This is going to be waaaay crazy and out-there, except my antennas are twitching that it might be reality within 18 months.
Don’t ask me where I got that date; my antennas told me. Serendipitously I read two articles tonight. (I love Thanksgiving weekend.) I’ve probably got the details wrong, but consider three technologies:
- Amazon’s new “Kindle Fire” tablet
- Amazon’s new “Silk” browser
- IBM’s Watson.
1. Kindle “Fire”
In case you missed it, there’s been enormous talk about Amazon’s latest Kindle product, the Fire. It bears little resemblance to previous “books only” Kindles – it’s a media-streaming tablet. To understand what a big difference this is, see the new Wired interview with Jeff Bezos, CEO of Amazon.
Note: all is not great – I’m a strong believer in David Pogue’s reviews at the New York Times, and he says this first version of Fire is ugly-bad:
You feel [the cheap] price tag with every swipe of your finger. Animations are sluggish and jerky — even the page turns that you’d think would be the pride of the Kindle team. Taps sometimes don’t register. There are no progress or “wait” indicators, so you frequently don’t know if the machine has even registered your touch commands. The momentum of the animations hasn’t been calculated right, so the whole thing feels ornery.
But he ends with this: “Then again, Amazon tends to keep chipping away at the clunkiness of its 1.0 creations until it sculptures a hit.” For instance, today’s conventional Kindles are far slicker book machines than the original; they’re beautiful.
So for this fantasy about healthcare in a year or two, think about Fire in a year or two.
2. Fire’s new “Silk” browser and cloud computing
I love my laptops, but I long for power in the palm of my hand. And I un-long for my the cost of my 3G and 4G data plans, a separate one for every device. In short, I want all the speed I can eat, via wifi.
But websites that are designed for hard-wired computers aren’t optimized for slower connections. The Wired interview explains what Amazon is doing to re-architect how data gets to your browser:
One of the things that makes mobile web browsing slow is the fact that the average website pulls content from 13 different places on the Internet. On a mobile device, even with a good Wi-Fi connection, each round trip is typically 100 milliseconds or more …
We’ve broken apart this process. If you can be clever enough to move the computation onto our cloud platform, you get these huge computational resources. Our cloud services are really fast. What takes 100 milliseconds on Wi-Fi takes less than 5 milliseconds on Amazon’s Elastic Compute Cloud.
The interviewer, Steven Levy, points to how this may shift power fundamentally – especially how much power you need in your palm:
It’s a hack that aims to make web pages download much faster. But it also has grand ramifications: It points to an era in which the device is so secondary that even computation takes place in the cloud.
Mixed emotions here:
- “Computation takes place in the cloud”: that has one level of impact if you’re talking about tablets, but it has quite another level if you think about smartphones (iPhone, Android) or even “app-phones” (e.g. Blackberry) and cheap plain-old cell phones. The lesser devices have less power. What if they can access power anyway? Even the most disadvantaged people?
- But Levy, what are you smoking in that Fire? Tests by Anandtech.com show no improvement at all with that feature enabled, and Pogue’s test of three sites says “The iPad took about half as long each time.”
But let’s say that too gets resolved in 1-2 years, and just think about this architecture – a potentially potent streaming tablet, connected to a cloud platform that does a ton of the computing for you, and delivers slick answers to your hand.
Now, what if we connect that to a wicked-smart health information platform? Hm, where would we find that?
3. IBM’s Watson
Nine months ago IBM’s “Watson” supercomputer played the Jeopardy quiz show against the two best Jeopardy champions ever, and stomped the snot out of them (NY Times). One of the champions was Ken Jennings, who had won 74 consecutive matches. Watson beat him by $77,147 to $24,000. The second-best champion ever, Brad Rutter, earned $21,600.
At the time, my antennas instantly told me to write a post titled “Watson, I need you“:
Did you know the first “phone call” was a cry for medical help? Alexander Graham Bell had spilled acid on himself, and hollered to his assistant, “Watson, come here. I need you.” In a remote room, Watson was astounded to hear the first sound come from the primitive speaker.
This week, the amazing victory by IBM’s Watson immediately made me think, “What could this thing do in healthcare?? If it stomped the fastest fact-retrieval minds we’ve ever found, what could it do in medicine?”
