OBJECTIVE: To demonstrate applying American Association for Thoracic Surgery Quality Gateway (AQG) outcomes models to a Surgeon Case Study of quality assurance in adult cardiac surgery.
METHODS: The case study includes 6,989 cardiac and thoracic aorta operations performed in adults at Cleveland Clinic by one surgeon from 2001 to 2023. AQG models were used to predict expected probabilities for operative mortality and major morbidity, and to compare hospital outcomes, surgery type, risk profile, and individual risk-factor levels using virtual (digital) twin causal inference. These models were based on postoperative procedural outcomes after 52,792 cardiac operations performed in 19 hospitals of 3 high-performing hospital systems with overall hospital mortality of 2.0%, analyzed by advanced machine learning for rare events.
RESULTS: For individual surgeons, their patients, hospitals, and hospital systems, the Surgeon Case Study demonstrated that AQG provides expected outcomes across the entire spectrum of cardiac surgery, from single-component primary operations to complex multi-component reoperations. Actionable opportunities for quality improvement based on virtual twins is illustrated for patients, surgeons, hospitals, risk profile groups, operations, and risk factors vis-à-vis other hospitals.
CONCLUSIONS: Using minimal data collection and models developed using advanced machine learning, this case study shows that probabilities can be generated for operative mortality and major morbidity after virtually all adult cardiac operations. It demonstrates the utility of 21st century causal inference (virtual [digital] twin) tools for assessing quality for surgeons asking \"How am I doing?\" their patients asking \"What are my chances?\" and the profession asking \"How can we get better?\"

BACKGROUND: A lack of consensus exists across guidelines as to which risk model should be used for the primary prevention of cardiovascular disease (CVD). Our objective was to determine potential improvements in the number needed to treat (NNT) and number of events prevented (NEP) using different risk models in patients eligible for risk stratification.
METHODS: A retrospective observational cohort was assembled from primary care patients in Ontario, Canada between January 1st, 2010, to December 31st, 2014 and followed for up to 5 years. Risk estimation was undertaken in patients 40-75 years of age, without CVD, diabetes, or chronic kidney disease using the Framingham Risk Score (FRS), Pooled Cohort Equations (PCEs), a recalibrated FRS (R-FRS), Systematic Coronary Risk Evaluation 2 (SCORE2), and the low-risk region recalibrated SCORE2 (LR-SCORE2).
RESULTS: The cohort consisted of 47,399 patients (59% women, mean age 54). The NNT with statins was lowest for SCORE2 at 40, followed by LR-SCORE2 at 41, R-FRS at 43, PCEs at 55, and FRS at 65. Models that selected for individuals with a lower NNT recommended statins to fewer, but higher risk patients. For instance, SCORE2 recommended statins to 7.9% of patients (5-year CVD incidence 5.92%). The FRS, however, recommended statins to 34.6% of patients (5-year CVD incidence 4.01%). Accordingly, the NEP was highest for the FRS at 406 and lowest for SCORE2 at 156.
CONCLUSIONS: Newer models such as SCORE2 may improve statin allocation to higher risk groups with a lower NNT but prevent fewer events at the population level.

