Development of a Wearable Haptic Feedback System for Post-Stroke Upper Limb Rehabilitation Using Real-Time Biofeedback and AI-Driven Progress Monitoring

Project Abstract

Stroke remains a leading cause of long-term upper-limb disability, with motor impairments limiting independence and quality of life. This study presents the development of a wearable haptic feedback system designed to enhance post-stroke rehabilitation by delivering real-time somatosensory cues synchronized with task-specific motor training and AI-driven progress monitoring. The system integrates a lightweight, sensorized glove with embedded actuators that provide tactile feedback across the fingers, palm, and wrist to reinforce correct movement trajectories, force modulation, and coordinated hand–eye tasks. In tandem, surface electromyography, inertial measurement units, and grip-force sensors capture multimodal data during therapy sessions, enabling fine-grained analysis of motor recovery dynamics. A patient-tailored rehabilitation protocol was implemented, combining repetitive task practice, constraint-induced movement therapy-inspired paradigms, and graded assistance-as-needed control. Real-time biofeedback is delivered through customizable haptic modalities, including vibration amplitude, pulse timing, and localized pressure, which are modulated by the AI engine based on ongoing performance metrics. The AI component employs supervised and reinforcement learning methods to model individual learning curves, predict plateaus, and adapt task difficulty, feedback intensity, and assistance levels to optimize engagement, endurance, and neuroplastic adaptation. Data visualization dashboards provide clinicians with actionable insights on trajectory deviations, compensatory strategies, and progression across impairment domains. The methodological framework comprises iterative prototyping, usability testing, and a longitudinal pilot study with a mixed-age cohort of post-stroke participants (n=40) undergoing eight weeks of intervention. Primary outcomes include standardized motor assessments (Fugl-Meyer Upper Extremity, Box and Blocks Test), functional performance measures (Action Research Arm Test, Wolf Motor Function Test), and functional independence scales. Secondary outcomes address user acceptance, perceived workload, and psychosocial factors via validated questionnaires. To ensure ecological validity, therapy tasks simulate daily activities such as reach-to-grasp, object manipulation, and bimanual coordination within a home-like environment, enabling smooth transition to community and home-based rehabilitation. Preliminary results indicate that the integration of haptic feedback with AI-adaptive progression accelerates motor recovery by promoting more accurate motor planning, reducing compensatory patterns, and enhancing motor learning efficiency. The system demonstrated high adherence and favorable usability scores, with clinicians reporting improved ability to tailor interventions and monitor progress remotely. Statistical analyses reveal significant improvements over baseline in primary motor scores and functional tasks, with effect sizes suggesting clinically meaningful gains. Subgroup analyses suggest greater benefits for individuals with moderate impairments and those engaging in higher-frequency sessions. The study addresses key translational challenges, including device comfort, battery life, data privacy, and integration with existing clinical workflows. This research contributes a scalable, closed-loop rehabilitation platform that leverages real-time haptic feedback and AI-driven personalization to optimize post-stroke upper-limb recovery, with potential applicability to other upper-limb impairments and neurorehabilitation contexts.

Project Overview

What This Project Is About

---

A straightforward exploration of using a wearable device to help people recover the use of their arm after a stroke. The project looks at adding gentle feedback to guide movements, tracking progress with smart artificial intelligence, and sharing simple results to help therapists and patients adjust therapy.

---

---

The Problem It Addresses

---

Many stroke survivors have trouble relearning arm movements, and current therapies can be repetitive or slow to adapt to a person’s needs. Without real-time feedback, patients may perform exercises incorrectly, slowing recovery. The project aims to provide an affordable, user-friendly way to improve practice quality and monitor progress over time.

---

---

Objectives of the Project

---

1. Design a comfortable wearable that can gently stimulate or guide arm movements.
2. Implement real-time feedback that tells the user if their movement is on target.
3. Incorporate AI to interpret progress and adjust difficulty automatically.
4. Create a simple data dashboard for therapists and patients.
5. Evaluate the system with a small group of users for safety and usefulness.

---

---

What You Will Do Step by Step

---

1) Review existing rehab tools and gather user needs. 2) Build the wearable hardware with sensors and haptic feedback. 3) Develop software that analyzes movement and provides real-time cues. 4) Train a basic AI model to track progress and personalize workouts. 5) Run a small study to collect user feedback and safety data. 6) Analyze results and refine the system. 7) Document design choices and potential improvements.

---

---

---

---

Expected Outcome

A functional wearable with easy-to-understand feedback and a simple AI-driven progress tracker. The project should show improved exercise quality, user satisfaction, and a clear path for future development or larger trials.