Development of a Machine Learning Algorithm for Prediction of Complications after Ankle Arthrodesis

P125. Development of a novel ensemble machine learning algorithm for prediction of complications and readmission after anterior cervical spinal fusion

The aging and multimorbid population and health personnel shortages pose a substantial burden on primary health care. While predictive machine learning (ML) algorithms have the potential to address these challenges, concerns include transparency and insufficient reporting of model validation and effectiveness of the implementation in the clinical workflow. To systematically identify predictive ML algorithms implemented in primary care from peer-reviewed literature and US Food and Drug Administration (FDA) and Conformité Européene (CE) registration databases and to ascertain the public availability of evidence, including peer-reviewed literature, gray literature, and technical reports across the artificial intelligence (AI) life cycle. PubMed, Embase, Web of Science, Cochrane Library, Emcare, Academic Search Premier, IEEE Xplore, ACM Digital Library, MathSciNet, AAAI.org (Association for the Advancement of Artificial Intelligence), arXiv, Epistemonikos, PsycINFO, and Google Scholar were searched for studies published between January 2000 and July 2023, with search terms that were related to AI, primary care, and implementation. The search extended to CE-marked or FDA-approved predictive ML algorithms obtained from relevant registration databases. Three reviewers gathered subsequent evidence involving strategies such as product searches, exploration of references, manufacturer website visits, and direct inquiries to authors and product owners. The extent to which the evidence for each predictive ML algorithm aligned with the Dutch AI predictive algorithm (AIPA) guideline requirements was assessed per AI life cycle phase, producing evidence availability scores. The systematic search identified 43 predictive ML algorithms, of which 25 were commercially available and CE-marked or FDA-approved. The predictive ML algorithms spanned multiple clinical domains, but most (27 [63%]) focused on cardiovascular diseases and diabetes. Most (35 [81%]) were published within the past 5 years. The availability of evidence varied across different phases of the predictive ML algorithm life cycle, with evidence being reported the least for phase 1 (preparation) and phase 5 (impact assessment) (19% and 30%, respectively). Twelve (28%) predictive ML algorithms achieved approximately half of their maximum individual evidence availability score. Overall, predictive ML algorithms from peer-reviewed literature showed higher evidence availability compared with those from FDA-approved or CE-marked databases (45% vs 29%). The findings indicate an urgent need to improve the availability of evidence regarding the predictive ML algorithms' quality criteria. Adopting the Dutch AIPA guideline could facilitate transparent and consistent reporting of the quality criteria that could foster trust among end users and facilitating large-scale implementation.

12074 Background: Skeletal-related events (SREs) are detrimental clinical events in bone metastasis from advanced cancer, defined by pathologic fracture, spinal cord compression, and inevitable surgical or radiational intervention to the bone. Given their negative impact on quality of life and prognosis and interpatient heterogeneity in the SRE risk, accurate identification of patients with high SRE risk is critical. Methods: The patient-level data from three randomized clinical trials that administered zoledronic acid (ZA) to patients with bone-metastatic breast cancer, castration-resistant prostate cancer (CRPC), and other types of cancer were analyzed. (N = 460, 452, and 315, respectively) Machine learning (ML) models to predict SREs within 18 months (breast cancer), 12 months (CRPC), and 9 months (other cancers) were developed based on more than 40 baseline clinical and laboratory data. Seven ML algorithms and five feature selection methods were utilized to develop multiple models. The ML model with the best performance was identified based on the F1 score and the area under the receiver operating characteristic curve (AUC-ROC) for each cancer type and interrogated for important features with Shapley additive explanations. Lastly, the ML models’ ability to stratify patients by the cumulative SRE risk was evaluated by calculating hazard ratio (HR) with Cox-proportional hazards models. Results: Among the multiple ML models developed with different algorithms and feature selection methods, the model developed utilizing the random forests algorithm and the Boruta method for selecting features demonstrated the best performance in all types of cancer (F1 0.70, 0.67, and 0.67, and AUC-ROC 0.72, 0.68, and 0.73 for breast cancer, CRPC, and other cancers, respectively). In the ML model for breast cancer, performance status (PS), history of SRE, serum alkaline phosphatase (ALP), history of anti-neoplastic surgery, radiation therapy, and pathologic fracture were included as important features. Serum ALP, albumin, sodium, Gleason scores, and geographic regions were shown to be relevant in the CRPC model. For other cancers, serum ALP, albumin, total protein, phosphorus, red blood cell count, white blood cell count, visceral metastases, and history of arthritis were incorporated in the ML model. The ML model prediction successfully stratified the patients for cumulative SRE risk in all three cohorts. (HR with 95% confidence interval: 2.43 [1.86 – 3.18], 1.92 [1.51 – 2.45], and 3.06 [2.29 – 4.09] for breast cancer, CRPC, and other types of cancer, respectively). Conclusions: ML models incorporating baseline clinical and laboratory data can identify patients with bone metastasis on ZA harboring a high SRE risk.