OBJECTIVE: mpMRI is routinely used to stratify the risk of clinically significant prostate cancer (csPCa) in men with elevated PSA values before biopsy. This study aimed to calculate a multivariable risk model incorporating standard risk factors and mpMRI findings for predicting csPCa on subsequent prostate biopsy.
METHODS: Data from 677 patients undergoing mpMRI ultrasound fusion biopsy of the prostate at the TUM University Hospital tertiary urological center between 2019 and 2023 were analyzed. Patient age at biopsy (67 (median); 33-88 (range) (years)), PSA (7.2; 0.3-439 (ng/ml)), prostate volume (45; 10-300 (ml)), PSA density (0.15; 0.01-8.4), PI-RADS (V.2.0 protocol) score of index lesion (92.2% ≥3), prior negative biopsy (12.9%), suspicious digital rectal examination (31.2%), biopsy cores taken (12; 2-22), and pathological biopsy outcome were analyzed with multivariable logistic regression for independent associations with the detection of csPCa defined as ISUP ≥ 3 (n = 212 (35.2%)) and ISUP ≥ 2 (n = 459 (67.8%) performed on 603 patients with complete information.
RESULTS: Older age (OR: 1.64 for a 10-year increase; p < 0.001), higher PSA density (OR: 1.60 for a doubling; p < 0.001), higher PI-RADS score of the index lesion (OR: 2.35 for an increase of 1; p < 0.001), and a prior negative biopsy (OR: 0.43; p = 0.01) were associated with csPCa.
CONCLUSIONS: mpMRI findings are the dominant predictor for csPCa on follow-up prostate biopsy. However, PSA density, age, and prior negative biopsy history are independent predictors. They must be considered when discussing the individual risk for csPCa following suspicious mpMRI and may help facilitate the further diagnostical approach.

OBJECTIVE: The multifactorial dynamic perfusion index was recently introduced as a predictor of cardiac surgery-associated acute kidney injury. The multifactorial dynamic perfusion index was developed based on retrospective data retrieved from the patient files. The present study aims to prospectively validate this index in an external series of patients, through an on-line measure of its various components.
METHODS: Inclusion criteria were adult patients undergoing cardiac surgery with cardiopulmonary bypass. Data collection included preoperative factors and cardiopulmonary bypass-related factors. These were collected on-line using a dedicated monitor. Factors composing the multifactorial dynamic perfusion index are the nadir haematocrit, the nadir oxygen delivery, the time of exposure to a low oxygen delivery, the nadir mean arterial pressure, cardiopulmonary bypass duration, the use of red blood cell transfusions and the peak arterial lactates.
RESULTS: Two hundred adult patients were investigated. The multifactorial dynamic perfusion index had a good (c-statistics 0.81) discrimination for cardiac surgery-associated acute kidney injury (any stage) and an excellent (c-statistics 0.93) discrimination for severe patterns (stage 2-3). Calibration was modest for cardiac surgery-associated acute kidney injury (any stage) and good for stage 2-3. The use of vasoconstrictors was an additional factor associated with cardiac surgery-associated acute kidney injury.
CONCLUSIONS: The multifactorial dynamic perfusion index is validated for discrimination of cardiac surgery-associated acute kidney injury risk. It incorporates modifiable risk factors, and may help in reducing the occurrence of cardiac surgery-associated acute kidney injury.

OBJECTIVE: Risk models for mortality after percutaneous coronary intervention (PCI) are underutilized in clinical practice though they may be useful during informed consent, risk mitigation planning, and risk adjustment of hospital and operator outcomes. This review analyzed contemporary risk models for in-hospital and 30-day mortality after PCI.
RESULTS: We reviewed eight contemporary risk models. Age, sex, hemodynamic status, acute coronary syndrome type, heart failure, and kidney disease were consistently found to be independent risk factors for mortality. These models provided good discrimination (C-statistic 0.85-0.95) for both pre-catheterization and comprehensive risk models that included anatomic variables. There are several excellent models for PCI mortality risk prediction. Choice of the model will depend on the use case and population, though the CathPCI model should be the default for in-hospital mortality risk prediction in the United States. Future interventions should focus on the integration of risk prediction into clinical care.