I want it. It reads constantly, remembers everything it’s read, and understands what it means. Imagine how useful it could be for physicians … and for members of my ACOR kidney cancer community.
I was going to speculate about applications: Watson might say “I see you mentioned Sutent. Last week at a conference in Bulgaria a doctor mentioned a new finding, during the live-streamed Q&A in session B42.” or “There’s a discussion in process in Inspire.com’s Women Heart community, with a reliability rating of 87.3…” (I made up “reliability rating.”)
At the time, circumstances came up and I never wrote the post, but I wondered if it had occurred to IBM. Ha: I should have known: with $2,000,000,000 a year of healthcare spending worldwide, IBM was already on it. Just eight weeks later they announced it, and five months after that (Sept. 12) they announced a major Watson deal with Wellpoint, corporate parent of Blue Cross:
For physicians, incorporating hundreds of thousands of articles into practice and applying them to patient care is a significant challenge. Watson can sift through an equivalent of about 1 million books or roughly 200 million pages of data, and analyze this information and provide precise responses in less than three seconds.
The announcement added this thought:
Watson may help physicians identify treatment options that balance the interactions of various drugs and narrow among a large group of treatment choices, enabling physicians to quickly select the more effective treatment plans for their patients. It is also expected to streamline communication between a patient’s physician and their health plan, helping to improve efficiency in clinical review of complex cases.
Help physicians identify treatment options? Um, it might help the rest of us, too. Especially those of us who are having a hard time affording a doctor visit.
Then this, which may not be welcome by certain providers in your area, but very welcome by others:
It could even be used to direct patients to the physician in their area with the best success in treating a particular illness.
I could go on at length, but you get the picture: a wicked fast system that understands human questions, continuously updates itself and delivers answers… and isn’t controlled by the medical establishment. Extremely disruptive, because it goes around existing vendors (providers) and delivers valuable facts directly to the buyer.
Who are the buyers who’d like that information? Depends on the patient:
- the insurance company, or
- government insurers (everything from Medicare to county health agencies), or
- self-insured employers (who have given up on insurance), or
- self-pay patients. (That’s me, for all practical purposes: I’m on $10,000 deductible insurance.)
How big a deal would that disruption be? Recently at an event in the Midwest, I heard an MD panelist at the region’s dominant hospital respond to talk of shared decision making by booming, “WE will tell you the best treatment! WE will tell you who the best surgeon is! WE are the experts! Don’t get that from some… blog!” He almost spat the word out.
I wonder how he’ll do in Q&A versus Watson. Because it seems that day is coming.
1+2+3: Silkfire Watson
Now imagine something with Watson’s power, thinking out there in the cloud somewhere, with enormous amounts of information about outcomes. Immediate and long-term outcomes.
Imagine it with an architecture like the Silk browser and Amazon’s “thinking cloud” (my term, not theirs). So all that power (and knowledge) gets delivered to however much power you have in your hand (Fire or other), or in your desktop computer, or whatever.
Or to a kiosk in a public health clinic. Health information accessible to everyone.
* * *
I imagine a mashup of Watson with the Amazon Computing Cloud is crazy, but consider the concept. I don’t know if any of it’s going to happen, and I certainly don’t know if it would involve Silk and Fire. Plus, arguably Microsoft’s own cloud computing is better positioned, because MS’s HealthVault has such a good footprint in healthcare. I don’t have a clue.
But on the other hand, I wouldn’t be surprised if IBM’s on the verge of announcing it, same as they voiced my February fantasies within eight weeks. We’ll see.
But really – just think what it would be like if you and your doctors could get the best information, without depending solely on the doctor’s capacity for reading.
Alexander Wilms says
The concept is sound and we will see it (some day). Just so a few things to consider
– putting people’s medical data to the cloud is considered a security and privacy risk (at least here in Europe). Law requires data encryption, but even Watson won’t be able to process on encrypted data
– Once the information is in Watson it may use its incredible capabilities for processing. But how do patient symptoms and objective findings get to Watson? It will have to rely on a physician who establishes this and feeds it into Watson.
– While Watson may be able to find the right therapy for a certain rare illness it does not guarantee that insurance will cover treatment. Especially with the severe budget cuts in healthcare you may end up in a situation where you can tell patients you have found a cure but they won’t get it because of budget cuts.
e-Patient Dave says
Good morning, Europe! I’m so happy to see your responses.
