DISQUS

Altgate: Cracking the Code on Patient Medical Record Search

  • TedHoward · 5 months ago
    There are so many subtleties of medical records that most people don't know about. Here's just one simple example.
    In many ethnic populations, people have many names. If someone's name is "Maria Jose Martinez Gonalez Gutierrez" (not to pick on Hispanics but I live in San Francisco) then what name are they entered as? Many systems only allow for 3 names, and practitioners will often only enter 2 names. Is the name of the lady's entry the same in each hospital? Are there other people with that name? If one matching entry shows a fully healthy person and one shows someone with seizure disorders and possible kidney problems, what do the doctors do? (In the USA they have practice defensive medicine and assume the worst of course.) What happens to names when people get married?
    The issues involved in medical records are so complex as to require a series of textbooks just to explain. Delineating and fixing these issues will require much more work, and both efforts need to start today so they can someday be solved.
  • fnazeeri · 5 months ago
    I completely agree, Ted. Medical records are probably the most complicated data set on earth when you consider the amount and nature of the data. What's scariest about this, though, is that the US $2.4 trillion spend on health care is driven by these data. Couldn't agree more that we need to get cracking on the effort to upgrade and improve patient medical record systems.
  • obxerve · 5 months ago
    I agree that this is interesting, up to a point. What I do not yet see from the presentation nor i2b2.org web site is the data collaboration aspect. That is the killer! At the end of the day, there are two main things (tired, so only can think of two this minute): how much (& how diverse) clean data is available, and how can that data be mined internally within companies by their medical researchers. The software is the least of the problem.

    For reference, see ehrcr.org and cdisc.org.
  • fnazeeri · 5 months ago
    I do know that i2b2 is designed for federated queries and in fact there are a number of sites that are already doing this. The collaborative aspect of that is quite significant. The trick, of course, is that the data is non-anonymized so access is limited appropriately. And you're right, the software, while not trivial, is only a piece of the solution.
  • Hashim Ibrahim · 3 months ago
    It is good medical coverage, regard,