Development of a wearable sensor-based biofeedback system for upper-limb stroke rehabilitation using real-time motor intent detection and adaptive rhythmic cueing

Project Abstract

A wearable sensor-based biofeedback system for upper-limb stroke rehabilitation was developed to enhance motor recovery by leveraging real-time motor intent detection and adaptive rhythmic cueing. The system integrates inertial measurement units (IMUs), surface electromyography (sEMG) sensors, and a flexible haptic feedback module to create a closed-loop therapy platform that can be deployed in clinical and home settings. Real-time motor intent detection uses a multimodal algorithm that fuses kinematic data from IMUs with myoelectric signals from targeted upper-limb muscles to discern intended movements prior to actual limb displacement. This anticipatory detection enables the system to synchronize cueing with movement planning, thereby reducing the latency between intention and action, which is critical for neuroplastic changes after stroke. The adaptive rhythmic cueing component delivers multimodal cues—auditory metronome-like pacing and tactile haptic pulses—that are modulated based on performance metrics such as movement accuracy, smoothness, speed, and inter-segment coordination. Cueing difficulty adapts in real time to challenge patients appropriately, promoting gradual progression from guided to autonomous movement while avoiding excessive cognitive load or frustration. The study employed a mixed-methods design with a cohort of post-stroke participants across the subacute to chronic phases. Quantitative outcomes focused on motor impairment and functional use, evaluated using standardized tools including the Fugl-Meyer Assessment for the upper extremity (FMA-UE), Wolf Motor Function Test (WMFT), Box and Block Test (BBT), and kinematic indices of precision, jerk, and movement smoothness. Secondary outcomes included adherence, user satisfaction, and fatigue, measured through validated questionnaires and device usage logs. A randomized crossover protocol compared the biofeedback system against conventional therapy and a sham feedback condition to isolate the contribution of motor intent detection and adaptive cueing. Key findings indicate that real-time motor intent detection significantly reduced reaction time to movement initiation and improved task-specific kinematic consistency, particularly in shoulder abduction and elbow extension patterns. Adaptive rhythmic cueing enhanced temporal coupling between intention and action, resulting in higher movement accuracy and increased reproducibility of functional tasks such as reaching and object manipulation. Neurophysiological markers collected via noninvasive EEG in a subset of participants suggested heightened cortical engagement in motor planning networks during sessions with biofeedback, consistent with enhanced neuroplastic potential. Usability metrics demonstrated high acceptability, with participants reporting intuitive sensor wearability, seamless feedback timing, and perceived motivation to practice. The system’s modular architecture supports extensibility, enabling the integration of additional sensors, AI-based personalization, and remote monitoring features for tele-rehabilitation. Limitations include variability in motor impairment and learning effects across individuals, potential sensor drift, and the need for longer-term studies to assess sustained transfer to daily living activities. This work contributes a scalable, patient-centered rehabilitation paradigm that harnesses real-time intent decoding and adaptive cueing to amplify motor recovery after stroke.

Project Overview

What This Project Is About

---

A simple, hands-on study that explores how wearable sensors can help people recovering from stroke regain use of their arms. The project looks at combining sensors that measure movement and muscle signals with real-time feedback to guide and motivate therapy, plus a rhythm-based cueing system to improve repetition and coordination.

---

---

The Problem It Addresses

---

Many stroke survivors have limited arm function and access to consistent therapy. Traditional therapy can be repetitive and hard to tailor to each person. This project seeks a portable solution that provides immediate feedback and encourages correct movements, potentially speeding recovery and making at-home practice easier.

---

---

Objectives of the Project

---

1. Design a wearable setup that collects movement and muscle activity data from the arm.
2. Develop a real-time feedback system that indicates performance and guides improvements.
3. Implement adaptive rhythmic cues to support pacing and coordination during exercises.
4. Evaluate the system with simple, repeatable tasks to measure improvement trends.

---

---

What You Will Do Step by Step

---

1. Review basic literature on wearable sensors and biofeedback in rehab.

2. Select sensors (motion and possible muscle signals) and assemble a wearable prototype.

3. Create software that collects data, detects movement intent, and provides feedback.

4. Add an adaptive rhythm cueing feature to guide therapy sessions.

5. Run small tests with volunteers to tune usability and safety.

6. Analyze data to see if feedback helps accuracy and consistency.

7. Refine the system based on feedback and results.

8. Prepare a simple user guide and present findings.

---

---

Expected Outcome

A functional wearable-enabled rehabilitation aid that offers real-time feedback and rhythm-guided exercises, along with initial evidence of improved motor performance and engagement in therapy.