Fine-tuning treatments using data mining and pattern matching
Hoping to customize medical treatments to individual patients--such as picking the best chemotherapy for a cancer patient with a specific genetic marker--Mayo Clinic and IBM are applying pattern recognition and data mining to the electronic records of about 4.4 million Mayo patients.
For the last three years, Mayo and IBM have been working to replicate into one database clinical information from five data sources that includes digitized patient files, lab results, X-rays, and electrocardiograms from Mayo's hospitals in Arizona, Florida, and Minnesota. In our lifetimes, each of us generates an estimated 1 terabyte of medical data.
The two companies last week said they're extending that project to apply custom algorithms to identify patterns that could lead to better, more personalized treatment or help researchers develop therapies. The goal is to find patterns --based on age, medical history, genetics, and other factors--related to how patients respond to various treatments and adapt care accordingly. Such applications might be available to practicing Mayo physicians by the end of the decade, says Drew Flaada, director of the IBM and Mayo collaboration and IBM life sciences.
IBM CEO Sam Palmisano and Mayo CEO Cortese with Blue Gene model.
Mayo researchers also will use pattern-recognition tools to search patient data for relationships among particular proteins, genetic makeup, and responses to specific treatments, says Dr. Nina Schwenk, a Mayo physician and chair of the Mayo Foundation IT committee. Schwenk tells of a situation in which a promising new chemotherapy treatment for lung cancer ended up producing disappointing results and was nearly abandoned. "The majority of patients didn't do well," she says. Then a researcher discovered that a small percentage of patients--all with the same rare genetic marker--responded extremely well to the drug. Now the treatment is used for lung-cancer patients with that genetic trait, she says.
With the help of the pattern-recognition and data-mining technologies, Schwenk believes many more such discoveries and correlations will be uncovered. Also, because medical knowledge is evolving, IT tools eventually could help doctors at the patient bedside make care decisions based on the latest findings and breakthroughs.
Parallel with the clinical data-mining and pattern-recognition project, Mayo Clinic will be the first medical institution to use IBM's Blue Gene supercomputer for molecular modeling in disease research. "We're at a point with standards in technology and new genomic-based analytic techniques where we can achieve more in the next 10 years than we've achieved in the last 100," Mayo CEO and president Denis Cortese said in a statement.
Personalizing health care is one of the most significant potential benefits from moving to electronic medical records. In his national plan for medical IT unveiled in July, federal health IT coordinator Dr. David Brailer cited personalizing care as one of four main goals, part of giving consumers better access to their own medical information.
Mayo patients have the option of whether to allow their data to be aggregated for analysis; so far, 95% of patients have agreed to do so, Schwenk says. She notes that the replicated database is called "Data Trust," to remind everyone that the project depends on protecting patient confidentiality, privacy, and security. The majority of the pattern recognition and data mining will be applied to aggregate data, and only a few select individuals at Mayo will have access to patient-identifiable information.
We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.