A Clinician-Driven Iteration Framework for Adaptive Music Instruments in Severe TBI Rehabilitation

Abstract

INTRODUCTION:"I want the software to follow the person, not the person follow the software." — Neurologic music therapist. Patients with severe TBI are excluded from rehabilitation because available tools cannot adapt to their physical, sensory, and cognitive function. Acoustic instruments require motor force. iPad apps produce false triggers. Speech-language therapy music accompaniment apps do not wait for uneven timing or hesitation. A design framework for clinician-driven personalization of adaptive instruments does not exist.OBJECTIVE:To describe whether a platform designed around three principles — hardware that responds to any gesture, modular code enabling rapid deployment, and weekly clinician feedback cycles — can increase access to music therapy for patients with severe TBI, through rapid iteration.METHODOLOGY:A neurologic music therapist and an assistive-technology occupational therapist at a US rehabilitation hospital used the platform during standard sessions with ten patients over four weeks. The pilot was deemed a practice-improvement and product-evaluation activity, outside IRB jurisdiction. Interventions were personalized or built between sessions.The platform combined: Commercial music interfaces: Expressive E Touché pressure-sensitive controller and Dubler 2 voice-to-MIDI microphone;Apple Mac running Logic Pro music software;Custom interactive code (Logic Pro Scripter). RESULTS: Over four weeks, the three design principles enabled the therapist to solve progressively complex access problems:  Week 1 — Adapting to patients' strength. Therapists explored different interventions — press, tap, slide — using the sensitivity knob on the Touché to adjust responsiveness. Minimal touch triggered sound; greater pressure unlocked richer output. "I like that sensitivity adjustment, it's right there." — Occupational therapist. The therapist described the shift:"'Oh my gosh, I'm finally doing something, something's happening' versus 'Oh my gosh, I'm trying so hard to pick up this ball or push a button and reach.'" Week 2 —Recognizing intentional movement. With a sliding intervention, therapists observed initial contact triggered sound, signaling success prematurely; patients stopped after the first touch. They requested that sound only trigger when a person starts to slide. This was delivered within the weekly cycle. Week 3 — Interventions that follow the patient. Therapists wanted an intervention for people only capable of gross arm movements. A four-movement intervention — push, pull, slide left, slide right — was built. The intervention waited for and followed the patient's timing. Week 4 — A novel intervention. Therapists wanted a singing intervention that patients could control independently. The team searched and found nothing comparable. An intervention was built that waits for the patient's uneven breath and pauses. CONCLUSIONS:Rapid feedback and iteration cycles increased access for patients excluded from existing tools and generated a novel intervention within weeks. The model was enabled by three design choices: hardware that responds to any gesture, a modular software architecture enabling rapid deployment, and a weekly feedback process. Minor environmental glitches with directional gestures were resolved after the pilot, and all requested interventions function reliably. Over four weeks, the platform supported personalization of interventions and the generation of a novel intervention — a trajectory that demonstrates clinician-driven iteration's capacity to increase access to rehabilitation.Further research with larger samples and quantitative measures is warranted.