CHEO Research Institute Logo
Decrease Text SizeIncrease Text SizeFacebookTwitterYoutube

 
What's New rss


19/4/2011

CHEO Researchers Protect Privacy in Heritage Health Prize Contest

OTTAWA - More than 2,000 participants from 52 countries are competing to create the best algorithm to predict patient hospitalization in one of the largest-ever health data mining competitions. Privacy experts from the Children's Hospital of Eastern Ontario (CHEO) Research Institute ensured the anonymity of this data before it went out.
 
The California-based Heritage Health Prize (HPN) is inviting teams from around the world to develop an algorithm to predict the number of days patients will spend in hospital within the next year. The expectation is that the innovations emerging during the two-year competition will produce a prediction model that allows healthcare professionals to proactively engage patients before a hospital visit is required, resulting in improved patient health, reduced strain on hospital resources and savings for the healthcare system. The winner gets a $3 million cash prize.
 
Anthony Goldbloom, the CEO of Kaggle Pty. Ltd, the company hired by the Heritage Provider Network to run the predictive analysis competition, notes that "great good can come from engaging many smart people to solve healthcare problems, and we would not be able to do this unless we can ensure that the identity of the individual members in the data is protected based on current legislation and best practices."
 
Health data can be used to address persistent but solvable problems in the healthcare system. But making large data sets available to data miners and researchers from all over the world presents complex privacy challenges. Dr. Khaled El Emam, at the CHEO Research Institute, and his team in the Electronic Health Information Laboratory have extensive expertise in dealing with the privacy problems inherent in such large data releases. They developed new anonymization methods which they applied for this important competition.
 
“When personal health information is properly anonymized and made available through these kinds of data mining competitions, there is a vast potential for unexpected solutions to emerge from the analysis that gets done,” says Dr. El Emam. “Our de-identification methods provide strong guarantees of patient privacy while ensuring access to the optimal amount of data.”
 
Jonathan Gluck, executive and legal counsel for Heritage Provider Network, said that Dr. El Emam‟s software and know-how were relied on to empirically measure the privacy risks involved in handling this massive data set of insurance claims.
 
"This competition has the potential to change today's sick care system into a true „health care system‟,” says Gluck. “HPN, however, is acutely aware of healthcare data privacy concerns, and takes them very seriously. Indeed, HPN wanted to assure that it went beyond its legal obligations when it released the data and this was the foremost reason it looked to Dr. El Emam to help anonymize the data."
Take Action
Quick Links

Our Researchmagnifying glass

abcefg hijklmnopqrst uvwxyz