Predicting soft tissue changes after orthognathic surgery: The sparse partial least squares method.

Area of Science:

• Orthognathic surgery
• Medical prediction algorithms
• Facial reconstructive surgery

Background:

• Accurate prediction of soft tissue changes is crucial for successful orthognathic surgery outcomes.
• Existing prediction models may require extensive input variables, limiting their practical application.
• Developing a versatile and efficient prediction algorithm is essential for improving patient care.

Purpose of the Study:

• To develop a prediction algorithm for soft tissue changes following orthognathic surgery.
• To ensure accurate predictions across diverse surgical types and complexities.
• To minimize the number of input variables required for prediction.

Main Methods:

• Utilized partial least squares (PLS) and sparse partial least squares (SPLS) methods.
• Applied these methods to a cohort of 318 patients undergoing Class II or III malocclusion correction.
• Compared the predictive accuracy of the full PLS model (232 variables) with the reduced SPLS model (34 variables).

Main Results:

• No significant differences in prediction accuracy were observed based on surgical movements, patient sex, or additional procedures.
• The SPLS method successfully reduced the number of input variables to 34.
• The predictive performance of the SPLS model was statistically indistinguishable from the full PLS model.

Conclusions:

• The proposed prediction method demonstrates accuracy across a wide spectrum of orthognathic surgeries.
• The SPLS method enables the selection of a reduced variable set while maintaining high prediction accuracy.
• This approach offers a more efficient and practical tool for predicting soft tissue changes in orthognathic surgery.