A learning algorithm designated by the right facing

  • No School
  • AA 1
  • 13

This preview shows page 6 - 8 out of 13 pages.

current medications aggregated into a count in which case we get a summary table of the type below. A learning algorithm, designated by the right facing blue arrow in Figure 3, could then be applied to discover the pattern shown on the right of the table. The pattern represents an abstraction of the data. Essentially, this is the type of question we should ask the database, if only we knew what to ask! Figure 3: Extracting Interesting Patterns of Health Outcomes From Healthcare System Use Why is the pattern on the right side interesting? To appreciate this, suppose the overall complication rate in the population is 5%. In other words, a random sample of the database would on average, contain 5% complications. Under this scenario, the snippet on the right hand side could be very interesting since its complication rate is many times greater than the average. The critical question here is whether this is a pattern that is robust and hence predictive , that is, likely to hold up on unseen cases in the future. If so, it is actionable. The issue of determining robustness has been addressed extensively in the machine learning literature. 14 If the above table is representative of the larger database, the box on the right tells us the interesting question to ask our database , namely, “What is the incidence of complications in Type 2 diabetes for people over 36 who are on more than five medications?” In terms of actionability, such a pattern would suggest being extra vigilant about people with this profile who do not currently have a complication because of their high susceptibility to complications. The general point is that when the data are large and multidimensional, it is practically impossible for us to know a priori that a query such as the one above is a good one, that is, one that provides a potentially interesting insight. Suitably designed machine learning help find such patterns for us. Equally importantly, these pat terns must be predictive. Typically, the emphasis on predictability favors Occam’s razor since simpler models generally have a higher chance of holding up on future observations than more complex models, all else being equal. 3 For example, consider the diabetes complication pattern above: Age > 50 and #Medication > 6 Complication_rate=100%
Image of page 6

Subscribe to view the full document.

A simpler competing model might ignore age altogether, stating that people over 50 develop complications. The goodness of such a model would become more apparent when applied to future data. Does simplicity lead to higher future predictive accuracy in terms of lower false positives and false negatives? If so, it is favored. The practice of “out of sample” and “out of time” testing is used to assess the robustness of patterns from a predictive standpoint. When predictive accuracy becomes a primary objective, the computer tends to play a significant role in model building and decision making. It builds predictive models through an intelligent “generate and test” process, th e end result of which is an assembled model that is the decision maker. In other words,
Image of page 7
Image of page 8

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern

Ask Expert Tutors You can ask 0 bonus questions You can ask 0 questions (0 expire soon) You can ask 0 questions (will expire )
Answers in as fast as 15 minutes