Abstract TMP44: Personalizing Rehabilitation for Stroke Survivors- A Big Data Approach
Introduction: Advances in connected health delivery provides a unique opportunity to maximize intervention effectiveness for stroke patients. It has also helped collect large sets of data to facilitate clinical decision making. This vastly insightful data is used to personalize neurorehabilitation where the evidence for gains in chronic stroke patients is weak.
Methods: Over a span of 2 years (2013-2015), data was anonymously aggregated and analyzed from over 2,500 patients with post-stroke aphasia. Data was collected using a mobile therapy platform, Constant Therapy, which has 60 evidenced-based language and cognitive therapy tasks. The program was used by patients in the home and clinic under the guidance of a clinician, or under their own volition if not regularly seeing a clinician. This program dynamically adapted to each patient’s progress. Patients who completed between 3 and 1000 treatment sessions were analyzed to determine which tasks showed statistically significant changes. These data were compared with control patients who completed tests at two separate time points but with no intervening treatment.
Results: Despite the older demographic of patients (median age = 64 yrs), they performed an average of over 20 minutes on home therapy every day. The analyses take into account the number of patients who completed a specific task and show a significant change (all p <.05) for accuracy or latency. For example, the 2-step Auditory Command task (which requires a user to follow 2-step directions) showed a 12 point gain in accuracy and 39% improvement in processing speed in 1,200 patients. The results also show that patients with higher initial severity scores showed significant gains in accuracy, and reach similar post-treatment accuracy to those with lower severity scores, provided that they received an appropriate dosage of therapy. In contrast, control patients showed minimal gains on tasks that were assigned.
Conclusion: These results show that all patients, including the most severe, can make progress in their rehabilitation when treatment is individualized for them. Analysis of large data sets can be used to inform rehabilitation by highlighting therapies that are effective while accounting for etiology and individual variability.
Author Disclosures: S. Kiran: Employment; Significant; Boston University. Research Grant; Significant; NIH/NIDCD. Ownership Interest; Significant; Constant Therapy. J. Godlove: Employment; Significant; Constant Therapy. M. Advani: Employment; Significant; Constant Therapy. V. Anantha: Employment; Significant; Constant Therapy.
- © 2017 by American Heart Association, Inc.