Development of a Machine Learning Algorithm for Prediction of Complications and Unplanned Readmission Following Reverse Total Shoulder Arthroplasty

BackgroundCervical cancer is a common malignant tumor of the female reproductive system and is considered a leading cause of mortality in women worldwide. The analysis of time to event, which is crucial for any clinical research, can be well done with the method of survival prediction. This study aims to systematically investigate the use of machine learning to predict survival in patients with cervical cancer.MethodAn electronic search of the PubMed, Scopus, and Web of Science databases was performed on October 1, 2022. All articles extracted from the databases were collected in an Excel file and duplicate articles were removed. The articles were screened twice based on the title and the abstract and checked again with the inclusion and exclusion criteria. The main inclusion criterion was machine learning algorithms for predicting cervical cancer survival. The information extracted from the articles included authors, publication year, dataset details, survival type, evaluation criteria, machine learning models, and the algorithm execution method.ResultsA total of 13 articles were included in this study, most of which were published from 2018 onwards. The most common machine learning models were random forest (6 articles, 46%), logistic regression (4 articles, 30%), support vector machines (3 articles, 23%), ensemble and hybrid learning (3 articles, 23%), and Deep Learning (3 articles, 23%). The number of sample datasets in the study varied between 85 and 14946 patients, and the models were internally validated except for two articles. The area under the curve (AUC) range for overall survival (0.40 to 0.99), disease-free survival (0.56 to 0.88), and progression-free survival (0.67 to 0.81), respectively from (lowest to highest) received. Finally, 15 variables with an effective role in predicting cervical cancer survival were identified.ConclusionCombining heterogeneous multidimensional data with machine learning techniques can play a very influential role in predicting cervical cancer survival. Despite the benefits of machine learning, the problem of interpretability, explainability, and imbalanced datasets is still one of the biggest challenges. Providing machine learning algorithms for survival prediction as a standard requires further studies.

