93. ESTIMATING MOTOR SCORES WITH ACCELEROMETERS IN THE NEURO ICU

Department: Electrical & Computer Engineering
Faculty Advisor(s): Vikash Gilja

Primary Student
Name: John Hermiz
Email: jhermiz@ucsd.edu
Phone: 513-497-7930
Grad Year: 2018

Student Collaborators
Alfredo Lucas, allucas@ucsd.edu | Venkatesh Elango, elangovenkatesh@gmail.com

Abstract
The neurological motor exam provides important information about a patient's clinical status. However, there are several limitations of the exam, which include assessment subjectivity and the frequency with which it can be performed. Frequent neurological exams during a patient's hospital stay disturbs sleep and can increase delirium, which worsens morbidity and mortality. We propose automating motor score assessment by passively measuring activity with accelerometers. In this work, we train a linear model against hourly motor scores and use past scores and accelerometer features as model inputs. In several cases, the model achieves significant correlation (P<2.68E-5, Bonferroni corrected) with actual scores, indicating that it may be feasible to estimate motor scores with accelerometers.

Industry Application Area(s)
Life Sciences/Medical Devices & Instruments

« Back to Posters or Search Results


Contact:   researchexpo@soe.ucsd.edu   (858) 534-6068