A Prediction Model for Functional Outcomes in Spinal Cord Disorder Patients Using Gaussian Process Regression

Area of Science:

• Neurosurgery
• Rehabilitation Medicine
• Machine Learning in Healthcare

Background:

• Accurate prediction of functional outcomes for spinal cord disorder patients post-treatment is crucial for effective clinical management and patient care.
• Existing prediction methods may not fully account for the specific constraints of outcome variables in spinal cord injury recovery.
• Objective assessment of functional status, such as through target tracking, is valuable but requires integration with predictive models.

Purpose of the Study:

• To develop and validate a novel prediction method for postoperative functional outcomes in patients with cervical spinal cord disorder.
• To leverage Gaussian process regression with a truncated Normal distribution for improved prediction accuracy, considering restricted variable ranges.
• To integrate data from preoperative patient information and handgrip target tracking examinations into the prediction model.

Main Methods:

• Utilized Gaussian process regression, a machine learning technique, for predicting patient outcomes.
• Modeled the Gaussian process using a truncated Normal distribution to accommodate the restricted value range of target variables (e.g., Oswestry Disability Index, target tracking scores).
• Incorporated preoperative patient data and results from handgrip target tracking examinations using a portable device.

Main Results:

• The proposed method demonstrated high accuracy in predicting postoperative functional outcomes.
• Achieved a mean absolute error of 0.079 for predicting Oswestry Disability Index scores.
• Achieved a mean absolute error of 0.014 for predicting target tracking scores (normalized to a 0-1 scale).

Conclusions:

• The novel Gaussian process regression method effectively predicts postoperative functional outcomes for spinal cord disorder patients.
• The use of a truncated Normal distribution significantly enhances prediction accuracy by respecting variable constraints.
• Integration of preoperative data and objective handgrip tracking measurements improves the reliability of functional outcome predictions.