From Preference to Performance: Patient-Centered Design of Multimodal Cueing in Parkinson’s Disease Gait Training
This is a strong CHI best-paper contribution because it links formative patient-centered design to a concrete adaptive rehabilitation system and then tests meaningful modality and cueing trade-offs empirically. The standout insight is that better objective performance does not automatically translate into better user acceptance, which is exactly the kind of result that should reshape rehabilitation technology design priorities.
Axes Lens
Rare contribution shape, typical evidence profile. The point here is not a score. It is to show what kind of claim the paper makes, and whether the evidence pattern is unusual or baseline in this 268 -review set.
Contribution shape
- Knowledge form
- generative knowledge typical · 35/268
- Novelty type
- framework typical · 59/268
- Abstraction level
- system typical · 61/268
- Generalization target
- user population typical · 75/268
- Validation mode
- mixed methods typical · 136/268
Evidence profile
- Evidence strength
- strong typical · 158/268
- Claim alignment
- strong typical · 231/268
- Overclaim risk
- medium typical · 210/268
Review Summary
This paper stands out because it does three things that are often separated in healthcare HCI work and integrates them into one coherent contribution. First, it grounds the system in formative inquiry with clinicians and patients rather than treating rehabilitation as a purely technical optimization problem. That matters because the resulting CARES principles are not generic design slogans; they directly encode accessibility, personalization, guidance, evidence, and safety concerns that are highly relevant for Parkinson’s gait training. Second, the authors actually instantiate those principles in a working adaptive multimodal system using IMUs, AR/audio presentation, and wearable vibration, so the contribution is not merely conceptual. Third, they evaluate the system in a way that produces actionable comparative knowledge rather than a simple “prototype works” claim. The two studies separate cueing type from cueing modality and show that the best cue depends on the movement-optimization strategy: process cueing is more effective for visual and auditory channels, while periodic cueing is better for somatosensory stimulation. Just as importantly, the paper surfaces a performance–preference dissociation in which auditory cueing often yields the strongest objective gains but visual cueing is preferred and feels less effortful. That is a valuable corrective to a common assumption in rehabilitation design that maximizing measurable performance is sufficient. The evidence is persuasive for the intended scope, but the authors appropriately acknowledge real limits: the participants were relatively high-functioning early-stage PD patients, and the study period was short and controlled, so accessibility, novelty, and long-term adherence remain open questions. Even with those constraints, the paper makes a field-shaping contribution by showing how patient-centered design can produce both a reusable framework and sharper empirical guidance for multimodal rehabilitation systems.
What Changed
Canon before
Prior rehabilitation training systems for Parkinson’s disease heavily rely on clinician guidance and often focus on standardized gait correction rather than patient-centered adaptive approaches; moreover, prior work has focused largely on velocity improvements with limited attention to gait stability and individual differences in cueing modality effectiveness and patient preferences.
Departure from common sense
Contrary to the common assumption that the most objectively effective rehabilitation cueing would also be the most preferred, this paper reports a performance–preference dissociation: auditory cueing often produced the largest objective improvement, but patients preferred visual cueing and reported lower workload. It also pushes against a purely velocity-centric framing by comparing cueing in terms of both gait performance and patient experience.
Actual novelty
The paper contributes a patient-centered CARES design framework derived from interviews, a co-designed adaptive multimodal gait-training system using IMUs plus visual, auditory, and somatosensory cueing, and comparative studies showing modality–paradigm specificity: process cueing works better for visual and auditory channels, whereas periodic cueing works better for somatosensory stimulation.
Evidence
Evidence comes from formative interview analysis used to derive the CARES principles, a concrete adaptive prototype that operationalizes those principles, and two controlled user studies with 16 early-stage PD patients. The studies compare cueing types and modalities using objective gait measures plus subjective workload and preference, supporting both the design contribution and the empirical claims. The paper also explicitly bounds its claims by noting participant inclusion limits and the short controlled study period.
“4 Implementation of the Prototype System Through the pilot study, we co-designed the rehabilitation system. This section introduces the cueing paradigm and the prototype system. First, we reviewed the neurophysiological mechanisms by which external cueing modulates gait performance, identified two movement-optimization strategies, and accordingly devised two types of cueing paradigms”
actual novelty · 4 Implementation of the Prototype System · confidence 0.97
“ A performance–preference dissociation emerged: auditory cueing often produced the largest objective improvement, but patients preferred visual cueing and reported lower workload.”
departure from common sense · 6 User Study 2: Exploring Differences Across Cueing Presentation Modalities · confidence 0.98
“ty and broader inclusion. Our current design and evaluation primarily targeted participants with moderate motor challenges and relatively preserved visual and cognitive function. Finally, the experimental period was relatively short and took place in a controlled environment, so novelty effects and device form factors may have influenced participants’ preferences and workload ratings”
limitation · 7.7 Limitations and Future Work · confidence 0.99
“l cueing, examining strategies for information delivery and gait correction. Results indicated that visual and auditory cueing were more effective for process-oriented adjustments, whereas somatosensory stimulation better supported periodic cueing. Moreover, a dissociation between performance outcomes and user preferences was observed. Th”
validation scope · Abstract · confidence 0.95
Limits
Method limits
The evaluation primarily targeted early-stage participants with moderate motor challenges and relatively preserved visual and cognitive function, and the experimental period was short and conducted in a controlled environment.
Deployment limits
The current evidence does not establish long-term home deployment performance or adherence. Device form factors and novelty effects may have influenced preference and workload, and broader accessibility for people with advanced PD, visual impairments, or cognitive changes remains to be tested.
Boundary conditions
The findings are best interpreted for early-stage PD rehabilitation contexts using wearable multimodal cueing, especially for users able to engage with AR, audio, and vibration interfaces. Effects may differ for more advanced PD, different impairment profiles, or longer-term unsupervised home use.
Position in field
This paper advances HCI healthcare research by connecting patient-centered design, adaptive wearable rehabilitation, and comparative multimodal evaluation. Its main field contribution is not only a prototype but a structured framework and empirical argument that rehabilitation systems should optimize for both gait outcomes and lived user acceptance rather than assuming those align automatically.