Automatic identification of solid-phase medication intake using wireless wearable accelerometers.

We have proposed a novel solution to a fundamental problem encountered in implementing non-ingestion based medical adherence monitoring systems, namely, how to reliably identify pill medication intake. We show how wireless wearable devices with tri-axial accelerometer can be used to detect and classify hand gestures of users during solid-phase medication intake. Two devices were worn on the wrists of each user. Users were asked to perform two activities in the way that is natural and most comfortable to them: (1) taking empty gelatin capsules with water, and (2) drinking water and wiping mouth. 25 users participated in this study. The signals obtained from the devices were filtered and the patterns were identified using dynamic time warping algorithm. Using hand gesture signals, we achieved 84.17 percent true positive rate and 13.33 percent false alarm rate, thus demonstrating that the hand gestures could be used to effectively identify pill taking activity.

On the feature selection criterion based on an approximation of multidimensional mutual information.

We derive the feature selection criterion presented in [CHECK END OF SENTENCE] and [CHECK END OF SENTENCE] from the multidimensional mutual information between features and the class. Our derivation: 1) specifies and validates the lower-order dependency assumptions of the criterion and 2) mathematically justifies the utility of the criterion by relating it to Bayes classification error.

On the relationship between dependence tree classification error and Bayes error rate.

Wong and Poon [1] showed that Chow and Liu's tree dependence approximation can be derived by minimizing an upper bound of the Bayes error rate. Wong and Poon's result was obtained by expanding the conditional entropy H(w|X). We derive the correct expansion of H(w|X) and present its implication.