Sunday, February 21, 2010

What happened to Deep Blue?

Check out this amazing article written by Kasparov, the chess grandmaster who battled Deep Blue, the supercomputer built by IBM to play chess. Kasparov's career spanned a period when computers progressed from a weak chess player to an unbeatable one. By observing how computers quickly overpowered the best grandmasters we have to offer, Kasparov drew important insights on the human minds and artificial intelligence.

The most interesting finding made by Kasparov came from his observation at a freestyle chess tournament in 2005 where anyone could compete in teams with other human players or computers. A relatively strong human player teamed up with a weak computer quickly overwhelmed the best grandmasters or even the strongest supercomputers. The strength of human strategic planning combined with accurate calculations from a computer teammate was unmatchable.

At this point in human history, a chess grandmaster will have a tough time beating even an average chess program - humans have succeeded at building a machine to play chess. As the field of computer engineering moved on to other problems, Kasparov lingered on and noted that the chess computers we have built was nothing similar to what engineers had in mind when they decided to build machines that play chess. We wanted to build computers that play chess and think like a human - an artificial intelligence of sort. Instead, what we have now is computers that win by the sheer brute force of calculations - they assign numbers to pieces and positions, process a whole lot of numbers and possible moves, then pick the best one. They have no imagination or intuition, only processing power.

In my head I struggle between "complement" computers and artificial intelligence. Building machines that complete tasks by complementing our weakness is definitely helpful - calculators was a humble but powerful start. But do we stop there and leave imagination to humans, or do we venture further to create computers that can think for us? What are the ramifications of the world where machines can dream? If our strength is the power of imagination, what happens when we give this gift to our computer counterpart? Computers that mimic the human minds will revolutionalize our productivity, but then what is left for humans to do?

Thursday, February 11, 2010

Ignored common sense

Going through medical school, I have spoken with many mid-career, rising faculty members at various institutions. I noticed that these somewhat prominent faculty physicians somehow made it without the most simple of common sense in human interaction - learn the names of the people you're talking to and listen to what they say.

It is so important that I will repeat it again - to achieve a desirable result with others, it is of utmost importance to:
1. Know their names, and
2. Listen to what they say

Sounds simple enough, but it is mind-boggling to learn that many successful people fail to do this, especially in the medical field, where everyone (patients AND providers) somehow thinks it's ok to be nasty to each other in a way that wouldn't fly outside the hospital. This is another area where other professions lead the medical field by leaps and bounds - Any good politician/leader/businessman knows the importance of these two etiquettes - it is how Joe-the-Plumber got media's attention. Most people follow these rules well when they talk to a more important person - it takes real character to follow these rules when talking to someone less important, and these faculty physicians have failed miserably in my opinion. By not learning my name and listening to what I'm saying, they come off as arrogant and uncaring, none of which makes for a good educator/physician.

Read more in "How to make friends and influence people" by Dale Carnegie, a time-tested book that creates successful leaders and businessmen over decades, but really it is something that everyone should do on a daily basis, aspiring businessman or not. Know people's names and listen to what they say - other human beings deserve this much from us.

Fun with HospitalCompare


I have mentioned HospitalCompare in many posts in the past - it is very enlightening to play with. In a prior post I claimed that elite hospitals in MA did not perform better than cheaper community hospitals in NYC despite the higher exorbitant prices charged. To substantiate that claim, we will use HospitalCompare to look at 4 hospitals:

1. Brigham and Women's Hospital
2. Massachusetts General Hospital
3. Bellevue - a community hospital in NYC attached to NYU school of medicine
4. Mount Sinai Hospital - a somewhat elite medical institution as a NYC control for Bellevue

Follow the instructions below to get outcome information for these hospitals:
1. Go to HospitalCompare
2. Click "Find and compare hospitals" button
3. Choose "Find a hospital withing a distance of a City"
4. Type "Boston" into the City box, select Massachusetts as the state, and select "In this city/zip" for distance. Click "continue" button
5. Choose "General search". Click "continue" button
6. Scroll down and choose Brigham and Massachusetts General Hospital. Click "compare" button.
7. You will have the same tables as posted in the photo above. Repeat the same process for New York, choosing Bellevue and Mount Sinai hospital for comparison.

As we scroll through the tables, we can see that Bellevue does similar if not outperforms all these elite hospitals in most measures except for pneumonia outcomes. It's stunning to see that %door-to-balloon time (D2B) at Brigham is as low as 87%, while MGH does even worse at 79%. This means that in a patient with ST-elevation myocardial ischemia (heart attacks) in which one of the most important prognostic factors is receiving cardiac catheterization within 90 minutes, only 87% of the patients get that optimal treatment at Brigham, 79% at MGH. For comparison, Mayo clinic achieves 100%, Kaiser Foundation hospitals in San Jose got 98%, in Santa Clara got 94%. Both Mayo and Kaiser have been applauded by Obama and the medical field as high-performing, cost-effective systems that do not charge nearly as much as the two elite Boston hospitals.

