This study investigates whether early changes observed during the first weeks of hand and upper extremity rehabilitation can predict patient outcomes months later. In rehabilitation practice, clinicians make numerous decisions each session regarding exercise type, frequency, duration, and treatment approach. Most of these decisions are currently made without systematic longitudinal data. This study addresses three fundamental questions using data collected during routine clinical care: (1) Can the rate of improvement in the first weeks of treatment predict functional status months later, across both session-based and milestone-based time points? (2) Are there meaningfully different recovery profiles among hand rehabilitation patients? (3) Is there measurable variation in clinical decision-making among patients with similar profiles, and does this variation relate to outcomes? A digital patient monitoring platform developed by the principal investigator, a physiotherapist and academic researcher specializing in hand rehabilitation serves as the data collection infrastructure. The platform records standard clinical assessment measures in a structured format and has been in active clinical use at Hacettepe University prior to this study. For research purposes, the system has been expanded to include structured capture of patient-reported outcomes, patient global impression of change, treatment protocol coding, home exercise adherence, and automated calculation of early response metrics.

This is a 12-month prospective observational cohort study enrolling a minimum of 60 patients. Data are stored securely on the university's institutional network. Patients are anonymized using identification codes. The study is subject to Hacettepe University Ethics Committee approval and participant informed consent. Findings are expected to generate evidence supporting data-driven clinical decision-making in rehabilitation and to provide a feasibility foundation for a larger multi-center study.

The study addresses three research questions: (1) predictive power of early functional response for later clinical outcomes across four cascaded models (M1-M4), (2) classification of patient recovery profiles, and (3) measurement of clinical decision variation. System and methodological details reported in study supplement.

Inclusion Criteria:

• Adults aged 18 years or older
• Hand or upper extremity pathology requiring rehabilitation, including but not limited to: flexor or extensor tendon injuries, peripheral nerve injuries or compression syndromes (including carpal tunnel syndrome, cubital tunnel syndrome), distal radius fractures, metacarpal or phalangeal fractures, joint injuries and instabilities, arthropathy (osteoarthritis, rheumatoid arthritis), trapeziometacarpal osteoarthritis, trigger finger, Dupuytren's disease, DeQuervain tenosynovitis, lateral epicondylitis and post-surgical upper extremity conditions
• Under clinical responsibility of the principal investigator at the Faculty of Physical Therapy and Rehabilitation, Hacettepe University
• Expected to attend minimum 8 rehabilitation sessions
• Able to provide written informed consent

Exclusion Criteria:

• • Active psychiatric illness precluding participation or completion of assessment tools

  • Severe cognitive impairment precluding understanding of patient-reported outcome measures
  • Literacy difficulties preventing self-completion of questionnaires
  • Concurrent enrollment in another interventional clinical study
  • Refusal to provide written informed consent

Adults with hand and upper extremity pathologies under the clinical responsibility of the principal investigator at Hacettepe University Faculty of Physical Therapy and Rehabilitation, Ankara, Turkey. Consecutive, non-probability sampling applied: all eligible patients enrolled in order of presentation until minimum n=60 is reached

Individual participant data will not be publicly shared to protect privacy. Aggregate data and complete R analysis code will be made available as supplementary material upon publication. De-identified study data may be made available upon reasonable written request to the corresponding author following publication.