apparently I was stuck on a 2G data channel when I started out.  24 kbps is actually pleasing news in that context.  Right now, I’m doing much better on 4G.

Fixing the click of death in Western Digital Black

I like using laptop hard drives in my desktops.  Something about the lower latency.  Call it a poor-man’s Raptor, if you will.

As it happens, I bought another drive from Western Digital, the WD Black WD5000BPKT, that had a decent 5-year warranty, but one constant and annoying issue:  the onboard power-saving features create an annoying *click* more or less randomly, as the drive seems to attempt a spin-down at strange moments.

I’ve had this issue since day one with this drive, and I’m still not sure I’ve fixed it.  Here are my workarounds so far:

Automatic Disk Access

Directory Listing

No joke – the way I fixed this under Ubuntu 12.04 was to add a directory search to my crontab, scheduling a disk use before the about 4 minutes or so it would take for the disk to spin down.

sudo crontab -e

*/3 * * * * ls /home/

Random Sector Access

I tried switching from this to having the disk read from a randomly selected sector, since I was concerned all those reads from the same place were going to do stress damage of a different kind.  So I thought perhaps I could get the disk to read random sectors on the plate with something like this…

*/3 * * * * sudo dd if=/dev/sdx of=/dev/null bs=4096 skip=$(( -1 + ( disksize * $RANDOM / 65536 ) ))

the command appears to work (that is, do essentially nothing) but the clicking returned rapidly.

Power Management Settings

Under Ubuntu 14.04, neither of these options appears to work.  So I dug down under the system settings to find the disk utility, which Trusty Tahr now simply calls the “Disks” program.


A few clicks later and I can set the disk’s APM setting to ‘254, Spin-down not permitted’,  You have to do it that way because setting the power management level to ‘255, disabled’ causes the drive to say “LOL, O RLY?” and still spin down anyway.

Flags for Minnesota

Minnesota’s flag is a bit of an eyesore, ranked at the bottom of subnational entities in North America.  We can fix that!

Quick, build on these SVG versions!

My concept shares a similar central element to the current flag, while being simpler and more distinctive. The North Star is represented by a golden eight-point star, similar to a compass rose.  Here, I’ve included some titling to de-emphasize the South, West, and East points.  The white ring is an element of continuity from the present flag; at a glance, at a distance, they will appear quite similar.

The way the star’s rays pierce the white is technically a violation of heraldry rules about ‘tincture’, but I think it’s distinctive enough  Titling is also generally frowned upon in proper heraldry.  A more revolutionary redesign might also axe the titling and use a less commonly-seen star, say ten points of equal weight for the land of ten thousand lakes.

Of course, I’m not the first to think that the Minnesota flag is terrible.  There was a strong push in the late 1980s to change it, but it didn’t quite work out.  The one with the Green is called the “North Star Flag” and was designed by Rev. William Becker and Mr. Lee Herold.

Lon’s Lair, best in show merchant at A-Cen

Recently, I attended Anime Central in where, among other things, I needed a set of disposable dice so I could run a few rounds of MAID RPG without worrying about my personal set running off. At the Lon’s Lair booth, I was able to pick up a solid set of d6s for less than $2!

As good as its at-con prices were, dice prices on Lon’s Lair website are even more reasonable! So I figured they were worth a shout-out.

My only disappointment was that the booth was mainly selling Chessex sets, and I am a total Koplow fanboy — so I was happy to see that the website is full of Koplow sets for sale!

Posted in Fun

The end of net neutrality?

All websites are equal, but some are more equal than others.

By itself, your computer doesn’t care whether you’re sending baby pictures or phone calls over the Internet.  It has to be carefully taught which kinds of content to censor and toll.  Today, the back end of the Internet is being clogged with toll booths, as a result of court rulings and a captive FCC that refuses to call ISPs and backbone providers common carriers.

The arrival of pay-to-play on the Internet wasn’t inevitable in and of itself — it has been a side effect of vertical integration between content producers, distributors, and last-mile data pipes.  If I had been born the heir to a Hollywood media empire, I could rejoice this week at my continued corner on the market.  Instead, I’m forced to ask why our society allows 5 corruptible people to decide everything about communications policy for between 0.3 to 7 billion others.

Bicycles as PRT

For decades, the holy grail of public transportation has been Personal Rapid Transit (PRT).  Essentially, a PRT system has all the convenience and privacy of a personal vehicle, without the unpleasant side effects of private automobiles, like excess pollution and traffic.

