FFTrees

Nathaniel Phillips

Cook County

  • In 1996, the Cook County Hospital had a major problem

As the city’s principal public hospital, Cook County was the place of last resort for the hundreds of thousands of Chicagoans without health insurance. Resources were stretched to the limit. The hospital’s cavernous wards were built for another century. There were no private rooms, and patients were separated by flimsy plywood dividers. There was no cafeteria or private telephone—just a payphone for everyone at the end of the hall. In one possibly apocryphal story, doctors once trained a homeless man to do routine lab tests because there was no one else available.

But the Emergency Department (the ED) seemed to cry out for special attention. The rooms were jammed. A staggering 250,000 patients came through the ED every year. How do you figure out who needs what? How do you figure out how to direct resources to those who need them the most?”

But from the beginning, the question of how to deal with heart attacks was front and center. A significant number of those people filing into the ED—on average, about thirty a day—were worried that they were having a heart attack. Chest-pain patients were resource-intensive. The treatment protocol was long and elaborate and—worst of all—maddeningly inconclusive.

plot of chunk unnamed-chunk-2

Cook County

  • In 1996, the Cook County Hospital had a major problem

As the city’s principal public hospital, Cook County was the place of last resort for the hundreds of thousands of Chicagoans without health insurance. Resources were stretched to the limit. The hospital’s cavernous wards were built for another century. There were no private rooms, and patients were separated by flimsy plywood dividers. There was no cafeteria or private telephone—just a payphone for everyone at the end of the hall. In one possibly apocryphal story, doctors once trained a homeless man to do routine lab tests because there was no one else available.

But the Emergency Department (the ED) seemed to cry out for special attention. The rooms were jammed. A staggering 250,000 patients came through the ED every year. How do you figure out who needs what? How do you figure out how to direct resources to those who need them the most?”

plot of chunk unnamed-chunk-3

Heart disease

From the beginning, the question of how to deal with heart attacks was front and center. 30 people a day coming in worried about a heart attack.

  • A bed in the coronary care unit costs $2,000 a night and requires a 3 day stay.
  • Multiple measures

    • Blood pressure, Stethescope,
    • Questions: How long? How much? During exercise? History? Cholesterol? Drugs? etc.
    • Electrocardiogram (ECG) reading.
  • Disagreement between doctors

    • 20 patient case histories
    • Asked doctors to estimate from 0 to 100 the probability of a heart attack.
    • "In each case the answers we got pretty much ranged from 0 to 100. It was extraordinary."

Solution

  • A decision tree developed by a cardiologist named Lee Goldman
  • Tree combines ECG with just 3 risk factors.

plot of chunk unnamed-chunk-4

  • Nobody beleived the tree could outperform expert judgments. But over two years, the performance of the tree was compared to the physician judgments

  • Results: Tree had a 70% higher specificity and a 95% sensitivity compared to the 75-90% rate of doctors.

What is a Fast and Frugal Tree (FFT)?

A fast and frugal tree (FFT) is a decision tree with exactly two branches from each node, where at least one branch is an exit branch.

Why restrict a tree?

  • Speed
  • Frugality
  • Easy of understanding and implimentation

Depression Tree

  • Jenny et al. (2013): Simple rules for detecting depression

plot of chunk unnamed-chunk-5

Bank failure

  • Neth et al. (2013): Homo heuristicus in the financial world: From risk management to managing uncertainty

plot of chunk unnamed-chunk-6

Problem

Problem

There is no off-the-shelf method to construct fast and frugal decision trees

Solution

FFTrees