This study tests whether an artificial intelligence (AI)-enhanced telerehabilitation program can effectively treat chronic non-specific low back pain in college students.

Low back pain affects 40-52% of university students due to prolonged sitting during lectures and study sessions, poor posture from laptop use, and lack of physical activity. While exercise therapy is the recommended treatment, many students cannot access traditional physiotherapy due to cost, scheduling conflicts, and location barriers.

This randomized controlled trial compares three treatment approaches: (1) AI-enhanced telerehabilitation with automated video analysis and personalized feedback, (2) standard telerehabilitation with video instructions only, and (3) usual care. The AI system uses computer vision technology (Google MediaPipe Pose) to analyze exercise videos through a standard webcam or smartphone, automatically tracking joint movements, counting repetitions, and providing real-time feedback on exercise form.

College students with chronic low back pain (lasting more than 3 months) will be randomly assigned to one of the three groups. The AI-enhanced group will receive personalized exercise programs delivered remotely, with the AI system monitoring their performance and physiotherapists providing guidance through video consultations.

The study will measure changes in pain levels, disability, physical function, trunk muscle endurance, and quality of life over 8 weeks of treatment and 3 months of follow-up. Researchers will also evaluate how well participants stick to their exercise programs and how easy the technology is to use.

This research aims to determine if AI technology can make remote physiotherapy more effective and accessible for college students, potentially transforming how young adults receive treatment for back pain and improving their long-term health outcomes.

Chronic non-specific low back pain (CNSLBP) has emerged as a significant health concern among college students, with international studies reporting prevalence rates between 40-52%. This high incidence is attributed to the modern academic environment, characterized by prolonged static postures during lectures and study sessions, extensive use of laptops and handheld devices leading to poor trunk alignment, and generally low levels of structured physical activity resulting in deconditioning of core and postural muscles.

The impact of CNSLBP in college students extends beyond physical discomfort, affecting academic performance, causing absenteeism, limiting recreational participation, and potentially leading to persistent pain patterns in adulthood. Current evidence-based management guidelines recommend multidisciplinary approaches emphasizing structured exercise therapy, self-management education, and postural retraining, with particular focus on flexibility, core stability, and functional strength exercises.

However, significant barriers prevent college students from accessing optimal care. These include logistical challenges such as academic scheduling conflicts, economic constraints related to repeated physiotherapy visits, and geographical accessibility issues. Consequently, many students resort to unsupervised home exercise programs that, while cost-effective and flexible, lack real-time monitoring and professional guidance, often resulting in incorrect technique, poor adherence, and suboptimal outcomes.

TECHNOLOGICAL INNOVATION

Recent advances in artificial intelligence (AI) and computer vision technology offer promising solutions to bridge this care gap. Markerless motion capture systems, particularly Google's MediaPipe Pose and OpenPose, can analyze human movement using standard cameras to identify skeletal landmarks, track joint angles, assess posture, and detect movement deviations in real-time. These systems demonstrate approximately 85% accuracy for gross movement tracking and exercise repetition counting, making them suitable for clinical rehabilitation applications.

When integrated into telerehabilitation platforms, AI-driven video analysis provides dual benefits: enhancing remote care quality by supplying therapists with quantitative performance feedback, and enabling patients to receive immediate automated correction cues, thereby improving engagement and self-efficacy.

STUDY DESIGN AND METHODOLOGY

This single-blind, three-arm parallel-group randomized controlled trial will compare the effectiveness of AI-enhanced telerehabilitation versus standard telerehabilitation and usual care in college students with CNSLBP. The study design addresses a critical research gap, as no published randomized controlled trials have specifically examined AI-enhanced telerehabilitation in this population.

INTERVENTION GROUPS

Group 1: AI-Enhanced Telerehabilitation Participants will receive personalized exercise programs delivered through a custom platform incorporating AI-based movement analysis. The system uses computer vision algorithms to monitor exercise performance through participants' webcams or smartphones, providing real-time feedback on form, automatically counting repetitions, measuring hold times, and flagging technique errors. Physiotherapists will review AI-generated performance data and provide personalized guidance through scheduled video consultations.

Group 2: Standard Telerehabilitation Participants will receive exercise programs via video instructions without AI monitoring or personalized feedback. This group represents current telerehabilitation practice, relying on pre-recorded exercise videos and periodic therapist consultations without objective movement analysis.

