Using Artificial Intelligence to Adapt Level of Difficulty in a Cognitive Training Program

Research output: Contribution to conferencePosterResearch

  • Inge Linda Wilms
Increasingly, computer-based training is being used within the field of rehabilitation in the effort to improve cognitive functions such as memory, attention and language2,4,5. As a complement to clinical training, computer-based training is a fairly inexpensive way of increasing the intensity and frequency of training received by patients. However, the performance of brain-injured patients may fluctuate during cognitive training due to factors such as fatigue, cognitive load and training-induced progress. With paper-and-pencil training, the therapist ordinarily monitors progress and adjusts the workload and level of difficulty presented to the patient to achieve the best results. Computer-based training may be administered at home, in groups or over the internet, without constant and direct intervention from a therapist. For this reason, it makes sense to build logic into the computer-based training systems, that secures a reasonable and individualized progression in the level of difficulty presented to the patient during training. This study investigates the possibility of using the actor/critic method from the discipline of reinforcement learning, to control the level of difficulty during training. The intention is for task difficulty to be increased or decreased during training depending on the performance of the patient, thus challenging the patient at an optimal level at all times.
Original languageEnglish
Publication date2009
Number of pages1
Publication statusPublished - 2009
EventInternational Symposium on Neurrehabilitation from basics to future - Valencia, Spain
Duration: 15 Oct 200916 Oct 2009

Conference

ConferenceInternational Symposium on Neurrehabilitation from basics to future
CountrySpain
CityValencia
Period15/10/200916/10/2009

ID: 19524367