Metabolic dysfunction-associated steatotic liver disease (MASLD) can progress to metabolic dysfunction-associated steatohepatitis (MASH) and liver fibrosis, contributing to a heavier global health burden. Non-invasive diagnostic tools developed using machine learning (ML) and deep learning (DL), two representative artificial intelligence algorithms, are increasingly being explored for MASH and its related fibrosis assessment. This study aimed to compare the diagnostic performance of different ML and DL models and identify the top-performing models for diagnosing MASH and associated liver fibrosis. A systematic review and meta-analysis were conducted across PubMed, Web of Science, Embase and Cochrane Library from inception to May 18, 2025. Pooled area under the receiver operator characteristic curve (AUROC) values with 95 % confidence interval (CI) were calculated. Accuracy, specificity, sensitivity, positive predictive values, and negative predictive values were also recorded. Of 4,314 studies initially identified, 106 met the inclusion criteria, with 35 studies (ML: n=28; DL: n=7) providing data for analysis. Logistic Regression and Neural Network are the most commonly algorithms applied in ML and DL, respectively. The pooled AUROCs for diagnosing MASH were 0.833 (95 %CI: 0.806-0.860) for ML models and 0.841 (95 %CI: 0.782-0.900) for DL models. Light Gradient Boosting Machine (LightGBM) and ResNet50 were the best-performing models for diagnosing MASH within ML and DL algorithms, respectively, achieving corresponding AUROCs of 0.920 (95 %CI: 0.916-0.924) and 0.960 (95 %CI: 0.951-0.969). For fibrosis diagnosis, ML models had a pooled AUROC of 0.826 (95 %CI: 0.792-0.860), with Categorical Boosting (CatBoost) achieving the highest AUROC of 0.960 (95 %CI: 0.950-0.970). DL models yielded the pooled AUROC of 0.875 (95 %CI: 0.816-0.934) for fibrosis diagnosis. Both ML and DL models demonstrated strong diagnostic performance for MASH and liver fibrosis, with DL achieving marginally higher AUROCs. AI-driven approaches show promise in MASLD management.

ObjectiveInferior vena cava (IVC) filter placement is associated with important long-term complications. Predictive models for filter-related complications may help guide clinical decision-making but remain limited. We developed machine learning (ML) algorithms that predict 1-year IVC filter complications using preoperative data. MethodsThe Vascular Quality Initiative database was used to identify patients who underwent IVC filter placement between 2013 and 2024. We identified 77 preoperative demographic and clinical features from the index hospitalization when the filter was placed. The primary outcome was 1-year filter-related complications (composite of filter thrombosis, migration, angulation, fracture, and embolization or fragmentation, vein perforation, new caval or iliac vein thrombosis, new pulmonary embolism, access site thrombosis, or failed retrieval). The data were divided into training (70%) and test (30%) sets. Six ML models were trained using preoperative features with 10-fold cross-validation (Extreme Gradient Boosting, random forest, Naïve Bayes classifier, support vector machine, artificial neural network, and logistic regression). The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was assessed using calibration plot and Brier score. Performance was evaluated across subgroups based on age, sex, race, ethnicity, rurality, median Area Deprivation Index, planned duration of filter, landing site of filter, and presence of prior IVC filter placement. ResultsOverall, 14,476 patients underwent IVC filter placement and 584 (4.0%) experienced 1-year filter-related complications. Patients with a primary outcome were younger (59.3 ± 16.7 years vs 63.8 ± 16.0 years; P < .001) and more likely to have thrombotic risk factors including thrombophilia, prior venous thromboembolism (VTE), and family history of VTE. The best prediction model was Extreme Gradient Boosting, achieving an AUROC of 0.93 (95% confidence interval, 0.92-0.94). In comparison, logistic regression had an AUROC of 0.63 (95% confidence interval, 0.61-0.65). Calibration plot showed good agreement between predicted/observed event probabilities with a Brier score of 0.07. The top 10 predictors of 1-year filter-related complications were (1) thrombophilia, (2) prior VTE, (3) antiphospholipid antibodies, (4) factor V Leiden mutation, (5) family history of VTE, (6) planned duration of IVC filter (temporary), (7) unable to maintain therapeutic anticoagulation, (8) malignancy, (9) recent or active bleeding, and (10) age. Model performance remained robust across all subgroups. ConclusionsWe developed ML models that can accurately predict 1-year IVC filter complications, performing better than logistic regression. These algorithms have potential to guide patient selection for filter placement, counselling, perioperative management, and follow-up to mitigate filter-related complications and improve outcomes.

