Automatic Assessment of Cognitive Impairment through Electronic Observation of Object Usage

Indications of cognitive impairments such as dementia and traumatic brain injury (TBI) are often subtle and may be frequently missed by primary care physicians. We describe an experiment where we unobtrusively collected sensor data as individuals with TBI performed a routine daily task (making coffee). We computed a series of four features of the sensor data that were increasingly representative of the task, and that we hypothesized might correlate with severity of cognitive impairment. Our main result is a significant correlation between the most representational feature and an apparent indicator of general neuropsychological integrity, namely, the first principal component of a standard suite of neuropsychological assessments. We also found suggestive but preliminary evidence of correlations between the computed features and a number of the individual tests in the assessment suite; this evidence can be used as the basis of larger-scale studies to validate significance.

• Type: Conference
• Year: 2010
• Where: PERVASIVE
• Authors: Mark R. Hodges, Ned Kirsch, Mark W. Newman, Martha E. Pollack
• Keywords: Applied Computing, Cognitive Impairment, PERVASIVE 2010, Routine Daily Task, Sensor Data