Falls in daily life are a serious risk for people with stroke. A new type of balance training, called reactive balance training (RBT). may help to reduce this risk of falling. In some previous studies participants improved their balance reactions a lot after RBT, whereas others did not improve at all. These studies used different types and amounts of training. Differences in training program features might explain differences in the study results.

Training intensity is the difficulty or challenge of the training program. For other types of exercise (like 'cardio' or strength training) if the intensity of exercise is increased, someone can get the same benefits in less time than with lower intensity exercise. Physiotherapists report that they have limited time in rehabilitation to do everything they need to do with their stroke patients, so it would be valuable to know if high-intensity RBT improves balance reactions quickly. The goal of this study is to see if more intense RBT improves balance reactions faster than less intense RBT.

People with chronic stroke will be randomly placed in one of three groups: high-intensity RBT, moderate-intensity RBT, or a walking control group. The investigators will find the fastest moving platform speed that participants can respond to with single step (multi-step threshold). There will then be 4 1-hour long training sessions in one week. Participants in the high-intensity group will experience platform movements that are 50% faster than the multi-step threshold. Participants in the moderate-intensity group will experience platform movements at the multi-step threshold. Participants in both RBT groups will experience 36 multi-directional platform movements in each training session, causing them to start to fall forwards, to the left, or to the right. Participants in the walking group will walk on the platform 36 times without any platform movements. The investigators will measure how quickly people improve their balance reactions over the training program.

Remaining active after stroke is essential to recovery, maintaining quality of life, and reducing secondary stroke risk. However, impaired balance control post-stroke increases fall risk, contributes to fear of falling, and reduces overall mobility, activity, and community integration. Therefore, effective interventions to improve balance control and reduce fall risk are essential to ongoing recovery and quality of life post-stroke. However, while there is unequivocal evidence that exercise, specifically balance training, is the most effective intervention for preventing falls in older adults without stroke, it is unclear if exercise prevents falls after stroke. Effective balance reactions (e.g., reactive stepping) are essential to avoid falling following a loss of balance, or balance perturbation. Reactive balance training (RBT), where clients experience repeated balance perturbations, is a novel type of exercise that aims to improve control of balance reactions. The investigators found that RBT improves reactive stepping ability in people with stroke. The investigators' recent meta-analysis found that RBT reduces rate of falls in daily life by ~40% among older adults and people with neurologic conditions, including stroke.

Clinicians are unsure how to optimally prescribe RBT. Exercise is often prescribed considering frequency, intensity, time, and type; these parameters are interdependent. For example, as intensity increases, frequency or time can decrease to achieve similar benefits. Clinicians report that limited time in rehabilitation services and competing rehabilitation goals and priorities mean that there is little time available to include RBT in a client's treatment plan. Therefore, it would be valuable to know if a short duration of high-intensity RBT can improve reactive balance control post-stroke. However, it is possible that people with stroke would not tolerate high intensity perturbations and, therefore, require lower-intensity but long-duration RBT. No study has directly compared different RBT training intensities among people with stroke.

The purpose of this study is to determine the optimal intensity of RBT post-stroke. The 'optimal' intensity is the intensity that improves reactive balance control in fewer sessions, without any apparent negative consequences (i.e., no increase in adverse outcomes). Responses to backward-fall perturbations will be assessed at the end of each session, and at the one-week retention time-point. An untrained perturbation (i.e., backward-fall perturbation) will be used to test transfer of learning to a novel context. Participants will experience 3 forward-directed platform translations, evoking a backward loss of balance, at the multi-step threshold. Because backward falls are more challenging than forward or lateral falls, a backward-fall perturbation at the forward-fall multi-step threshold should evoke a multi-step reaction, at least prior to training. Testing at each participant's multi-step threshold ensures that participants will initially experience challenges responding to the perturbation and there is room for improvement with training. The primary outcome will be number of steps taken to recover balance.

The investigators will calculate the learning rate for each participant by fitting an exponential function, a+b(e^(- t/x) ), to their data (average number of steps taken at each assessment time point), where a and b are constants, t is the assessment time point, and x is the learning rate.

The investigators expect that high-intensity training (50% above the multi-step threshold) will improve reactive stepping ability faster than moderate-intensity training (at the multi-step threshold). If the first reactive step is not effective in avoiding a fall after a loss of balance, additional steps must be taken; therefore, the number of steps needed to recover from a loss of balance is a global indicator of the effectiveness of balance reactions. The investigators will assess reactions to novel, untrained, balance perturbations throughout training, and calculate the rate of decline in average number of steps taken to recover balance (i.e., learning rate). Retention of learning will be assessed one week post-training. The investigators' primary hypotheses are:

1. Adaptation rate will be faster for high-intensity RBT than moderate-intensity RBT;
2. Adaptation rate will be faster for both RBT groups than a walking control group; and
3. Both RBT groups will have better retention of learning than the walking control group; i.e., the RBT groups will take fewer steps to respond to the novel perturbation than the walking control group one week post-training.

Secondary objectives are to compare the rate of adverse outcomes between groups, and to determine the effect of different intensities of RBT on the mechanisms underlying improved reactive stepping ability, functional balance, falls efficacy, and participation in daily activities.

Inclusion Criteria:

• Chronic stroke (>6 months post-stroke)

Exclusion Criteria:

• Are unable to stand independently without upper-limb support for >30 seconds and/or walk independently (without a gait aid) for ≥10 metres;
• Have another neurological condition that could affect balance control (e.g., Parkinson's disease);
• Have cognitive impairment (Montreal Cognitive Assessment8 score <26), or severe language or communication difficulties affecting understanding instructions;
• Have contraindications to RBT (has described by Mansfield et al., 2021), such as osteoporosis, activity restrictions due to cardiac event/surgery, or severe spasticity in the lower extremity; and/or
• Are currently attending in- or out-patient physiotherapy or supervised exercise.