Group 3: Usual Care Control Participants will receive standard medical care as typically provided for CNSLBP, which may include general advice on activity modification, over-the-counter pain medications, and basic exercise recommendations without structured supervision.

PARTICIPANT SELECTION

The study will recruit college students aged 18-25 years with chronic non-specific low back pain (duration >3 months) from university health services and campus recruitment. Inclusion criteria ensure participants have clinically significant symptoms warranting intervention, while exclusion criteria eliminate cases with specific pathologies requiring specialized medical management.

OUTCOME MEASURES

Primary outcomes include pain intensity measured using the Numerical Rating Scale (NRS), functional disability assessed with the Roland-Morris Disability Questionnaire (RMDQ), functional mobility evaluated through the Timed Up and Go (TUG) test, and trunk muscular endurance measured via the prone plank test. These validated instruments are sensitive to clinical changes in low back pain populations.

Secondary outcomes encompass adherence rates to prescribed exercises, platform usability assessed through standardized questionnaires, quality of life measures, healthcare utilization patterns, and long-term follow-up assessments at 3 months post-intervention.

STATISTICAL ANALYSIS

The study will employ intention-to-treat analysis as the primary approach, with per-protocol analysis as secondary. Power calculations indicate adequate sample size to detect clinically meaningful differences between groups. Mixed-effects models will account for repeated measurements and potential confounding variables.

EXPECTED OUTCOMES AND SIGNIFICANCE

This research aims to establish whether AI-enhanced telerehabilitation can provide superior clinical outcomes compared to standard approaches while maintaining high usability and adherence rates. The findings have potential to inform university health services, influence physiotherapy practice guidelines, and support broader integration of AI technology into telehealth delivery.

The study aligns with the World Health Organization's Global Strategy on Digital Health 2020-2025, which encourages leveraging digital innovation to improve healthcare accessibility and equity. By focusing on a tech-savvy demographic experiencing significant barriers to traditional care, this research addresses both immediate clinical needs and broader healthcare delivery challenges.

INNOVATION AND FUTURE IMPLICATIONS

This study represents a pioneering application of AI technology in rehabilitation for young adults, potentially establishing a scalable model for remote physiotherapy delivery. The integration of objective movement analysis with personalized professional guidance offers a novel approach that maintains therapeutic relationships while leveraging technological capabilities for enhanced monitoring and feedback.

The research contributes to the growing body of evidence supporting digital health interventions while specifically addressing the unique needs and circumstances of college students with chronic low back pain. Success could lead to broader implementation across university health systems and expansion to other musculoskeletal conditions and age groups.

QUALITY ASSURANCE

The study incorporates rigorous methodological standards including randomization procedures, blinding where possible, validated outcome measures, standardized intervention protocols, and comprehensive statistical analysis plans. Regular monitoring ensures protocol adherence and participant safety throughout the study period.

Inclusion Criteria

• Age 18-30 years, currently enrolled in undergraduate or postgraduate study.
• Diagnosis of non-specific LBP for at least 3 months.
• Baseline pain intensity between 3 and 7 (NRS).
• Ability and willingness to perform prescribed exercises and participate in video conferencing.
• Access to suitable device and reliable internet.
• Informed consent obtained. Exclusion Criteria
• Specific causes of LBP (e.g., fracture, tumor, infection, inflammatory disease).
• Recent spinal surgery or confirmed disc herniation (within past year).
• Neurological deficits or severe comorbid conditions contraindicating exercise.
• Pregnancy or current participation in another structured LBP program.
• BMI ≥ 35 kg/m² (could impair AI pose detection).
• Inability to understand English instructions or complete measures.

Outcome assessors evaluating primary endpoints (pain, disability, mobility, trunk endurance) are blinded to group allocation to prevent measurement bias. Independent research assistants conducting assessments remain unaware of participants' intervention assignments.

Participants cannot be blinded due to intervention nature - they know whether receiving AI-enhanced telerehabilitation, standard telerehabilitation, or usual care. Physiotherapists delivering interventions cannot be blinded as they provide group-specific treatments.

Data analysts remain blinded to group codes during statistical analysis until primary analyses complete. Principal investigator maintains randomization knowledge for safety monitoring but doesn't participate in outcome measurements.