▲ 36 points
back
56 comments
Healthcare technology has always been a siren song for the Silicon Valley. The thinking repeats over and over:
1) It is an antiquated industry
2) There has to be a better way to do it with IT
3) My company has the magic solution that will save everyone and make me rich
You can make the best ML algorithm to diagnose and treat disease and doctors will spend decades arguing about why it is unsafe and they should always be the final authority. You won't be able to get a diagnosis and a prescription without them. You might think otherwise, but they have real power in denying you medical choice via setting professional guidelines through medical boards and lobbying politicians who are simply not brave enough to pit themselves against doctor's medical judgement.
It is harder to think of a more counter-productive and self-defeating way to market any healthcare solution then by saying it is going to "replace doctors". Here is a great example (DNA testing) of how doctor lobby can restrict innovation: http://www.nytimes.com/2010/03/20/business/20consumergenebar...
EDIT: I actually believe Vinod Khosla knows better and he just made a troll statement on purpose. It worked, we are talking about it!
This is pretty spot on ( and I would not discount Vinod trolling ). We know healthcare can be done for less because folks like Doctors without Borders provide health care on a much smaller budget. But it is very hard to reconcile health care providers, health care insurers, and health care outfitters which have such a cozy relationship. Breaking all of those bonds at once is political suicide and thus not often embraced.
Its not spot on.
This debate is similar to debates about computer automation taking away jobs. The final result is, automation is inevitable. The biggest losers are those who resist it, the winners are those who learn before time and move to something else more productive. The same argument about unions, suppliers, OEM's etc can be applied to manufacturing industry. But delta improvements in robotics have been replacing humans and will continue to. There is little you can do against bit by bit delta automation happening every day. You can't protest against it, you can't stop it. But when look at it in a decades scale timeline you will see it is likely to be changing the whole scenario like never before.
Heck in our own industry IDE's have decreasing barrier to entry rapidly. The quality of skills required to write programs today is far lesser compared what it was two decades back.
You could argue every point in the world about why the doctors say in a judgement must be final. But that point is true if you prove doctors themselves don't make mistakes. Doctors make mistakes all the time. In fact ordinary down-the-lane doctors years into their practice aren't anymore helpful than what vinod describes them to be.
The point he is trying to is not that there would be a massive change tomorrow and doctors will be out of jobs. But automation in this area is inevitable. How many testing labs have lost blood sugar testing business to glucometers? That sort of innovation is inevitable. There is no point fighting it. Human beings will inevitably move to a higher levels in the productivity curve endlessly.
If the doctors don't want that to happen, they have to work to be better than machines. Not expect machines to be dumber than them.
I think a better strategy than to "replace 80% doctors" is to help doctors "replace 80% of their tasks". So focus on a business that makes a doctor's job much easier by identifying or confirming a diagnostic, or helping him reach a diagnostic much easier. The point is to get doctors to want this technology, rather than fight against it.
There are worse sources of waste in the health sector. In Sweden, and most other developed countries I imagine, the introduction of IT has resulted in less patient time simply because doctors are now legally required to work on filling out forms, documentation and other acts of IT powered bureaucracy. What's astonishing is that we don't know if all this documentation is doing any good, it's just assumed that it does.
So startups don't have to work on hard AI problems to make it big in the health sector. Removing the bureaucracy caused by IT is a much more attainable goal provided you're ready to fight policy makers who have created the regulations in the first place.
But we do know that good IT implementation do help:
"There is significant evidence that CDSS can positively impact healthcare providers' performance with drug ordering and preventive care reminder systems as most clear examples." (from a literature review)[1]
The other 20% of the tasks that aren't easily automatable or done by a computer ironically are often the tasks that doctors have already delegated to nurses (RNs and NPs).
Personally I see a future of healthcare with more nurses and fewer doctors. Technology bridges the knowledge gap between nurses and 80% of doctors.
Don't forget number 4 - sheer hubris on the part of people who think that just because they've been to a doctor that they know what a doctor does.
