Title
Towards true personalized medicine: Statistical platform for individualized behavioral analysis, dynamic diagnosis and real-time tracking of intervention outcomes
Elizabeth B Torres, Rutgers University – Psychology, Computer Science and Cognitive Science
Abstract
Physical motions can be continuously captured with wearable sensing technology (e.g. accelerometers in smart phones or lifestyle bracelets outputting caloric intake, sleep patterns, etc.) Such wearable sensors have opened a broad market which will have implications for personalized medicine. However, there is a current gap between the output data that these sensors provide and the analytical tools available to report their outcome. The reports are currently based on discrete averages of data chunks taken a posteriori (at the end of the day or the week, month, etc.) to build a trend. The statistics from such trends are computed relative to normative population averages that do not truly reflect the uniqueness of the person’s nervous systems. Lastly no analytical tools provide true real-time capability to continuously track changes in the person’s physiologically relevant states so as to detect or even shift their specific signatures.
We have discovered that human motions have non-stationary statistics. These shifts in the stochastic signatures of the motor output variability manifest at the various functional levels of the nervous system (from voluntary, to automatic to autonomic). The rates at which they occur provide a sort of “fingerprint” of the individual’s adaptive capacity and can be tracked in real time to automatically reveal the different levels of intent in the person’s behaviors. This real-time signal can also be paired with other forms of sensory input to augment sensory feedback capabilities or to sensory-substitute missing, corrupted or decreased inputs. This feature of our technology enables the controlled re-parameterization of the motor output as (re-afferent) sensory feedback input to the nervous system so as to lower somato-sensory-motor uncertainty (noise) and to reliably enhance the predictive and anticipatory capacity of the person.
We here present examples of the use of this technology in the clinical domain including diagnosis, sub-typing of severity and interventions in Autism Spectrum disorders, Parkinson’s disease, and de-afferentation. We also provide examples of the use of this platform in sports, the performing arts (modern dance) and in an ongoing clinical trial.