AI-driven Personalized Exercise Feedback Program on Exercise Adherence in Traumatic Brain Injury

National Defense Medical Center, Taiwan

Status: Not yet enrolling

Conditions: Traumatic Brain Injury; Digital Health; Exercise; AI (Artificial Intelligence)

Treatments: Behavioral: Active Control; Behavioral: Theory-based digital exercise; Behavioral: AI-PEF

Study type: Interventional

Funder types: Other

Identifiers

NCT06817564

AI feedback digital exercise

Details and patient eligibility

About

This study aims to develop and evaluate an AI-driven Personalized Exercise Feedback Program (AI-PEF) to enhance exercise adherence and health outcomes in mTBI patients.

Methods: AI-PEF integrates the transtheoretical model and self-determination theory with machine learning algorithms to provide real-time, personalized feedback. A phased randomized controlled trial will be conducted: Phase I evaluates feasibility and acceptability through Delphi methods with expert consensus and patient feedback; Phase II validates preliminary outcomes with 30 participants in a 2-arm randomized trial; and Phase III assesses the program's impact on adherence, sleep quality, depressive symptoms, and quality of life with 90 participants in a 3-arm randomized trial.

Full description

The study will employ a stepwise, multi-phase design, combining a two- parallel-group pilot study and a three-arm randomized controlled trial (RCT) to evaluate the appropriateness, feasibility, acceptability, and effectiveness of the AI-PEF (Figure 5). Participants will be recruited from the neurosurgery clinics at Tri-Service General Hospital, Taipei. Recruitment will be facilitated through referrals by attending physicians and registered nurses, who will be briefed on the study protocol.

Study procedures

1. Phase I (Year 1): Development and Pilot Testing (Aim 1) Phase I will focus on developing and refining the AI-PEF through expert validation and pilot testing. This phase will use a sequential-parallel hybrid design to develop and evaluate the appropriateness of the AI-PEF to deliver personalized exercise empowerment. Stage 1, Delphi Process for AI-PEF Development: Ten experts from diverse fields will participate in two Delphi rounds to finalize the AI-PEF framework. This includes refining algorithms, educational materials, and preliminary validation of key components, such as personalization algorithms and behavior-change strategies. Stage 2, Small-Scale Pilot Study (Early Feedback): A single-arm pilot study with 5 patients will be conducted over four weeks to gather feedback on usability, engagement, and content clarity.
2. Phase II : Larger Pilot Study (Aim 2) Phase II will focus on evaluating the feasibility, acceptability, and preliminary effectiveness of the refined AI-PEF intervention. A two-parallel-group design will be employed, with 30 participants, 15 participants per group, randomly assigned in a 1:1 ratio to the AI-PEF group and Active control group. Over three months, AI-PEF participants will engage in personalized exercise guided by AI, while active control participants will follow standard exercise recommendations. Both qualitative and quantitative data will be collected. Assessments will occur at baseline (T0) and 3 months (T1), including fitness tracker data, questionnaires (motivation, sleep, symptoms), and semi-structured interviews. The primary outcome of interest will be adherence rates, while secondary outcomes will focus on motivation and various health metrics.
3. Phase III : Full- scale three-arm RCT (Aim 3) Phase III will assess the long-term impact of the AI- PEF on exercise adherence, motivation, and health outcomes through a 3-arm RCT randomly assigned to one of three groups: (1) the AI-PEF group, receiving a machine learning-powered personalized exercise program; (2) the Theory-based digital exercise group, engaging in structured digital exercise without machine learning- powered feedback; or (3) the Active control group, receiving general exercise recommendations as part of standard care in a 1:1:1 ratio. Over six months, participants will receive group-specific interventions, with assessments conducted at baseline (T0), 3 months (T1), and 6 months (T2). Primary outcomes, including exercise adherence, will be objectively measured via Garmin fitness trackers, while secondary outcomes, such as motivation, sleep quality, and symptom reduction, will be assessed through standardized questionnaires and qualitative interviews. These standardized questionnaires are being used in our current digital remote exercise trial in patients with mild TBI (NSTC 112-2314-B-016-007-MY2) and have performed well on physiological performance. Data will be collected and analyzed using both quantitative and qualitative methods to compare adherence, motivation, and health outcomes across the three groups, identifying the efficacy of AI-PEF relative to standard interventions. Procedures will adhere to the blinding and randomization protocols described in Intervention Fidelity, ensuring unbiased assignment and data collection processes. This proposed study protocol is closely aligned with those that have been successfully implemented in our previous digital exercise trial.

Enrollment: 125 estimated patients

Sex: All

Ages: 18+ years old

Volunteers: No Healthy Volunteers

Inclusion criteria

• Eligible participants are patients aged over 18 with mild TBI (GCS 13-15)
• who can walk independently,
• reside in the Greater Taipei area,
• and possess sufficient Chinese or Taiwanese language proficiency to understand the trial
• complete self-administered questionnaires.

Exclusion criteria

• Exclusion criteria include individuals with severe medical conditions (e.g., respiratory failure, epilepsy, psychiatric disorders), musculoskeletal or neurological impairments
• hindering physical activity in the 6-minute walk test,
• cognitive impairments (MMSE < 24),
• frontal lobe injuries or penetrating injury causing significant psychological dysfunction.
• Patients regularly engaging in moderate-to-high-intensity aerobic exercise or participating in other studies will also be excluded to avoid bias.

Trial design

Primary purpose: Supportive Care

Allocation: Randomized

Interventional model: Parallel Assignment

Masking: Triple Blind

---

125 participants in 3 patient groups

AI-PEF

Experimental group

Description:

the AI-PEF group, receiving a machine learning-powered personalized exercise program.

Treatment:

Behavioral: AI-PEF

Theory-based digital exercise

Experimental group

Description:

Theory-based digital exercise group, engaging in structured digital exercise without machine learning- powered feedback.

Treatment:

Behavioral: Theory-based digital exercise

Active control group

Active Comparator group

Description:

the Active control group, receiving general exercise recommendations as part of standard care in a 1:1:1 ratio.

Treatment:

Behavioral: Active Control