Using deep learning to predict postoperative pain in reverse shoulder arthroplasty patients.

Area of Science:

• Orthopedic Surgery
• Artificial Intelligence
• Pain Management

Background:

• Postsurgical pain is a critical outcome in shoulder arthroplasty, yet receives less research focus than functional recovery.
• Predicting and managing postsurgical pain is essential for patient well-being after orthopedic procedures.

Purpose of the Study:

• To develop a deep learning algorithm for predicting postsurgical pain following reverse total shoulder arthroplasty (rTSA).
• To identify key preoperative factors associated with low postsurgical pain after rTSA.

Main Methods:

• An artificial neural network was trained using preoperative data from 1707 rTSA patients.
• Data included demographics, disease information, clinical assessments, and self-report measures.
• Postsurgical pain was a binary outcome (pain score ≥3 vs. ≤2), and the model was validated internally.

Main Results:

• The deep learning model achieved 63% accuracy in predicting postsurgical pain at 2 years post-rTSA.
• SHapley Additive exPlanations identified factors for low pain: low ASA score, low DASH score, private insurance, primary OA, and lower baseline pain.

Conclusions:

• A deep learning model was successfully developed to predict postsurgical pain after rTSA.
• Further refinement is needed, but the model can optimize surgical indications and manage patient expectations.
• This predictive tool holds potential for improving clinical decision-making in shoulder arthroplasty.