Background This study aimed to establish and validate a machine learning (ML) model for predicting in-hospital mortality in critically ill patients with chronic kidney disease (CKD). Methods This study collected data on CKD patients from 2008 to 2019 using the Medical Information Mart for Intensive Care IV. Six ML approaches were used to build the model. Accuracy and area under the curve (AUC) were used to choose the best model. In addition, the best model was interpreted using SHapley Additive exPlanations (SHAP) values. Results There were 8527 CKD patients eligible for participation; the median age was 75.1 (interquartile range: 65.0–83.5) years, and 61.7% (5259/8527) were male. We developed six ML models with clinical variables as input factors. Among the six models developed, the eXtreme Gradient Boosting (XGBoost) model had the highest AUC, at 0.860. According to the SHAP values, the sequential organ failure assessment score, urine output, respiratory rate, and simplified acute physiology score II were the four most influential variables in the XGBoost model. Conclusions In conclusion, we successfully developed and validated ML models for predicting mortality in critically ill patients with CKD. Among all ML models, the XGBoost model is the most effective ML model that can help clinicians accurately manage and implement early interventions, which may reduce mortality in critically ill CKD patients with a high risk of death.

Introduction: Early and accurate diagnosis of transthyretin amyloid cardiomyopathy (ATTR-CM) is known to improve prognosis. However, it is sometimes challenging to distinguish ATTR-CM from other cardiac conditions that cause left ventricular hypertrophy (LVH), such as hypertrophic cardiomyopathy (HCM), hypertensive LVH (h-LVH), and aortic stenosis (AS). The Mayo ATTR-CM score is commonly used to distinguish ATTR-CM from these mimickers. Herein, we constructed a machine learning (ML) model and compared its performance to the Mayo score. Hypothesis: A ML model based on readily available clinical variables will outperform the Mayo score in distinguishing ATTR-CM from other cardiac conditions that cause LVH. Methods: We performed a multicenter case-control study of 1,605 patients, including 303 ATTR-CM cases and 1,302 controls – i.e., 935 with HCM, 331 with h-LVH, and 36 with AS. We allocated the former two-thirds of each disease group to the training set and the latter one-third to the prospective test set for validation. We developed a main ML model in the training set using XGBoost with 18 clinical variables including demographics, comorbidities, laboratory values, and echocardiographic data. We examined its performance in the test set using the area under the receiver-operating-characteristic curve (AUC) and decision curve analysis (DCA). In addition, we developed a simplified ML model using only the top 5 variables identified by feature importance and tested its performance. Results: The top 5 most important variables were age, LV diameter in systole, LV posterior wall thickness in diastole, LV ejection fraction, and left atrial diameter ( Figure 1 ). The main ML model (AUC: 0.96, 95% CI: 0.93–0.98) outperformed the Mayo score (AUC: 0.84, 95% CI: 0.79–0.89, DeLong’s P &lt;0.0001; Figure 2 ) in the test set. DCA demonstrated that the ML model provided greater net benefit across all threshold probabilities ( Figure 3 ). The simplified 5-variable model also outperformed the Mayo score (AUC: 0.94, 95% CI: 0.91–0.97, DeLong’s P &lt;0.0001). Conclusions: In this multicenter study, we developed ML models that effectively distinguish ATTR-CM from its mimickers that cause LVH using readily available clinical variables. Given the simplicity and accuracy of our models, the present study serves as an important initial step toward the development of a robust, data-driven tool that could be readily automated and integrated into electronic medical record system to screen for ATTR-CM.

