mRUST Estimation of Tibial Fracture Healing After Intramedullary Nailing Using Deep Forest Model with a Genetically...

Area of Science:

• Orthopedics
• Biomedical Engineering
• Data Science

Background:

• Tibial shaft fracture healing assessment traditionally relies on radiographic methods (mRUST score), which involve radiation exposure and subjective interpretation.
• There is a need for objective, non-invasive, and cost-effective methods to monitor fracture healing.
• Portable insole systems offer a promising avenue for continuous gait data collection.

Purpose of the Study:

• To develop a quantitative model estimating mRUST scores using plantar pressure data.
• To identify an optimal, cost-effective sensor layout for a portable insole system.
• To provide an objective and radiation-free method for assessing tibial shaft fracture healing.

Main Methods:

• A Deep Forest Regression (DFR) model was developed to estimate mRUST scores from continuous plantar pressure data collected from 23 patients with tibial shaft fractures.
• Data from 515 gait analysis segments across 103 follow-up visits were utilized.
• A Genetic Algorithm (GA) optimized the sensor layout, with model interpretability assessed using Shapley Additive Explanations (SHAP).

Main Results:

• An optimal 6-sensor layout was identified, achieving a high coefficient of determination (R² = 0.902), comparable to a full 99-sensor array.
• The DFR model demonstrated high accuracy in estimating healing stages (early, intermediate, late).
• SHAP analysis confirmed clinical relevance, showing a shift in sensor contribution from heel to forefoot as healing progressed.

Conclusions:

• A DFR model utilizing a GA-optimized plantar pressure insole provides accurate and objective assessment of tibial fracture healing post-intramedullary nailing.
• This portable, data-driven approach serves as a viable alternative to radiographic methods.
• The system enables timely and convenient clinical monitoring of fracture recovery.