ProsthetiX-AI: An LLM-based clinical decision support system for evidence-based prosthetic recommendations.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Rehabilitation Technology

Background:

• Conventional prosthetic selection for lower-limb amputees relies on subjective judgment and static protocols.
• Individualized patient factors are often overlooked, impacting rehabilitation outcomes.
• Need for objective, evidence-based decision support systems in prosthetic fitting.

Purpose of the Study:

• To introduce ProsthetiX-AI, a clinical decision support system for personalized prosthetic recommendations.
• To integrate deterministic policy, evidence-based reasoning, and large language models for transparent justification.
• To dynamically analyze patient-specific parameters for aligned prosthetic choices.

Main Methods:

• Developed ProsthetiX-AI, a web-based system analyzing amputation level, mobility, comorbidities, weight, and biomechanics.
• Integrated a large language model for evidence retrieval and explanation generation.
• Evaluated system performance on transtibial and transfemoral amputations with complex profiles.

Main Results:

• ProsthetiX-AI demonstrated sub-millisecond latency and reliable multi-user handling.
• Quantitative evaluation showed high accuracy (0.72-0.89) and Cohen's kappa (0.64-0.85), comparable to clinician variability.
• Clinicians and users reported high accuracy (4.76/5) and usability (4.54/5) ratings.

Conclusions:

• ProsthetiX-AI offers a transparent, scalable, and patient-centric approach to prosthetic selection.
• Citation-linked recommendations can augment clinical decision-making, enhancing interpretability.
• The system provides a foundation for AI adoption in healthcare, especially in resource-constrained settings.