Alexander, I agree with your points. It all depends on benevolent management – and I believe that’s true for every single bit of healthcare.
I’m especially interested in how we will integrate personal data: the feeds from our Withings blood pressure and scales, our Fitbits, our Zeos, our CGM. But I didn’t go down that path here, to avoid making it longer than it was.
(btw, re privacy, encryption, etc: I’m going to change subjects for a moment. In the U.S. our famous TV “columnist” Andy Rooney died recently of “serious complications after minor surgery.” It seems clear to me that this fatal error should be openly discussed, so we can learn how to prevent it, the same as if it were a plane crash with one death. But at the moment, all we know is that it was minor surgery, he died, and nobody will say more. Where is the benevolent authority today?)
Bart de Witte says
Alexander Wilms your comment seems to be a general fear in Europe. Bringing in issues with data privacy, without knowing how the system works, does not seem to accelerate the innovation the healthcare system needs in order to overcome the challenges. Watson does not use or store any patient related data, it just needs data that is relevant for the case. The difference between case and patient related data should be obvious. Then there is also the difference between private and public clouds, not all cloud solution have data protection issues.
IBM always communicated that it does not intend to replace the doctor, but Watson could be used as a decision support system. I sure do hope that the data entry process for Watson will not be limited to the doctor feeding the system, as this would limit the amount of case related data. Just imagine what would happen if the have a longitudinal record of data, which includes data sets such as genomic and lab results.
e-Patient Dave says
Thanks, Bart. I don’t know a lot about what is planned and isn’t planned for Watson so I’m grateful for your notes.
I did expect that IBM wouldn’t want Watson to be the final diagnostic tool – I suspect that will always require a human doctor, presuming one is available and affordable. But in May I proposed that Watson could be the “information valet” for a physician, scanning the entire online planet, and delivering it neatly organized.
And for the people who can’t even afford a doctor, I hope Watson will be made available somehow. That’s 50 million uninsured people in the US, and countless millions around the world. I’m sure nobody in our culture would deny those people access. Think what Watson might do for the productivity of a WHO public health worker.
(I, for one, would gladly say “Sure, Watson, look at my data. Look in my underpants, if you want. Jeopardy, heck; let’s play doctor.”)
See, regardless of what IBM is already thinking, I have a lot of other wishes.:-)
Alexander Wilms says
Bart, it is true that Europeans have a general fear of losing data privacy, and we often wonder about the lax handling of these issues in the US. Actually I earn a living designing software for physicians, I do fully agree that improved decision support is one major innovation of the coming decade and it must become part of the next software generation. On the other hand we cannot ignore people’s wish to keep critical and sensitive data private.
If Watson is to apply it’s knowledge base to real medical cases it needs a set of medical and personal data describing the case; even if that set is anonymous it still may enable a malicious attacker to profile it’s owner – gender, age, weight, height, diagnoses… By adding additional information an attacker may be able to dissolve anonymity – over here our government takes these topics serious, and we really have to discuss with them how to prevent malicious attackers from trailing your connection back to your PC using the trail of IP addresses…
Of course some cloud types may present less security issues, but we can’t rule out even an evil server admin who sells personal data – that happened many times even to Swiss banks, whose account data got stolen and sold to governments searching for black money – if you can sell the data somebody will take his hand on them…
Don’t get me wrong; I would love to have a system like Watson cheap enough to incorporate it to everyday life (and I think it’s not that far away…). But I am able to imagine what happens if my longitudinal set of data, including my genomic and lab tests, is available and it shows a huge probability of me falling ill from a deadly genetic disease – what will my employer and my health insurance company say to this? I really would love to treat this benevolently, but I have my doubts…
e-Patient Dave says
Again it’s very interesting to me that when I wrote this, I wasn’t at all thinking about Watson as a diagnostic too – I was thinking about it as an information valet.
I know IBM’s already advertising Watson (on US television) as a tool to suggest diagnoses to doctors, but that’s not what I’m writing about here. I’m just considering, what if we could remove one major barrier that affects everyone in healthcare, every day: knowing the latest about a medical topic?