Misdiagnose your hypothyroidism for months, only testing you once you've done your own research and pointed it out to them, and then signing their name to a piece of paper and getting you to sign your name to a large check?
</bitter>
The computer has decided you don't have hypothyroidism either, the doctor you can at least persuade otherwise.
I've read my chart. It's full of the doctors' logic based on facts. Computers are good at that. Medicine isn't magic.
I've read my charts as well, they're made up of a mix of facts, intuition and conjecture. The later two being things which computers aren't very good at.
Medicine isn't magic, but it's also more than a checklist which gives you a definitive answer at the end, that's why the process of becoming a doctor is as much an apprenticeship as it is rote learning.
You don't have to go through doctors to improve the healthcare system. You can around them, to the health insurance companies that pay their checks.
Insurance companies are much more interested in statistics than doctors are, especially predictive models. If someone can find out how to use technology to make health insurance companies more money, the doctors will follow.
That is what IBM is trying to do with Watson initially.
These constraints are US-specific. Have you heard of Narayana Hrudalaya?
http://knowledge.wharton.upenn.edu/india/article.cfm?article...
The US healthcare "market" is completely distorted ranging from no one having any idea what anything actually costs, multiple layers of disintermediation from consumers to payments, a doctor's cartel that keeps numbers low thereby increasing compensation, but giving an incentive to replace them, income not based on outcomes, the government spending ~50% of all healthcare money but everyone ignoring it, and the list goes on.
Atul Gawande's articles in the New Yorker are fascinating insight into how things really work - full list at http://www.newyorker.com/magazine/bios/atul_gawande/search?c...
Some highlights:
Reliably delivering care (and cheaper) http://www.newyorker.com/reporting/2012/08/13/120813fa_fact_...
Pit crews http://www.newyorker.com/online/blogs/newsdesk/2011/05/atul-...
Hot spots http://www.newyorker.com/reporting/2011/01/24/110124fa_fact_...
Costs http://www.newyorker.com/online/blogs/newsdesk/2010/12/the-c...
Hey man!
You forgot the fact that despite the push to fix insurance, the price of pharmaceuticals won't be pushed down anytime soon. Since the cost of health care has much more to do with the drugs than what hospitals/doctors charge, but nobody likes it when that is mentioned.
http://www.nytimes.com/2009/08/08/us/08lobby.html
If anybody can find a relatively recent estimation of how much our medicare obligation is pharmaceutical spending, I'd love to see it.
Pharmaceuticals represent approximately 10% of total health care expenditures versus hospital, physician and clinical services representing approximately 51% of total health care expenditures (for the latest estimated year of spending, 2010) [1].
Within Medicare, prescription drugs account for 11% of total Medicare expenditures and are projected to be 13.5% of total expenditures by 2021 [1].
Further, the prescription drug benefit ("Part D") has cost about 30% less than estimated by the CBO when the Medicare Modernization Act was passed (2003) [2]. This is due to a variety of factors, but a huge one has been lower-than-expected growth rates in per capita spending on prescription drugs, estimated at 4% per year since 2006 [2], much lower than overall growth in health spending.
[1] http://www.cms.gov/Research-Statistics-Data-and-Systems/Stat...
According to [1] we spend only 16% on drugs and devices. Maybe they didn't include meds you get at a hospital but still it seem that medical services take a big share of the cost.
Also, assuming technology can give better treatments , it can lower the drugs needed and their costs.
[1]http://www.examiner.com/article/health-care-reform-historica...
One thing I've been thinking of lately is whether there are technologies that could fundamentally change the incentive structure of the US healthcare market to be more in line with patient interests. Are there technologies or attitudes or approaches that could make feasible new business models that would accomplish such a change?
I don't know, but it's something that's been on my mind.
It isn't just individual patient interests but rather everyone. For example providing sporting facilities at schools and exercise/active activity facilities for everyone would do a lot to improve overall group health. Making cities more walkable would do a lot too.
