Low back pain expert systems: Clinical resolution through probabilistic considerations and poset

Area of Science:

• Artificial Intelligence in Healthcare
• Medical Informatics
• Clinical Decision Support Systems

Background:

• Low Back Pain (LBP) diagnosis is challenging, especially in developing countries like India, due to implementation difficulties of existing guidelines.
• Diverse management approaches across medical specialties complicate standardized LBP evaluation.
• Existing AI-based LBP diagnostic methods have limitations in clinical and computational efficiency.

Purpose of the Study:

• To develop an Artificial Intelligence (AI)-based technique for expert-level diagnosis of Low Back Pain (LBP).
• To create a clinically justified and highly sensitive LBP resolution methodology.
• To enhance the efficiency and accuracy of LBP diagnosis in clinical practice.

Main Methods:

• Utilized lattice structures to represent exhaustive knowledge of LBP disorders, ensuring completeness and optimality.
• Developed a hierarchical network, RuleNet, using partially-ordered sets (poset) for knowledge enhancement.
• Incorporated probability within RuleNet for reliable resolution logic and uncertainty management in clinical decision-making.

Main Results:

• Validated the proposed methodology using clinical records of 77 LBP patients.
• Achieved 83% sensitivity, deemed clinically satisfactory by pain experts.
• Inferred outcomes demonstrated homogeneity with actual diagnoses.

Conclusions:

• The AI-driven LBP resolution technique offers clinical and computational efficiency, overcoming limitations of existing methods.
• The system provides a fast, reliable, and affordable healthcare solution for LBP when embedded in medical expert systems.
• The proposed scheme is expected to reduce treatment controversies, confusion, and costs associated with unnecessary referrals.