Pattern Analysis and Clinical Reasoning by Prof. Gordon Parker

Posted on April 13, 2016

Prof. Gordon Parker, Professor of Psychiatry, author and business executive, brings us a personal and fascinating discussion about pattern analysis. This section of his talk opens with personal examples of how pattern analysis has altered his way of thinking between explicit and implicit modes of reasoning.

If medicine was a science, a well programmed computer could substitute for the best physician

Montgomery

Key Points:

1. Pattern Analysis –

It is applying a label to multiple differing given input values to generate a “most likely” or “probabilistic” match of the individual inputs handled in a Bayesian way

2. Bayesian analysis is a useful tool to consider outcomes when the outcome is uncertain. When several outcomes are possible, one can assign a probabilistic value to each outcome on the basis of evidence available. These probabilities change when additional information becomes available. For example: The probability of rolling a 3 from a single roll of a die is 1/6 = 16% probability. If one were to then ask for only even numbers then this probability changes to 3/6 =50%. Thus as additional information is integrated, the probability changes.

3. Experienced chess players after thousand’s of hours of practice begin to see the chess board different to others. They can spot probabilities and can hence work several moves ahead. They essentially recognise patterns. This ‘intuition’ is nothing more than pattern recognition.

4. Pattern analysis is a fast process and can be subject to errors. It is part of the System 1 thinking described by Kahnemann.

5. System 1 thinking is used frequently in diagnosis and clinical reasoning. A set of criteria may look like depression (when in fact it may not be) because system 1 represents categories by prototypes or typical exemplars.

  • generates impressions and feelings and is gullible
  • Operates automatically, quickly and effortlessly
  • Excels at constructing the best possible story incorporating currently active ideas.
  • neglects ambiguity and suppresses doubt.

6. System 2 thinking: May look for evidence that shows that the diagnosis is not depression when a certain set of criteria are fulfilled.

  • slow thinking process which can be used to check if System 1 thinking will work
  • required to check that intuition is not idiosyncratic.

7. An iterative process works best.

Want to learn more? Click here to see Prof Parker’s talk on the neurobiology of brain plasticity and pattern recognition.

References:

Kahneman, D. (2011). Thinking, fast and slow. Macmillan.