But right now my primary care doctor (who is most valuable to me) gets paid zero when I am healthy and more the more tests and procedures she does. Other specialists only get paid when I am referred to them. Drug manufacturers only get paid when their items are prescribed. Essentially compensation is based on consumption not on wellbeing which is why the system is so messed up.
The US market definitely needs to be split into two. A baseline coverage that is based on wellbeing, effectiveness and does take price into account, done by the government available to all residents. It is easiest to gather the money via taxes (which already happens to a large degree) and to provide the actual care via combinations of government employees/facilities and contractors. And a top up private system where the people who want to pay for or provide gold plated coverage and procedures can do so (in addition to the base).
Most countries have different systems which amounts to running experiments in effectiveness and efficient. Heck worst case just do what the French do http://en.wikipedia.org/wiki/Health_care_in_France
Your primary care doctor does not get paid for tests and procedures. Labs and specialists do.
Are there technologies that give advantages to large groups of doctors over small clinics or make it preferable to doctors to create large groups ? Large groups have access to a large pool of customers so they can offer different business models.
Are there technologies that enable medical "self-service" or less dependance on doctors ?
I think medicare has started a "fee for package" program(instead of paying for services, you pay for a complete treatment around a single medical event - let's say knee surgery + including all services - for a single fee). Are there technologies that can help such implementations ?
Are there any technologies that can help places more receptive to medical change(like africa[1])? with time the experience gained there could be integrated in other places.
How can you make medical tourism a more popular choice or better choice? if it would be popular , more insurance companies would offer it and would pressure the medical establishment for improvement.
studying the process that resulted in https://questions.cms.gov could give you some insight into what makes up the health care costs of today. Lookup coding for a first look at why things are the way they are.
No Vinod. Technology will make doctors more efficient so they can handle 3x-5x the workload and serve more patients.
I've known doctors/nurses who answer emails both between other health professionals, and between patients. Sometimes a 2-minute email can take the place of a regular checkup for somebody with a chronic condition.
Where technology can make drastic improvements is detecting problems when they are minor, and not becoming major. If a primary-care doctor can only do 5 minute consults, and that results in missing a major diagnosis like cancer, which is then treated at a later stage, then they've blown out any cost saving by rushing consults.
Thinking of innovation as coming from people outside the industry or people inside the industry is a false dichotomy.
I reckon that one of the catalyzing factors will be smart people who naively go into school to study medicine thinking that is the career for them, but drop out of that career path just before, during or just after their residency, either because they can no longer imagine living the medical lifestyle or because they realize how backwards it is. Those that drop out will know more enough about what is wrong with the industry but not have the horse blinders that those immersed in the industry for years will have.
These people and those who work with the medical industry but are doctors are the people the medical industry discounts.
The key word here is "eventually". Easy to prognosticate with no timeline.
That being said, technology will likely eventually disrupt/replace many, many current industries/jobs/practices. This isn't news.
The last three times I've contacted Kaiser (In California) to get a doctors appointment, (Once, when I had stabbing chest pains and my left arm had gone completely numb - I seriously thought I was having a heart attack) - their expert system (staffed by a human, still - but that will change) - asked me a series of questions, and came to the conclusion that I didn't really need to see a doctor. "Drink some tea with honey for a cough", "Stretch your arm, and take breaks from the keyboard" for the numb arm. Oh, and "Stress Pains"
So - some significant number of doctors are already starting to be replaced by the experts systems at Kaiser right now.
Kaiser is a rare gem in the u.s. healthcare system, and the reason is that they are integrated system: they integrate insurance with treatment.
Since Kaiser if payed a fixed sum and not a fee for service it has motivation to reduce costs, hence the expert system.
Patients at Kaiser also stay for the long haul(who wants to switch doctors?), so it makes sense for Kaiser to invest in prevention, and they do.
Finally, Kaiser is an NGO, which i'm sure helps to keep they're priorities straight.
Maybe Kaiser is a good target for startups ?
