Patrick F. Spear

Unsurprising. Though the average (and completely understandable) opinion is that a trained person’s intuition can’t be matched by a unfeeling, “methodical” machine, I would say that this type of AI is actually the perfection of a trained person’s intuition- it’s a way of arriving at a best guess, put together by the unconscious weighting a large amount of data accumulated over a long period of time.

I would imagine there there will be enormous pushback to this method of diagnosis, because these AI systems are being applied to a very critical knowledge industry, where people’s live’s are at stake, without fist gaining the public’s trust in less critical applications. So then why are these systems being developed for medical use before they’re able to do something more “simple”, like decide what kind of campaign a marketing company should employ for their client?

Quite simply, it’s the dataset. I can’t think of any other knowledgebase that would be so broad, detailed and complete than medical records. And any real-world AI application, no matter how advanced the algorithm, will always be bound by dataset quality.

Next question: how does facility layout and room design change when this job is outsourced to an entity with no human-scale spatial requirements?

Easy choice for my top pick of CES this year. This is the endgame of the Futureward project- this is what the new breed of connected ward is being designed to actually connect to.

In the last few years, I’ve been astonished at how quickly theorized developments that formed the basis of Futureward have come to fruition; this project takes the cake in terms of bringing that future into reality today.

3D printing was one of the cornerstones of the original Futureward project. In more recent iterations, the 3D printing aspect has been de-emphasized, as the benefits they brought to the project were minimal compared to the even the optimistic projections of costs.

However, the rapid uptake of interest in 3D printing in the last two years has me thinking that de-empahsis might not have been necessary. The Form 1 is the latest of a string of cheap, consumer-oriented available 3D printers and its production is being funded by backers at Kickstarter.

The promise represented by this machine is best summed up by user tweakingforjesus on reddit:

In 1991 my roommate bought a new laser printer for his tech writing business. It was one of the first affordable laser printers. In terms of quality it blew away the dot matrix printers the rest of us used. On simple text it would spit out about 4 pages per minute. Cost was $2K.

This is the 3D printer equivalent of that laser printer.

Couldn’t have said it better.

Finally, we’re witnessing the beginning of the end of the last two awkward decades of environment “automation” being handled by attaching wires to everything we want to measure.

That kind of “manual” automation, where you actually have to do the majority of the work yourself, is best typified by the mid-00s promise of smart refrigerators that would tell you what food you had available and when the milk was running low. The catch? You have to scan the barcodes of anything you put in the fridge for it know its contents. Wonder why that never caught on.

But what about a fridge that uses computer vision to actually look at the milk and see how much is left, without the user having to change their habits at all? Now we’re talking.

So the wired (and even wireless) devices of the last 20 years can finally be tossed in the bin, and instead we’ll put a camera in the corner of the room and attach it something smart enough to know what it’s looking at. The technology has finally caught up with the vision for a “smart” environment.

Of course, I’d rather see this tech cataloging medical supplies than measuring milk.

It will be nice if hospitals of the future captialize on voice control as a means getting computing done without having to touch any surfaces, but there will always be a need for traditional desktop computing workstations due to the speed and accuracy they afford while doing big tasks.

However, these are keyboard and mouse setups are also hubs of transmission for nosocomal infection, so it’s nice to see Logitech taking a step towards more easily cleanable equipment.

Ideally, keyboards would be treated like linens- collected three times a day and sent down to the CSSD for sterilization, while freshly cleaned ones are distributed to take their places.

Anyone tried the best-of-both-worlds approach of in-person care via an online tool?

The story presented in the link above is largely one-sided, and I’d be interested to hear the doctor’s/hospital’s side of it, but the point remains: patients need access to the raw medical data that is captured about them by medical devices.

The patient in the video admits that he is not a cardiologist, but simply has an interest in being expert about his own body. Most patients may not even be that interested in their own medical data, but they are subject to the risks of risks of being denied access to their data just the same: as soon as patients can be denied that access, they can be locked into a device or healthcare provider ecosystem.

The solution? Get people interested in capturing their own medical data with their personal digital devices now. Make it easy and brainless to capture sleep cycles, caloric intake, blood sugar levels; anything that is doable for cheap. We basically want what Wired discussed in this classic article five years ago. People may not roll their own pacemakers, but getting people accustomed to owning their data, and pressuring medical infrastructure to incoporate a BYOD model will ensure that the medical data field will develop into- and remain- a standards-based system for years to come.

I’m willing to bet that the same thing will happen with artificial intelligence systems.

This is like something directly out of Ghost in the Shell or Neuromancer.

Not related to healthcare or architecture in any strict way, but it’s a hell of a reminder of how quickly even the wildest science fiction scenarios are becoming commonplace.

I’ve always looked to Kinect-like hardware as the future of monitoring patient movement and agitiation, but this is a far better solution. Looking forward to seeing how this develops.