Smart 3D-Printed Assistive Device for Hand Rehabilitation Based on Real-Time Motion Tracking and AI-Guided Therapy Protocols

Project Abstract

This study presents the development and evaluation of a smart 3D-printed assistive device for hand rehabilitation that integrates real-time motion tracking with AI-guided therapy protocols to enhance motor recovery after neurological injury. The device combines a lightweight, modular hand exoskeleton printed from biocompatible polymers with embedded flex sensors and a compact inertial measurement unit (IMU) array to capture finger and hand kinematics at high resolution. Real-time motion data are streamed to a detachable processing unit running edge AI models that classify movement primitives, detect compensatory strategies, and quantify performance metrics such as range of motion, grip strength, coordination, and dexterity. The AI component employs a multi-task architecture that fuses kinematic features with patient-specific data, including baseline impairment level, fatigue state, and therapy history, to personalize therapy trajectories. A reinforcement learning module dynamically adapts exercise difficulty, progression rules, and feedback modalities to maximize engagement, adherence, and neuroplastic potential, while a safety layer monitors joint torques and range limits to prevent discomfort or injury. The therapeutic protocol comprises progressively challenging tasks aligned with conventional occupational therapy goals, such as precise finger coupling, grasp-and-release sequences, and functional object manipulation. Real-time feedback is delivered through multimodal cues—haptic micro-stimulation, proprioceptive actuators, visual progress dashboards, and gamified interfaces—to reinforce correct movement patterns and maintain motivation. The system supports remote monitoring and data sharing with clinicians via a secure cloud platform, enabling iterative adjustments to the therapy plan without in-person visits. A rigorous pilot study with a diverse cohort of post-stroke and spinal cord injury patients evaluated usability, safety, and preliminary efficacy over a 6-week intervention period. Primary outcomes included improvements in Functional Independence Measure (FIM) scores related to hand function, Box and Block Test performance, and grip strength. Secondary outcomes encompassed adherence rates, user satisfaction, sensor accuracy, and computational latency. Results indicate high user acceptance, with mean System Usability Scale (SUS) scores exceeding benchmark thresholds and minimal adverse events reported. Quantitative gains were observed in active finger ROM, coordination indexes, and task completion times, demonstrating statistically significant improvements compared to baseline and to conventional therapy in matched controls. The AI-driven personalization scheme contributed to more rapid attainment of therapy milestones and reduced sedentariness by sustaining optimal challenge levels. Robust cross-validation confirmed the model's generalizability across etiologies and severities, while ablation analyses highlighted the critical roles of real-time feedback, adaptive progression, and precise motion sensing in driving functional gains. The 3D-printed fabrication workflow enables rapid customization, cost-effective production, and straightforward maintenance, making the device scalable for clinical and home-based rehabilitation. This work advances hand rehabilitation by delivering an integrated, patient-centric platform that harmonizes precise motion tracking, intelligent therapy modulation, and accessible fabrication. The findings support broader adoption of intelligent assistive devices in neurorehabilitation and lay groundwork for future enhancements, including multimodal physiological sensing, cloud-based data-driven optimization, and augmented reality-assisted therapy.

Project Overview

What This Project Is About

---

A simple device and software system designed to help people regain finger and hand motion after injury or stroke. The project combines a small, printable device that assists hand movement with sensors that track how the hand moves in real time, and software that uses artificial intelligence to tailor therapy exercises to each learner's progress.

---

---

The Problem It Addresses

---

Many patients struggle with repetitive, boring therapy and limited access to personalized guidance. Without frequent, guided practice, recovery can stall. This project aims to provide an affordable, engaging way to do daily hand therapy at home or in clinics, with feedback that adapts to the user’s abilities. It also helps therapists monitor progress remotely.

---

---

Objectives of the Project

---

1. Design a 3D-printed assistive device that is comfortable, safe, and easy to assemble.
2. Integrate sensors to capture real-time hand movement data.
3. Develop AI-based guidance to personalize therapy workouts.
4. Create a user-friendly interface for patients and therapists.
5. Evaluate usability and basic effectiveness with a small group of participants.

---

---

What You Will Do Step by Step

---

1. Review existing hand rehabilitation devices and therapy approaches.
2. Design the mechanical hand device and select suitable sensors.
3. 3D-print the components and assemble the prototype.
4. Collect motion data while participants perform simple tasks.
5. Develop an AI model that recommends exercises based on progress.
6. Implement a basic software app for feedback and progress tracking.
7. Test with volunteers to gather usability feedback.
8. Analyze data to look for trends in improvement and engagement.

---

---

Expected Outcome

A functional, low-cost hand rehabilitation device with real-time motion tracking and personalized AI-guided therapy plans, plus initial data on usability and potential effectiveness to inform larger studies.