Patient Specific Framework for Biomedical Signal Management 16.02


Project Start Date: Jul 1, 2016
Funding: Member Funded
Project Tags: ,

Project Summary

 Personalised  healthcare  services  are  striving  for  improving  the  quality  of  individuals’  life  by  proposing  the  customization  of  medical  decisions   and  products.  Interpreting  the  biomedical  signals,  such  as  Electroencephalogram  (EEG)  and  Electrocardiogram  (ECG),  falls  into  these  types  of  services,  and   covers  numerous  applications  from  real-­‐time  monitoring  to  diagnosis  of  different  health  conditions.  The  conventional  time  series  analyses  lack  the  required   robustness  to  address  the  need  for  interpretation  of  biomedical  signals. In  order  to  address  this  need, our  ambition  is  focused  in  increasing  the  classification   accuracy  along  with  the  minimum  feedback  from  experts  within  a  patient-­‐specific  framework.  The  main  goal  is  to  study  and  develop  the  state-­‐of-­‐the-­‐art   nonlinear  methods  in  order  to  interpret  the  complex  behaviour  of  the  brain  and  heart  using  bioelectric  signals  and  provide  complementary  information  to   characterize  specific  underlying  patterns  such  as  providing  early  warning  for  brain  seizures  or  heart  arrhythmias. More  specifically,  we  aim  to  explore  and   develop  robust  and  novel  method  in  a  sense  of  reconstruction  and  characterization  of  the  signals’  dynamics.  In  order  to  achieve  crucial  improvements  over   the  state-­‐of-­‐the-­‐art  techniques,  the  focus  will  be  particularly  drawn  onto  the  Deep  vs.  “Divide  and  Conquer”  Learning  using  Operational  Neural  Networks   (ONNs).

PERSONNEL

Principal Investigator(s)

Student(s)