Here's a recent anecdote:
A few months back I got a stomach ulcer and before going to the doctor, I decided to do some research. I came to realize that the primary cause is not stress or excess spicy food, but a particular stomach bacteria, H. Pylori, going rogue and attacking the mucus lining of the stomach that protects underlying tissue from the strong acid [1]. For this discovery, Drs. Marshall and Warren were awarded the 2005 Nobel prize in medicine [2], as it meant that ulcers were curable with the use of suitable antibiotics.
However, my doctor (and good friend) insisted that there was no permanent cure for stomach ulcers and they could only be managed by taking acid reducing medication (Prilosec) everyday for as long as I lived. I couldn't help wondering if his information was corrupted by big pharma, who always want to keep you hooked onto some drug or the other. So, I had to use extra persuasion and tact to get him to prescribe antibiotics for H. Pylori elimination (and thus permanently curing the ulcer).
I'm a bit skeptical that big pharma's tentacles will NOT reach Watson's doctor successors [3].
[1] http://en.wikipedia.org/wiki/Peptic_ulcer#History
[2] http://en.wikipedia.org/wiki/List_of_Nobel_laureates_in_Phys...
[3] http://www-03.ibm.com/innovation/us/watson/watson_in_healthc...
This is because your average doctor is no match to a Google search.
In my own case it has been true. I lost around 35 kgs rapidly to tropical sprue around 3 years back(By God's grace- I'm healthy now). None of the doctors could diagnose it. In fact most of them so badly managed my situation, I had to finally decide to work on it myself. Needlessly to say simply researching the topic on Google gave me far more perspective than they ever had.
There were even a few days when I could tell them straight on their face, why they weren't going with a particular medication with existing ones. I even described to a doctor how a particular anti biotic works, and why the one he is giving regularly to patients is not working. I showed them their flaws in diagnostics, results and clinical procedures.
Nothing that I did was magic. It was all decision procedures, if-else scenarios with some careful reading and analysis. The point is no doctor hears to your problems completely. They have preconceived ideas of what your problem might be and go with that. They cut you off while you are speaking and that is all it. You are their mercy now. Many of them just experiment with trial and error methods.
Some just prescribe what the salesman from the pharmacy company sells them.
This is what your ordinary doctor down the lane is. He is not god/super human who can do miracles to save your life.
10 years into practice your ordinary doctor knows nothing more than what you would know if you read Merck Manual of diagnosis and therapy: http://en.wikipedia.org/wiki/Merck_Manual_of_Diagnosis_and_T...
If you can code up that Manual as software- I am not sure if 80% of them can be replaced. But I can tell you the number will pretty close to more than 50%.
I am not a doctor, but I have visited many. With machine learning and big data analytics, I can see why machines can be better than 80% of the doctors out there. It may be possible for simple iPhone compatible devices to take the vitals, send the data to a web service, which returns a diagnostic that is more accurate than what most doctors can come up with.
It is necessary to stir up the industry with bold statements. Go Vinod Khosla!
Vitals are for monitoring you are vital. I'd think they have very little to no meaning for the majority of diagnoses a GP makes daily. There are many diagnostical tools where computer analysis can be tremendously helpful (like EKG), but most of these are very much out of the iPhone-accessory league.
I know you were just giving an example, but any practitioner who relies on a computer to interpret an EKG is not one I would trust. Many times my own EKG interpretations are far, far more accurate than whatever the monitor thinks is wrong
All this indicates is that the monitors aren't good enough yet. My understanding is that EKG interpretations are a function of rules and probability. Computers do those things well. What are you adding that computers can't? Experience? That's probability.
Do you know a primary-care doctor who can properly interpret an EKG? Most can't beyond the basics. If it is for cardiac diagnostic purposes, then the recording is sent to a cardiologist for diagnosis.
I have seen a room full of doctors with licenses and who are generally competent their in practice, who can't read an EEG either. They were not trained in medical school/residency to read an EEG and so their skill is limited. The neurologist with an EEG fellowship does the interpretation. This is after it is digitally processed by software that does the quantitative analysis.
Note: I am not a physician, but I'm quite familiar with EEG technology including Quantitative EEG analysis.
