By John Russell
July 11, 2002 | At least one HMO has contracted with an informatics firm to cull its vast patient record database to identify which drugs are most effective for specific patient segments, and it seems likely that other HMOs will soon follow.
“We are six to 12 months from delivering results to this customer, and are in the advanced stages of negotiations with several others.” said Mor Amitai, president and CEO of Compugen, during the BIO 2002 conference in Toronto last month.
This kind of patient record data mining is likely to increase friction between HMOs and pharmaceutical companies about which drugs to prescribe. It may also represent a new business opportunity for the battered informatics industry, which has faced slumping software sales, forcing many companies to turn to services or drug discovery initiatives to survive.
Compugen has labeled the practice, “computational medicine,” which it defines as the attempt to divide populations into subsets according to drug efficacy and adverse side effects. Currently, the data mining is based on phenotype and clinical trial data but may include genotype data later. The company says it has access to millions of patient records from the HMO it is working with and that it has struck another deal that gives it access to clinical trial data.
Amitai declined to name the HMOs it is working for, and other informatics firms contacted for this article indicated they did not have similar projects at present. The idea isn’t new, but technology stumbling blocks and debate over patient privacy concerns have slowed efforts to actually embark on such patient record data mining projects.
“I’ve not heard of this particular practice before, but if it provides more effective medicines, it’s a good thing, and it could help contain costs,” noted Caroline Kovac, general manager of IBM Life Science Solutions. IBM is working with the Mayo Clinic in Minnesota to create a database of 6 million patient records (see June Bio-IT World, page 1).
“The HMOs have the information but it must be consented information. A spokesmen from one clinic recently told me they now have blanket consent for patients, however for every new study we must go back to the IRB [institutional review board],” said Kovac.
Compugen’s effort began a year and a half ago with a small research group and has since swelled into a major initiative. “We have very promising results,” said Amitai. “In each patient record there is a huge amount of data and it was not written to be easy for computers to understand. Merging the different data types hasn’t been easy.
“The HMOs are extremely interested in knowing which drugs are effective and which are not. The HMOs and pharmaceutical companies look at this differently. Maybe there are four drugs used for a particular indication, but the HMO wants to choose just one,” says Amitai. “It may not be in the pharmaceutical companies’ best interest to find these sub-populations which benefit most from a particular drug.”
Computational medicine, for now, is a small part of Compugen’s business. Its two main businesses are computational biology (drug target identification and validation) and computational chemistry (merging computation with experimental organic chemistry).