Genetic testing can determine family screening strategies and has prognostic and diagnostic value in hypertrophic cardiomyopathy (HCM). However, it can also pose a significant psychosocial burden. Conventional scoring systems offer modest ability to predict genotype positivity. The aim of our study was to develop a novel prediction model for genotype positivity in patients with HCM by applying machine learning (ML) algorithms. We constructed 3 ML models using readily available clinical and cardiac imaging data of 102 patients from Columbia University with HCM who had undergone genetic testing (the training set). We validated model performance on 76 patients with HCM from Massachusetts General Hospital (the test set). Within the test set, we compared the area under the receiver operating characteristic curves (AUROCs) for the ML models against the AUROCs generated by the Toronto HCM Genotype Score (the Toronto score) and Mayo HCM Genotype Predictor (the Mayo score) using the Delong test and net reclassification improvement. Overall, 63 of the 178 patients (35%) were genotype positive. The random forest ML model developed in the training set demonstrated an AUROC of 0.92 (95% CI, 0.85-0.99) in predicting genotype positivity in the test set, significantly outperforming the Toronto score (AUROC, 0.77 [95% CI, 0.65-0.90], P=0.004, net reclassification improvement: P<0.001) and the Mayo score (AUROC, 0.79 [95% CI, 0.67-0.92], P=0.01, net reclassification improvement: P=0.001). The gradient boosted decision tree ML model also achieved significant net reclassification improvement over the Toronto score (P<0.001) and the Mayo score (P=0.03), with an AUROC of 0.87 (95% CI, 0.75-0.99). Compared with the Toronto and Mayo scores, all 3 ML models had higher sensitivity, positive predictive value, and negative predictive value. Our ML models demonstrated a superior ability to predict genotype positivity in patients with HCM compared with conventional scoring systems in an external validation test set.

P68. Machine learning algorithms for predicting patient-controlled analgesia use after minimally invasive spine surgery

Introduction: Falls occur in all age-groups and represent a significant public health concern. Previous studies have implemented artificial intelligence, including machine learning (ML) and deep learning (DL) algorithms for fall risk prediction, but the comparative performance between models and the applicability for younger populations remains unclear. This study aims to develop and compare different ML/DL models and identify key predictive features across age-groups. Methods: We enrolled 1,441 community-dwelling adults aged over 20 years in southern Taiwan and collected demographic, clinical, and physical performance data. Participants were categorized based on fall history. Five ML models (KNN, RF, GBDT, XGBoost, and CatBoost) and two DL models (GRU, AGRU) were trained and evaluated using accuracy, precision, recall, F1 score, and the area under the receiver operating characteristic curve (AUROC). Feature importance was interpreted using SHapley Additive exPlanations values in the best-performing model. Age-stratified subgroup analyses were conducted for groups aged 20–45, 46–65, and >65 years. Results: The AGRU model achieved the highest accuracy (91.39%) and AUROC (0.934) in the overall group and outperformed other models across all subgroups. Feature importance analysis revealed pulse rate, living alone, systolic blood pressure, 5-times Sit-to-Stand test, and sex as major predictors of falls in the overall group. The top five predictive factors varied across age-groups. Conclusion: We developed a robust and interpretable DL model to identify fall risk across different age-groups. Age-specific risk factors highlight the need for tailored preventive strategies. External validation using an independent dataset demonstrated moderate generalizability. Validating the model on larger, more diverse datasets and integration of sequential or sensor-based data are essential for practical applications.

Stent malapposition (SM) following percutaneous coronary intervention (PCI) for myocardial infarction continues to present significant clinical challenges. In recent years, machine learning (ML) models have demonstrated potential in disease risk stratification and predictive modeling. ML models based on optical coherence tomography (OCT) imaging, laboratory tests, and clinical characteristics can predict the occurrence of SM. We studied 337 patients from the Affiliated Hospital of Zunyi Medical University, China, who had PCI and coronary OCT from May to October 2023. We employed nested cross-validation to partition patients into training and test sets. We developed five ML models: XGBoost, LR, RF, SVM, and NB based on calcification features. Performance was assessed using ROC curves. Lasso regression selected features from 46 clinical and 21 OCT imaging features, which were optimized with the five ML algorithms. In the prediction model based on calcification features, the XGBoost model and SVM model exhibited higher AUC values. Lasso regression identified five key features from clinical and imaging data. After incorporating selected features into the model for optimization, the AUC values of all algorithmic models showed significant improvements. The XGBoost model demonstrated the highest calibration accuracy. SHAP values revealed that the top five ranked features influencing the XGBoost model were calcification length, age, coronary dissection, lipid angle, and troponin. ML models developed using plaque imaging features and clinical characteristics can predict the occurrence of SM. ML models based on clinical and imaging features exhibited better performance.