Using bike sharing as PRT solves most of the issues, with a few caveats:  Only the dedicated will use the system in adverse weather, traffic will get worse without sidewalks or bike lanes, and without cargo capacity, it’s only good for point-A-to-point-B.

Compared to other transit options, bike lanes are extremely cheap.  And people hate running errands in the rain anyway, so really the killer app is being able to do a grocery run on a bike.  The only folks I’ve seen able to do this are those with third-wheels, or plastic tubs strapped to their rear-wheels. If cargo space was already built in — or at least available on a considerable number of fleet bikes — there’d be few barriers to public acceptance of a bike sharing system.

Bikes as PRT do have one key weakness: diurnal capacity.  Without enough bikes in the system, it will become impossible to find a residential bike during rush hour, and without enough bike docks, it’ll become impossible to stop at the user’s proper destination.  Active PRT systems like ULTra can automatically move units to balance things out; bikes have to be moved by people.

But this is also a solvable problem:  Just introduce route planning into the Bike Sharing system — when people sign in, have them choose their destination, and incentivize nearby sites that need more bikes.  Or conversely, you can highlight nearby sites that would be better to take a bike from.  It could be done in the form of membership “points” and/or ride discounts.  Ideally, there should be a way to chain up two or three bikes together to maximize the movement potential when a counter-cyclical user enters the system.

Twin Cities Twinwheels

Bikes are a hassle for the well-funded; why invest in a bike when your car needs yet more expensive maintenance?  If you’re keeping the car, and yet pressed for cash, you’re certainly going to want to take care of it first, and your bicycle, if ever, second.  This is the situation my bike is in.  It’s certainly not ideal.

Professional bike maintenance is especially expensive in Grand Forks, where service and repairs are only offered by a single full-line shop.

So I went and used the Nice Ride Minnesota project in St. Paul to see if something like that could work elsewhere.


Maintenance is handled on the backend, and the bikes are decent and well-maintained.  If there was a problem with a bike, you can signal it to the system.

Bikes and open docks (more about this later) are plentiful where the system exists.

It’s all weatherproof!  Even with all the electronics involved, it works in the rain.

As it happens, you can do better time on a Nice Ride Bike than a transit bus  — more than partially because you have to ride your bike in the street, and the bus can’t easily pass you..

Present System Limitations

Payment.  Quick and easy for credit card holders,  mixed results for debit cards and impossible for chip-and-PIN cards.  “Legal tender” — what’s that?

Pricing.  2.5 hour trip on Metro Transit:  $1.75.  Rush hour round trip:  $4.50.  Least you can pay for Nice Ride: $6.00  2.5 hour trip on Nice Rice: $22.50?!!

Narrow time limits.  A 30-minute ride policy is designed to keep you within a 15-minute ride of the dock network.

Limited dock network.  Dense in downtown Minneapolis and St. Paul, and thinly spread between. Totally absent from Bloomington and suburbs.

Limited transit integration.  Seems odd that Nice Ride isn’t at every Light Rail Station.  Intermodal should be a clear goal.

Limited ridership.  By my estimates Nice Ride has between 7000 and 20000 riders in an urban area of 3.3 million.   One a drizzly day in west Stl Paul, I saw a couple other bikes, but nobody on Nice Ride units.

Prairie Public will air governor’s address on tape delay

Prairie PublicI tried my hardest to find out who was doing live coverage of the State of the State address today, without much luck.  Though I did hear that Prairie Public Radio was going to be there, I hoped for a TV feed.

But the Television staff at Prairie Public was there getting video, as it happens, and the address will air in its entirety tonight at 7pm.  Kudos!

CKMW’s new transmitter fuzzes out Prairie Public

CKMW in Winkler is getting a transmitter boost that will get it decent coverage all across northeast North Dakota. This follows their switch to FM from the AM band, started last year.

Being on 88.9 MHz, It will probably interfere with fringe reception of KMPR Minot, particularly east of Turtle Mountain, a small blow to Prairie Public, though with alternative coverage available from KPPD Devils Lake, it’s not as big as when the CBC built its 89.3 transmitter in Winnipeg that fuzzes out KUND in Pembina, Cavalier, and Kittson Counties.

With data from Industry Canada, and their handy Longley-Rice generator! (And some thanks to Kevin McDougal)