Challenge #1 is to get sufficient data. There are surprisingly few medical data available that have sufficient parameters to let you do any useful training.
I'm always somewhat leery about these sort of predictions since to my mind they seem to miss that being a doctor is often more than just making diagnoses and deciding on a treatment - it's also about having a relationship with the patient, especially when it comes to primary care practitioners (like GPs in the UK) and specialists for long term conditions (such as my father, an oncologist, who has patients he's been seeing for ten or fifteen years since their initial diagnosis and treatment).
Not withstanding that people are generally more comfortable and willing to submit personal information to a fellow human who they know and trust, there's also things like how does a computer tell the difference between a little old lady with a problem, and a little old lady who's just lonely and uses going to her GP as a way to get some human contact every few weeks. Her GP will likely know the difference though, because she's known the little old lady for the last five years and has realised that her visits aren't about the vague symptoms she's presenting, but the ten minute chat she has with the doctor.
>> people are generally more comfortable and willing to submit personal information to a fellow human who they know and trust
There was some research(can't find it now) , that show that filling an automated form is a good way to make patient elicit personal embarrassing info that they wouldn't offer via chat.
>> difference between a little old lady with a problem, and a little old lady who's just lonely
First that's a nurse can easily do. And maybe, hopefully we'll find ways to solve the loneliness issue.
Proper care of the elderly is another problem that society should address without racking up insane health care costs just so a little old lady has someone to chit chat with. There are certainly more fruitful (and cheaper) ways to solve that.
I look forward to the day that machines can practice art the way doctors do. One of the most underrated/underappreciated aspects of being in medicine is realizing it is a true amalgamation of art and science, relying heavily on both education, experience, wisdom, and instincts.
Being in medicine, I don't agree.
Over 80% of the job could be done by properly designed algorithms - and 80% would be a conservative guess. The "interesting" cases are very rare, and mostly consist in easy-to-do diagnosis of unfrequent illnesses (I once diagnosed an insulinoma! hurrah!)
Just look at the growing number of tasks given to non-MD- say RN. It's a trend that will go on.
Human knowledge is nothing but the application of algorithms - deciding which symptoms are important, which are to be discarded, which require more investigation, etc.
Education is building algorithms. Experience is machine learning. Wisdom is statistics. Instinct is prioritization.
The faster we can implement all that in software, the better.
It seems to me that things like "intuition" and "instinct" are just subconscious algorithms that you can't articulate (example instinct: I pull my hand away from a fire; I can't tell you why, but it isn't magic; somewhere in the wiring of the brain there is an algorithm that makes that happen).
Fundamentally, the brain processes information to make decisions, which machines can always in principle be designed to do. Thus as long is medicine is fundamentally about making diagnostic and treatment decisions, it can be solved by machines.
hurrah for insulinoma diagnosis:)
I think that even rare disease diagnosis can be done better by algorithms by using Isabel healthcare's diagnostic system[1] or Watson(in the future).
The place where a human will contribute the most is the interpersonal stuff:motivating people, digging data from people(although computers compete there[2]) and administering placebo.
And of course doing the physical stuff.
But that seems very different from the job of a doctor and closer to the job of a nurse.
[1]http://www.isabelhealthcare.com/home/default [2]There was some research that showed that people give more personal information to a computed form than via face to face chat with a doc.
You present an excellent argument! Integration of software/computing with healthcare/medicine is what I hope to focus the rest of my life on, so I surely hope we can make significant progress in that area.
As a fellow doctor I agree with everything you say.
What's your take on checklists?
As an investor, Vinod Khosla's job is to make such bold statements. But his statement will never come true.
Even if you assuming that 'doctors' and 'humans' are old technolgy, whenever a disruptive technology has come to market the old technology has not gone away -- it is still used and is often a part of a growing market (this is partly why incumbents are often so slow to move).
Regardless, I think the right way of phrasing the above is that in the future 80% of medical needs will be met by technology. This statement is actually different from implying that 80% of doctors will loose their jobs, it is more that technology will help them do their jobs - either to service more people or to service more complex situations.