Evaluation of an ontology-based system for computerized cognitive rehabilitation

This paper describes the results of a randomized clinical trial about the effectiveness of a computerized rehabilitation treatment on a sample of 31 patients affected by Parkinson disease.

Computerized exercises were administered by the therapists to the intervention group (n = 17) through the CoRe tool, which automatically generates a big variety of exercises leveraging on a stimuli set (words, sounds and images) organized into a dedicated ontology. A battery of standard neuropsychological tests was performed for patients' assessment at baseline, after the treatment (that lasted 1 month), and after 6 months from the treatment stop. The control group underwent a sham intervention.

Results show a statistically significant clinical benefit from computerized rehabilitation with respect to sham treatment. For the intervention group, response time and response accuracy were integrated into a weighted score that accounts also for the specific cognitive burden of each exercise. Differently from the control group, the majority of patients in the intervention group showed an improvement in that score, more marked in the first week of treatment, and which lasts for the entire treatment period, which could account both for a quick learning effect and for an improvement of cognitive conditions. Good usability of CoRe, already observed in previous studies, was confirmed by the present trial, where the percentage of protocol completion in the intervention group is very high (all but one patient are above 90%).

The CoRe system showed to be effective to improve some cognitive abilities in patients with Parkinson disease. However, after the end of the training, the benefit is hardly maintained over time. These findings support the implementation of CoRe in the clinical routine and the continuation of the treatment after discharge through the use of a homecare version of the system.