Glioma is a prevalent type of malignant tumor. To date, there is a lack of literature reports that have examined the association between sulfatase modifying factor 1 (SUMF1) and glioma.
The levels of SUMF1 were examined, and their relationships with the diagnosis, prognosis, and immune microenvironment of patients with glioma were investigated. Cox and Lasso regression analysis were employed to construct nomograms and risk models associated with SUMF1. The functions and mechanisms of SUMF1 were explored and verified using gene ontology, cell counting kit-8, wound healing, western blotting, and transwell experiments.
SUMF1 expression tended to increase in glioma tissues. SUMF1 overexpression was linked to the diagnosis of cancer, survival events, isocitrate dehydrogenase status, age, and histological subtype and was positively correlated with poor prognosis in patients with glioma. SUMF1 overexpression was an independent risk factor for poor prognosis. SUMF1-related nomograms and high-risk scores could predict the outcome of patients with glioma. SUMF1 co-expressed genes were involved in cytokine, T-cell activation, and lymphocyte proliferation. Inhibiting the expression of SUMF1 could deter the proliferation, migration, and invasion of glioma cells through epithelial mesenchymal transition. SUMF1 overexpression was significantly associated with the stromal score, immune cells (such as macrophages, neutrophils, activated dendritic cells), estimate score, immune score, and the expression of the programmed cell death 1, cytotoxic T-lymphocyte associated protein 4, CD79A and other immune cell marker.
SUMF1 overexpression was found to be correlated with adverse prognosis, cancer detection, and immune status in patients with glioma. Inhibiting the expression of SUMF1 was observed to deter the proliferation, migration, and invasion of cancer cells. The nomograms and risk models associated with SUMF1 could predict the prognosis of patients with glioma.

In the last 30 years, there has been an increasing incidence of oral cancer worldwide. Earlier detection of oral cancer has been shown to improve survival rates. However, given the relatively low prevalence of this disease, population-wide screening is likely to be inefficient. Risk prediction models could be used to target screening to those at highest risk or to select individuals for preventative interventions. This review (a) systematically identified published models that predict the development of oral cancer and are suitable for use in the general population and (b) described and compared the identified models, focusing on their development, including risk factors, performance and applicability to risk-stratified screening. A search was carried out in November 2022 in the Medline, Embase and Cochrane Library databases to identify primary research papers that report the development or validation of models predicting the risk of developing oral cancer (cancers of the oral cavity or oropharynx). The PROBAST tool was used to evaluate the risk of bias in the identified studies and the applicability of the models they describe. The search identified 11,222 articles, of which 14 studies (describing 23 models), satisfied the eligibility criteria of this review. The most commonly included risk factors were age (n = 20), alcohol consumption (n = 18) and smoking (n = 17). Six of the included models incorporated genetic information and three used biomarkers as predictors. Including information on human papillomavirus status was shown to improve model performance; however, this was only included in a small number of models. Most of the identified models (n = 13) showed good or excellent discrimination (AUROC > 0.7). Only fourteen models had been validated and only two of these validations were carried out in populations distinct from the model development population (external validation). Conclusions: Several risk prediction models have been identified that could be used to identify individuals at the highest risk of oral cancer within the context of screening programmes. However, external validation of these models in the target population is required, and, subsequently, an assessment of the feasibility of implementation with a risk-stratified screening programme for oral cancer.

OBJECTIVE: While several predictive models for the development of hepatocellular carcinoma (HCC) have been proposed, including those for patients with chronic hepatitis C virus (HCV) infection who have achieved sustained virologic response (SVR), the best model may differ between regions. We compared the ability of six reported models to stratify the risk of post-SVR HCC in Japan, where rigorous surveillance and early detection of HCC is common.
METHODS: A total of 6048 patients with no history of HCC who achieved SVR by oral direct-acting antiviral drugs were enrolled in this nationwide study. Patients continued HCC surveillance every 6 months after SVR. The incidence of post-SVR HCC was compared between risk groups using the aMAP score, FIB-4 index, Tahata model, GAF4 criteria, GES score, and ADRES score.
RESULTS: During the observation period with a median duration of 4.0 years after SVR, post-SVR HCC developed in 332 patients (5.5%). All six models performed significantly at stratifying the incidence of HCC. However, Harrell\'s C-index was below 0.8 for all models (range, 0.660-0.748), indicating insufficient stratification ability.
CONCLUSIONS: Although all six proposed models demonstrated a good ability to predict the development of post-SVR HCC, their ability to stratify the risk of post-SVRHCC was unsatisfactory. Further studies are necessary to identify the best model for assessing the risk of post-SVR HCC in regions where early detection of HCC is common.

