Wednesday, September 18, 2013
Risk of depression: the past predicts the future
In recent posts, I have noted recent attempts to develop algorithms capable of predicting recurrences in major depression. My colleague, Dr. Jian Li Wang, has been working on developing such algorithms and his work extends similar work done previously by the PredictD study group. However, these algorithms are complex and can be difficult to apply. Dr. Andrew Bulloch, at the University of Calgary has recently shown (link to article synopsis here) that a very simple rule can work quite well. Using Canadian epidemiological data, he has shown that the number of prior episodes is a good predictor and that the risk of future episodes has a linear relationship to the number of past episodes. While this is not surprising, the strength of association is very strong and this may be a strong and simple predictive rule.