Related Experiment Video
Updated: Aug 20, 2025
09:51
The Transition to an Anterior-Based Muscle Sparing Approach Improves Early Postoperative Function but is Associated with a Learning Curve
Published on: September 7, 2022
2.8K
Highly accurate acoustical prediction using support vector machine algorithm for post-operative subsidence after
Yasuhiro Homma1,2, Xu Zhuang3, Hiroshi Ohtsu4
1Department of Medicine for Orthopaedics and Motor Organ, Graduate School of Medicine, Juntendo University, Tokyo, Japan. yhomma@juntendo.ac.jp.
International Orthopaedics
|November 23, 2022
Summary
Acoustic analysis and machine learning accurately predict post-operative subsidence in cementless total hip arthroplasty (THA). This technology can help avoid complications by monitoring the procedure acoustically.
Area of Science:
- Orthopedic Surgery
- Biomedical Engineering
- Machine Learning
Background:
- Acoustic analysis is a novel approach for evaluating cementless total hip arthroplasty (THA).
- Post-operative subsidence is a potential complication following THA that requires monitoring.
Purpose of the Study:
- To develop a machine learning algorithm for accurately predicting post-operative subsidence in cementless THA.
- To investigate the utility of acoustic parameters from the broaching procedure for subsidence prediction.
Main Methods:
- Analyzed acoustic parameters (normalized sound pressure in 24 frequency ranges) from hammering sounds during broaching in 62 THAs.
- Utilized a support vector machine with models incorporating acoustic data, patient demographics, and femoral morphology.
- Compared prediction accuracy across three models: acoustic data only, acoustic + patient data, and acoustic + patient + morphology data.
Main Results:
- The machine learning models demonstrated high accuracy in predicting subsidence (≥3 mm).
- Model C, incorporating acoustic parameters, patient demographics, and femoral morphology, achieved the highest accuracy (0.957) and AUC (0.998).
- All models showed excellent performance, with Model B and C achieving 100% specificity and positive predictive value.
Conclusions:
- A high-accuracy machine learning algorithm for predicting post-operative subsidence in cementless THA was successfully developed.
- Acoustic monitoring during THA broaching, combined with pre-operative features, shows promise for complication avoidance.
- This study represents a significant step towards implementing acoustic monitoring in clinical practice for cementless THA.