Overall, Brigham and MGH do not appear to be national leaders in any of these process outcomes. Maybe they are better at other things, but these outcomes are simple, objective measures that have proven to improve outcomes regardless of sickness or patient population - hospitals do not have any excuse not to excel at these outcome measures. It will be interesting to hear alternative explanations why these elite hospitals deserve the exorbitant amount of reimbursement squeezed out of the over-strained MA health system.

Wednesday, February 10, 2010

GoogleBuzz! Don't knock it and wait for tweaks

Buzz has good potential - integration with Google products is key. With a few tweaks it can potentially take over facebook - user profile should contain more details/personalization, the interface could be more attractive, and lack of targeted advertisement is nice. Taking over twitter is more difficult because it is a completely different platform - word limit sets it apart from facebook-like social media, making twitter more succinct and less personal, which means strangers feel comfortable tweeting to one another. However, this can be worked out easily with the option to post publicly and the option to search/follow users that are not in your Gmail contact. Overall, an interesting start!

Tuesday, February 9, 2010

What are we paying for?

Massachusetts Attorney General Martha Coakley recently released a report which showed that prices paid to healthcare providers in MA vary significantly based on market leverage/position - Partners Health Care (consisting of two biggest healthcare bully in the state - Brigham and Women's and Massachusetts General Hospital) being the exemplary culprit. Due to size and its brand name, Partners Health Care is a successful oligopoly that has pressured other Boston hospitals down a competitive, cost-inflating path, charging exorbitant costs and initiating medical arms race including acquisitions of surgical robots which have not been shown to be superior to human surgeons.

More importantly, Coakley's report showed what the higher price tags DID NOT correlate to. Price variations do not reflect:

(1) higher quality of care
(2) the sickness or complexity of the population being served
(3) the extent to which a provider is responsible for caring for a large portion of patients on Medicare or Medicaid
(4) whether a provider is an academic teaching or research facility
(5) differences in hospital costs of delivering similar services at similar facilities.
HealthBeat broke the report down into chewable pieces here.

In the medical field, Partners Health Care hospitals set the bar in medicine. They train the best and brightest residents and house world experts in many medical fields - their faculty members write game-changing articles and author respectable medical sources like Uptodate. In my personal experience, working in a Partners hospital resemble working in a perfect bubble - all the medical staff were top-notch, the wealth of resources made the most cutting edge tests/studies possible, almost all patients were insured, white, English-speaking middle-class who could advocate for themselves.

It's counter-intuitive to think that this perfect medical setup does not necessarily improve outcomes, but Coakley's report is not alone - try playing around with HospitalCompare and you'll find that Mass General or Brigham have worse outcomes than a cheaper institution like Bellevue, a community hospital in NYC. It is important to note that the outcomes measured are simple, objective yardsticks that should apply to all hospitals regardless of the complexity of their patient population. For example, smoking cessation counseling could and should be given to all smokers regardless of how complicated their sickness is.

Some may argue that these measures are too simple and do not reflect the ability to treat complicated cases that are often referred to brand name hospitals, like saving patients from an extremely rare malignancy. In medicine we have a saying that if you hear hoof sounds, they probably belong to horses and not zebras. Curing extremely rare malignancies is a zebra - most cases are horses of patients coming in with a heart attack, a common condition that most hospitals can treat effectively without a perfect setup, the most cutting-edge medical technology or world experts in cardiology, which is how things should be. Any average medical facility should be able to treat common conditions as well as more famous institutions, because that is what an average hospital should be able to do - treat common conditions. Not everyone can or should go to Brigham for an x-ray of a fractured wrist - we have gotten so good with this age-old technology that most hospitals can produce the same, accurate results, for much less than what is charged at brand name hospitals.

What we need to realize is this: most common causes of morbidity and mortality are, for lack of a better word, common. What we see in this healthcare system are horses, not zebras, and because they are common, we have devised cost-effective, time-tested treatments that we should use instead of shiny, experimental treatments that cost much more without proven benefit. An inflamed appendix can be as safely and effectively removed by human surgeons, much more cheaply than using a robot. This is why brand name hospitals providing the most advanced medical technology are not necessarily better for your health or your pocket.

In medicine, less is more. A good doctor should be good at keeping you out of the hospital - they should not aim to excel at providing the most invasive of treatments often advertised by marquee hospitals, but unfortunately doctors and hospitals get paid much more doing the latter. The only thing preventing this blatant conflict of interest from completely wrecking our health care system is human conscience, which is not without fallacy. A better system would reward outcomes, not volume or technology used. Until we get a new carrot on the stick, our healthcare system will continue to be overused and overpriced.