ASSIST AS NEEDED NANOROBOTIC NEUROREHABILITATION Motor-state estimation × minimum necessary assistance × transfer without assistance STARLI Research Works | Institute XXIII | Flagship 08 Author: Feng Cheng-en 33 × Starli Edition: Research Edition v1.0 | Nonclinical exploratory monograph PUBLICATION DESCRIPTION Can an adaptive rehabilitation device improve learning without quietly replacing the very human capability it aims to restore? This research edition investigates the distinction between successful machine-assisted task completion, independent motor capacity, and retained skill after assistance is withdrawn. The volume places assist-as-needed control inside a falsifiable scientific architecture. It considers latent fatigue and capability estimation, compliant actuation, feedback delay, voluntary effort, treatment-policy confounding, masked assistance withdrawal, cross-task transfer, fairness of comparisons, and reproducible nonclinical research. It evaluates whether complex adaptive strategies confer meaningful advantage over fixed-gain guidance, conventional rehabilitation robotics, dose-matched approaches, and carefully specified no-assistance controls. Its speculative micro- and nano-scale extension is intentionally bounded. The available references support motor learning and existing macroscopic robotic rehabilitation; they do not establish a safe or effective autonomous nanorobotic clinical intervention. No self-experimentation, human stimulation settings, invasive device instructions, or therapeutic claims are presented. RESEARCH CONTENT • Five-layer conceptual architecture: state estimation, minimum support, withdrawal, external transfer and successor reconstruction. • Thirty model derivation cards linked to individual chapters and Research Gates. • Eighty-two chapter-opening research theses and 410 separately paginated research sections. • Two hundred detailed Research Gates, each with a disprovable question, experimental sketch, competitor, stopping condition and follow-up route. • Twenty-four contextual DOI references with explicit distinctions between existing evidence and open hypotheses. • 111,011 machine-counted English words within the 410 chapter sections, plus 200 expanded Research Gates and mathematical prelude; phrase-level repetition remains substantial and independent originality certification is pending. The central principle is that the last scientific arbiter of rehabilitation is not how effectively a device moves, but what remains possible when the device no longer supplies the movement. Negative findings are treated as productive limits that can guide future investigators toward simpler, safer and more independently verifiable solutions.