# Variants of active assist robotic therapy: Feasibility of Virtual Assistance and Proprioceptive Training as gauged by their effects on success and motivation during finger movement training after stroke

## TL;DR

This study tests new robotic therapy methods for stroke survivors, finding that physical assistance boosts motivation more than virtual assistance, while proprioceptive training remains engaging despite challenges.

## Contribution

The paper introduces and evaluates two novel robotic therapy modes—Virtual Assistance and Proprioceptive Training—for post-stroke rehabilitation.

## Key findings

- Participants achieved ~80% success across all three therapy modes.
- Virtual Assistance led to lower motivation due to reduced perceived competence.
- Proprioceptive Training was engaging even for those with impaired proprioception, though requiring more assistance.

## Abstract

Standard robotic therapy for upper extremity stroke rehabilitation provides physical assistance during video game-based training. While effective, it requires complex equipment, focuses on visuo-motor control, and yields variable results. This raises the question whether more pragmatic or targeted variants can be developed. To explore this, we pilot tested two novel robotic therapy modes. In the Virtual Assistance mode, game parameters are modulated to promote task completion without physical assistance – potentially useful when only sensors are available, although its acceptability for people with more severe motor impairment is unclear. In the Proprioceptive Training mode, participants play video games through a combination of visual and haptic cues. This approach aims to retrain proprioception, but may also be too challenging for individuals with proprioceptive impairment to find it motivating. This study tested the feasibility of both variants across a range of motor and proprioceptive impairments, comparing them to training with Standard robotic therapy.

Chronic stroke participants (N = 46) were randomized to receive Standard, Virtual, or Proprioceptive Training. Participants used the FINGER robot to train in three sessions across one week, during which an adaptive algorithm attempted to titrate success to ~ 80%. Baseline proprioceptive and motor function were assessed prior to training, and motivation for training was assessed using the Intrinsic Motivation Inventory. Feasibility was evaluated by levels of gameplay success and motivation for training.

Participants of widely varying motor and proprioceptive ability achieved ~ 80% success for all three modes. However, Virtual Assistance resulted in significantly diminished motivation, due to lower perceived competence when participants were not provided with physical assistance. Participants with impaired proprioception rated Proprioceptive Training engaging, although it was more challenging for them and they required an increased level of assistance.

Both training paradigms were feasible for use with chronic stroke survivors and were able to achieve high gameplay success and motivation for training. However, physical assistance appeared to have an advantage over Virtual Assistance in raising training motivation. Proprioceptive Training required high levels of assistance, but was motivating even for people with poor proprioception.