eduzhai > Applied Sciences > Engineering >

Adaptive multi-channel event segmentation and feature extraction for monitoring health outcomes

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 37 pages

Abstract: $ textbf{Objective}$: To develop a multi-channel device event segmentationand feature extraction algorithm that is robust to changes in datadistribution. $ textbf{Methods}$: We introduce an adaptive transfer learningalgorithm to classify and segment events from non-stationary multi-channeltemporal data. Using a multivariate hidden Markov model (HMM) and Fisher slinear discriminant analysis (FLDA) the algorithm adaptively adjusts to shiftsin distribution over time. The proposed algorithm is unsupervised and learns tolabel events without requiring $ textit{a priori}$ information about true eventstates. The procedure is illustrated on experimental data collected from acohort in a human viral challenge (HVC) study, where certain subjects havedisrupted wake and sleep patterns after exposure to a H1N1 influenza pathogen.$ textbf{Results}$: Simulations establish that the proposed adaptive algorithmsignificantly outperforms other event classification methods. When applied toearly time points in the HVC data the algorithm extracts sleep wake featuresthat are predictive of both infection and infection onset time.$ textbf{Conclusion}$: The proposed transfer learning event segmentation methodis robust to temporal shifts in data distribution and can be used to producehighly discriminative event-labeled features for health monitoring.$ textbf{Significance}$: Our integrated multisensor signal processing andtransfer learning method is applicable to many ambulatory monitoringapplications.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×