Development of an adaptive virtual assistant to support virtual rehabilitation of the upper limb after stroke: A co-design approach

Research Objectives: To co-design an adaptive virtual assistant (in the form of an Avatar) to support virtual rehabilitation of the upper limbs to improve functional recovery after stroke

Design: Prototype development
A multi-step, iterative co-design process was used to develop the movement of an adaptive virtual assistant represented by an on-screen avatar. The research team initially developed the movement training components of the avatar-based virtual rehabilitation of the upper limb and sought the perspectives of a patient partner after demonstration of the avatar prototypes. Using a co-design approach, the avatar prototypes were amended with input from the patient partner and the research team.

Setting: Academic Research Institute

Participants: One person with stroke and four researchers (one clinician, two engineers, and one health technologist)

Interventions: avatar-based upper limb rehabilitation after stroke

Main Outcome Measures: N/A

Results: Six avatar prototypes were co-designed to demonstrate the following two-dimensional planar movements:(1)shoulder abduction-adduction; (2)shoulder flexion-extension; (3)shoulder external rotation; (4)elbow flexion-extension and; (5)wrist flexion-extension. Our patient-partner found the avatar helpful, engaging and fun to interact with. The following parameters of the avatar design were emphasized by the patient-partner and the research team: (1)preference for avatars with visible joint movements and muscle actions during targeted movements; (2)provision of real-time visual or auditory feedback to assist in self-correction of movements; (3)provision of summative feedback with results of performance such as providing scores to enhance motivation; (4)gamification of targeted upper limb movements; (5)receiving daily and weekly progress reports for self-monitoring; (6)personalization of avatars with different genders and ethnicities; (7)use of web application compatible with portable devices such as phones to increase accessibility and; (8)familiarization with the avatar-based exercises prior to patient discharge to facilitate self-management at home.

Conclusions: This co-design approach in the development of the adaptive virtual assistant will shape the transition from ideas and concepts to the next version of avatar prototype. Self-directed rehabilitation programs using avatars can provide an alternative means for community-living people with stroke to receive rehabilitation at home and continue their recovery.

Author(s) Disclosures: We have no disclosures.