Empirical mode decomposition analysis of alcohol withdrawal tremor signals

In this paper, we have introduced a novel method to extract involuntary tremor movement activity in patients with Alcohol Withdrawal Syndrome (AWS). Using the Empirical Mode Decomposition (EMD), we show that the variations of energy of the tremor and voluntary activity can be distinguished in different Intrinsic Mode Functions (IMF) of the recorded tremor signals. To measure the performance of our method in extracting the tremor activity and eventually developing a shortened, more objective AWS assessment tool, we compared the electronic tremor assessment employing our new technique with the consensus rating from 3 expert physicians on a 7-point scale. Based on a 3-fold cross-validation on 104 recordings from 64 patients with AW, we found that our proposed method achieves an average Root Mean Squared Error (RMSE) of 0.71 with respect to the consensus rating, a significant improvement over prior work.