Providers, Researchers Turn to Data Mining To Inform Care Decisions
October 7, 2014 in News
Health care providers and researchers increasingly are analyzing existing troves of medical data to identify treatments and potential complication risks for rare medical conditions, rather than relying on large, time-consuming and costly clinical trials, New York Times Magazine reports.
According to the Times Magazine, such data troves include electronic health records, which store the medical information of millions of patients. The data can be used to identify patterns that could inform care decisions.
For example, Jennifer Frankovich, a rheumatologist at Lucile Packard Children’s Hospital in California, analyzed the hospital’s database of lupus patients to pinpoint the likelihood of her patient developing a life-threatening blood clot. Frankovich said after pulling the data, physicians she had consulted with on the case changed their minds and agreed that “[i]t was very clear, based on the database, that [the patient] could be at an increased risk for a clot.” The patient was then given anti-clotting medications.
Data analysis also has been used to identify previously unknown adverse drug reactions. Nicholas Tatonetti, an assistant professor at Columbia University’s Tatonetti Laboratory, created an algorithm that allowed him to examine FDA’s database of adverse drug events and detect an adverse interaction between two commonly prescribed drugs.
While data mining has the potential to inform care decisions, observers have expressed concerns about patient privacy in the databases. According to the Times Magazine, hospitals might permit such analyses if they are deemed to be part of patient care. However, if the analyses are used for medical research, each patient whose data are accessed must give their consent.
In addition, experts have warned that databases can provide incomplete research results, since they include only individuals who sought treatment and not healthy individuals.
Further, Stanford University bioinformatics professor Daniel Rubin said that no studies have been conducted on the benefits of data-based medicine, which means care decisions based on the practice could be of “inferior quality” (Greenwood, New York Times Magazine, 10/3).