Objective: To develop a risk score model for the occurrence of composite cardiovascular events (CVE) in patients with stable angina pectoris (SA) combined with coronary heart disease (CHD) by comparing the modeling effects of various machine learning (ML) algorithms. Methods: In this prospective study, 690 patients with SA combined with CHD attending the Department of Integrative Cardiology, China-Japan Friendship Hospital, from October 2020 to October 2021 were included. The data set was randomly divided into a training group and a testing group in a 7:3 ratio in the per-protocol set (PPS). Model variables were screened using the least absolute shrinkage selection operator (LASSO) regression, univariate analysis, and multifactor logistic regression. Then, nine ML algorithms are integrated to build the model and compare the model effects. Individualized risk assessment was performed using the SHapley Additive exPlanation (SHAP) and nomograms, respectively. The model discrimination was evaluated by receiver operating characteristic curve (ROC), the calibration ability of the model was evaluated by calibration plot, and the clinical applicability of the model was evaluated by decision curve analysis (DCA). This study was approved by the Clinical Research Ethics Committee of China-Japan Friendship Hospital (2020-114-K73). Results: 690 patients were eligible to finish the complete follow-up in the PPS. After LASSO screening and multifactorial logistic regression analysis, physical activity level, taking antiplatelets, Traditional Chinese medicine treatment, Gensini score, Seattle Angina Questionnaire (SAQ)-exercise capacity score, and SAQ-anginal stability score were found to be predictors of the occurrence of CVE. The above predictors are modeled, and a comprehensive comparison of the modeling effectiveness of multiple ML algorithms is performed. The results show that the Light Gradient Boosting Machine (LightGBM) model is the best model, with an area under the curve (AUC) of 0.95 (95% CI = 0.91-1.00) for the test set, Accuracy: 0.90, Sensitivity: 0.87, and Specificity: 0.96. Interpretation of the model using SHAP highlighted the Gensini score as the most important predictor. Based on the multifactorial logistic regression modeling, a nomogram, and online calculators have been developed for clinical applications. Conclusion: We developed the LightGBM optimization model and the multifactor logistic regression model, respectively. The model is interpreted using SHAP and nomogram. This provides an option for early prediction of CVE in patients with SA combined with CHD.

The incidence of venous thromboembolism (VTE) in patients with lung cancer is relatively high, and risk stratification models are vital for the targeted application of thromboprophylaxis. We aimed to review VTE risk prediction models that have been developed in patients with lung cancer and evaluated their performance.
Twenty-four eligible studies involving 123,493 patients were included. The pooled incidence of VTE within 12 months was 11 % (95 % CI 8 %-14 %). With the identified four VTE risk assessment tools, meta-analyses did not show a significant discriminatory capability of stratifying VTE risk for Khorana, PROTECHT and CONKO scores. The pooled sensitivity and specificity of the Khorana score were 24 % (95 % CI 11 %-44 %) and 84 % (95 % CI 73 %-91 %) at the 3-point cut-off, and 43 % (95 % CI 35 %-52 %) and 61 % (95 % CI 52 %-69 %) at the 2-point cut-off. However, a COMPASS-CAT score of ≥ 7 points indicated a significantly high VTE risk, with a RR of 4.68 (95 % CI 1.05-20.80).
The Khorana score lacked discriminatory capability in identifying patients with lung cancer at high VTE risk, regardless of the cut-off value. The COMPASS-CAT score had better performance, but further validation is needed. The results indicate the need for robust VTE risk assessment tools specifically designed and validated for lung cancer patients. Future research should include relevant biomarkers as important predictors and consider the combined use of risk tools. PROSPERO registration number: CRD42021245907.