Enhancing Large Medical Equipment Health-Aware Control with Bayesian Graph Attention Transformer-Based Probabilistic...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Predictive Maintenance

Background:

• Large medical equipment (MRI, CT, linear accelerators) is vital for healthcare quality.
• Current maintenance strategies (scheduled, reactive) are inefficient, causing downtime and costs.
• Uncertainty in equipment health data hinders effective health-aware control (HAC) and maintenance.

Purpose of the Study:

• To develop a probabilistic remaining useful life (RUL) prediction method for large medical equipment.
• To improve health-aware control (HAC) and maintenance scheduling by addressing prediction uncertainty.
• To enhance the reliability and longevity of critical healthcare infrastructure.

Main Methods:

• A Bayesian graph attention transformer model was developed for RUL prediction.
• Graph attention networks extracted spatial relationships from sensor data.
• Transformer models captured temporal dependencies, enabling joint spatiotemporal analysis.
• An improved Bayesian network quantified RUL prediction uncertainty (confidence intervals).

Main Results:

• The proposed method achieved more accurate and reliable RUL predictions for CT and MRI equipment.
• Experimental results demonstrated superiority over existing RUL prediction techniques.
• The framework effectively supports HAC decisions by balancing equipment performance and longevity.

Conclusions:

• The novel probabilistic RUL prediction method enhances maintenance scheduling and HAC strategies.
• Accurate RUL prediction with uncertainty quantification improves the management of critical medical equipment.
• This approach contributes to optimizing healthcare infrastructure performance and patient care.