Related Experiment Video
Updated: Apr 13, 2026
Predicting Amputation using Local Circulating Mononuclear Progenitor Cells in Angioplasty-treated Patients with Critical Limb Ischemia
Published on: September 22, 2020
Utility of Machine Learning in Predicting Catastrophic Cardiac Complications Following Hip and Knee Periprosthetic
William T Sampson1, Isaiah A Freeman1, Michelle R Shimizu1
1Department of Orthopaedic Surgery, Bioengineering Laboratory, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts.
Insights
Machine learning models accurately predict rare cardiac complications like myocardial infarction and cardiac arrest after periprosthetic fracture surgery. This aids in identifying high-risk patients and developing interventions to improve outcomes.
Area of Science:
- Orthopedic Surgery
- Cardiology
- Data Science
Background:
- Rising prevalence of total hip (THA) and knee arthroplasties (TKA) leads to increased periprosthetic fractures (PPF).
- Myocardial infarction (MI) and cardiac arrest (CA) are severe complications following PPF surgery.
- Limited research exists on using machine learning (ML) for predicting rare outcomes in revision arthroplasty.
Purpose of the Study:
- Develop and evaluate ML models to predict cardiac complications (MI and CA) after PPF surgery.
- Assess the accuracy and utility of different ML algorithms for rare event prediction in orthopedic surgery.
- Identify key predictors of cardiac complications in patients undergoing PPF surgery.
Main Methods:
- Extracted data from a national quality improvement database for patients with PPF undergoing revision arthroplasty or open reduction internal fixation (ORIF).
- Developed four ML models: artificial neural network, random forest, histogram-based gradient boosting, and k-nearest neighbor.
- Assessed model performance using Area Under the Curve (AUC), calibration slopes, intercepts, and Brier scores.
Main Results:
- Included 3,805 PPF patients (1,869 revision, 1,936 ORIF).
- Reported low incidence of MI (1.81-1.97%) and CA (0.67-0.92%) post-surgery.
- All ML models demonstrated excellent predictive power (AUCs 0.76-0.94) and accuracy (Brier scores 0.013-0.044).
- Key predictors included elevated serum creatinine, white blood cell count, sodium, and lower body mass index.
Conclusions:
- ML models show high accuracy in predicting rare postoperative cardiac complications after PPF surgery.
- Demonstrates the effectiveness of ML in identifying high-risk patients for adverse cardiac events.
- Suggests potential for targeted interventions to mitigate cardiac risks in PPF patients.
Background:
With the rising prevalence of primary total hip (THA) and knee arthroplasties (TKAs), the number of periprosthetic fractures (PPFs) is expected to increase. Myocardial infarction (MI) and cardiac arrest (CA) are devastating complications following PPF surgery. There is a paucity of literature demonstrating the utility of machine learning (ML) in predicting rare clinical outcomes in revision arthroplasty. This study aimed to develop ML models to predict cardiac complications following PPF surgery.
Methods:
Patients who underwent revision arthroplasty or open reduction internal fixation (ORIF) due to PPF were extracted from a national quality improvement database. There were four ML models that were developed to predict MI and CA: artificial neural network, random forest, histogram-based gradient boosting, and k-nearest neighbor. Area under the curve (AUC) was used to assess discriminatory ability. Validation was performed with calibration slopes, intercepts, and Brier scores.
Results:
The final population included 3,805 PPF surgery patients, consisting of 1,869 revision arthroplasties (1,378 THAs, 491 TKAs) and 1,936 ORIF (811 THAs, 1,125 TKAs) procedures. Among revisions, 1.97% suffered MI (27 THAs, 10 TKAs) and 0.92% suffered CA (13 THAs, five TKAs). In ORIF patients, 1.81 and 0.67% experienced MI (14 THAs, 21 TKAs) and CA (one THA, 12 TKAs), respectively. All MI models yielded AUCs and Brier scores between 0.82 and 0.91 and 0.031 and 0.044, respectively. The CA analyses were similar, with AUCs between 0.76 and 0.94 and Brier scores between 0.013 and 0.037. Elevated preoperative serum creatinine, white blood cell count, and sodium, as well as lower body mass index were the most critical predictors.
Conclusions:
All models demonstrated excellent predictive power and accuracy in assessing postoperative cardiac complications. This highlights the effectiveness of ML in predicting rare, catastrophic outcomes, and offers potential areas of intervention to mitigate the risk of adverse cardiac events.