A $3m x-prize competition has been launched by the US Heritage Provider Network to develop a ‘breakthrough’ algorithm that can predict and prevent unnecessary hospitalisations.
The $3m prize, open to all comers, will be awarded to whoever can develop and prove a predictive algorithm able to identify patients at risk for hospital admissions. The prize is worth almost twice as much as the Nobel prize for medicine.
The intention of the open competition is to spur innovation in a similar fashion to the x-prize competition to develop a re-usable commercial space vehicle.
The idea is to design a suitable predictive model, so programs and resources can be focused to prevent those admissions – and readmissions.
Teams competing for the prize will have to show that the algorithm developed can reduce hospitalisations using real patient data, including health records and claims data.
Heritage says it believes that the incentivised competition is the best way to achieve the radical breakthroughs and innovations necessary to reform our health care system.
According to the American Hospital Association more than 71m individuals in the US are admitted to hospitals each year. Studies indicate that in 2006 well over $30 billion was spent on unnecessary hospital admissions.
So far, 200 teams of contestants from as far as China and India, as well as Harvard University and Georgia Tech, have expressed interest in the contest, which will officially open in the next three months.
Dr Richard Merkin, president and chief executive of Heritage, said he thinks teams might have the best chance. "The best way to achieve radical breakthroughs and innovation will be to encourage these youngsters to team up, collaborate, and work together to come up with a great technology."
The organisers considered having an entry fee to make sure entrants were serious but decided against it to encourage more people to have a crack at the problem.
Entrants will receive claims information, lab and pharmacy data, hospitalisations, and diagnostic history for 100,000 real but de-identified patients seen in 2009.
Data on patient’s race, ethnicity, weight, age and other socio-economic data will not be provided.
Using this anonymous data, entrants will face the challenge of developing an algorithm to predict which patients were hospitalised.
© 2010 E-HEALTH-MEDIA LTD. ALL RIGHTS RESERVED.
Why exclude the most important?layton 224 weeks ago
Nice idea. But excluding race, ethnicity, weight, age and other socio-economic data means the end result will be pretty meaningless as it is already clearly established that all of these have a major impact on health and hospitalisation rates.
Right objective, wrong premisetimbenson 224 weeks ago
Of course it is right to target hospital admissions, because that's where the money is. Jack Wennberg in his superb book "Tracking Medicine" (OUP 2010) has used existing data to understand that the causes of unwarranted variation in admissions for elective surgery and for long term conditions are completely different. The former is preference sensitive, the latter is supply sensitive.
Wennberg showed how much could be done with historical claims data, but it has serious defects. It was collected primarily to support payment claims and refers only to previous healthcare activity. It is not up to date and is silent about patient outcomes, the patient's own perception of their health status.
The Heritage prize might make even more progress if it widened its scope to encompass data that is not currently collected, but probably should be, such as each patient's own perception of how they are feeling and how much they can do.
CPM ResultsAndytheAnimal 224 weeks ago
I've just written a paper on this very subject and the King's Fund have no quantitative evidence that the models a) predict the correct patients for intensive case-management nor b) that intensive case-management in turn reduces the likelihood of an emergency admission.
Whilst there is some research on the latter, nothing can be unearthed on the former either here or in the US yet PCTs seem happy to utilise the likes of United Health to reduce admissions with no quantitative proof that it works - someone please direct me to figures if they exist?!
X-prize for NHS ITJon Hoeksma 224 weeks ago
If you had £3m spare to fund an X-prize style open competition for NHS IT what would you run it on?
Would it be on a breakthrough algorithm to prevent avoidable admissions or would it be something completely different?
Quick win for the UK?Lyn from Digital Health 224 weeks ago
The King's Fund (and then the Nuffield Trust) led work to develop "predictive risk" algorythms a couple of years back and made the results freely available to the NHS. So perhaps the UK would be well on track to win this prize, if the work were taken up again? More interesting questions might be: did those PCTs that adopted the algorythms see much benefit. And if they did, what were the barriers to others using them>