Introduction: Cardiovascular-kidney-metabolic (CKM) syndrome significantly impacts clinical outcomes, though evidence linking integrated cardiometabolic-kidney biomarkers to prognosis remains sparse. This study evaluated prognostic associations of these biomarkers and developed machine learning (ML)-based mortality prediction models for CKM patients. Methods: Using NHANES data (1999-2018) and death records from 10,616 stage 0–3 CKM patients, we analyzed cardiometabolic-kidney indices: cardiometabolic index (CMI), atherogenic index of plasma (AIP), estimated glomerular filtration rate (eGFR), and urinary albumin-creatinine ratio (uACR). Survival analysis incorporated the Kaplan-Meier curves, Cox regression, and restricted cubic splines to evaluate nonlinear associations. Risk reclassification was quantified via net reclassification index (NRI) and integrated discrimination improvement (IDI). Optimal mortality thresholds were determined using survival cut-point analysis, and inflammation’s mediating role was explored. Seven ML models were trained, with performance assessed by area under the receiver operating characteristic curve (AUC-ROC), Brier score, and net clinical benefit. Results: Over a median 96-month follow-up, 847 deaths occurred. Elevated CMI, AIP, and uACR, along with reduced eGFR, independently predicted mortality (all p < 0.05), with nonlinear trends for CMI, eGFR, and uACR (p-nonlinearity < 0.05). High-risk thresholds for these indices increased mortality risk by 1.19–1.91-fold. Combining all indices improved risk stratification (NRI = 15.8%, IDI = 3.4%). Inflammation mediated 1.1–5.0% of biomarker-mortality associations. Among ML models, XGBoost achieved optimal performance (AUC = 0.852, 95% CI: 0.829–0.877), with Brier score of 0.063 (95% CI: 0.056–0.069) and provided clinical net benefits across risk thresholds from 0 to 0.6. Conclusion: Cardiometabolic-kidney indices significantly associated with prognosis in CKM patients, highlighting the importance of heart-kidney-metabolism crosstalk. Combining easily accessible biomarkers with the XGBoost model may facilitate risk stratification.

The screening and diagnosis of osteoporosis in patients with type 2 diabetes mellitus (T2DM) based on bone mineral density remains challenging because of the limited availability and accessibility of dual-energy X-ray absorptiometry. We aimed to develop and validate models to predict the risk of osteoporosis in postmenopausal women with T2DM based on machine learning (ML) algorithms. This retrospective study included 303 postmenopausal women with T2DM. To develop prediction models for osteoporosis, we applied nine ML algorithms combined with demographic, clinical, and laboratory data. The least absolute shrinkage and selection operator were used to perform feature selection. We used the bootstrap resampling technique for model training and validation. To test the performance of the models, we calculated indices including the area under the receiver operating characteristic curve (AUROC), accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, calibration curve, and decision curve analysis. Furthermore, we conducted fivefold cross-validation for parameter optimization and model validation. Feature importance was assessed using the SHapley additive explanation (SHAP). We identified 10 independent predictors as the most valuable features. An AUROC of 0.616-1.000 was observed for nine ML algorithms. The extreme gradient boosting (XGBoost) model exhibited the best performance, outperforming conventional risk assessment tools and registering 0.993 in the training set, 0.798 in the validation set, and 0.786 in the test set for fivefold cross-validation. Using SHAP, we found that the explanatory variables contributed to the model and their relationship with osteoporosis occurrence. Furthermore, we developed a user-friendly tool for calculating the risk of osteoporosis. With the integration of demographic and clinical risk factors, ML algorithms can accurately predict osteoporosis. The XGBoost model showed ideal performance. With the incorporation of these models in the clinic, patients may benefit from early osteoporosis diagnosis and treatment.