Vitamin D deficiency is strongly associated with the development of several diseases. In the current context of a global pandemic of vitamin D deficiency, it is critical to identify people at high risk of vitamin D deficiency. There are no prediction tools for predicting the risk of vitamin D deficiency in the general community population, and this study aims to use machine learning to predict the risk of vitamin D deficiency using data that can be obtained through simple interviews in the community.
The National Health and Nutrition Examination Survey 2001-2018 dataset is used for the analysis which is randomly divided into training and validation sets in the ratio of 70:30. GBM, LR, NNet, RF, SVM, XGBoost methods are used to construct the models and their performance is evaluated. The best performed model was interpreted using the SHAP value and further development of the online web calculator.
There were 62,919 participants enrolled in the study, and all participants included in the study were 2 years old and above, of which 20,204 (32.1%) participants had vitamin D deficiency. The models constructed by each method were evaluated using AUC as the primary evaluation statistic and ACC, PPV, NPV, SEN, SPE, F1 score, MCC, Kappa, and Brier score as secondary evaluation statistics. Finally, the XGBoost-based model has the best and near-perfect performance. The summary plot of SHAP values shows that the top three important features for this model are race, age, and BMI. An online web calculator based on this model can easily and quickly predict the risk of vitamin D deficiency.
In this study, the XGBoost-based prediction tool performs flawlessly and is highly accurate in predicting the risk of vitamin D deficiency in community populations.

BACKGROUND: This study aims to develop a post-procedural risk prediction model for permanent pacemaker implantation (PPMI) in patients treated with transcatheter aortic valve replacement (TAVR).
METHODS: 336 patients undergoing TAVR at a single institution were included for model derivation. For primary analysis, multivariate logistic regression model was used to evaluate predictors and a risk score system was devised based on the prediction model. For secondary analysis, a Cox proportion hazard model was performed to assess characteristics associated with the time from TAVR to PPMI. The model was validated internally via bootstrap and externally using an independent cohort.
RESULTS: 48 (14.3%) patients in the derivation set had PPMI after TAVR. Prior right bundle branch block (RBBB, OR: 10.46; p < 0.001), pre-procedural aortic valve area (AVA, OR: 1.41; p = 0.004) and post- to pre-procedural AVA ratio (OR: 1.72; p = 0.043) were identified as independent predictors for PPMI. AUC was 0.7 and 0.71 in the derivation and external validation set. Prior RBBB (HR: 5.07; p < 0.001), pre-procedural AVA (HR: 1.33; p = 0.001), post-procedural AVA to prosthetic nominal area ratio (HR: 0.02; p = 0.039) and post- to pre-procedural troponin-T difference (HR: 1.72; p = 0.017) are independently associated with time to PPMI.
CONCLUSIONS: The post-procedural prediction model achieved high discriminative power and accuracy for PPMI. The risk score system was constructed and validated, providing an accessible tool in clinical setting regarding the Chinese population.

OBJECTIVE: This study aimed to explore the influencing factors of quality of life (QOL) and establish a prediction model in patients with primary nephrotic syndrome (PNS).
METHODS: A single-centre retrospective study was conducted on 245 patients with PNS admitted to Zibo Central Hospital from August 2020 to August 2022. According to the 36-Item Short-Form Health Survey (SF-36) for QOL evaluation, the patients were divided into the good QOL group (the total score ≥50 points) and poor QOL group (the total score <50 points). Univariate analysis was conducted by collecting clinical data from patients, and multiple logistic regression analysis was carried out on single factors with statistically significant differences to construct a clinical prediction model. The diagnostic efficacy of the prediction model was evaluated using the receiver operating characteristic (ROC) curve.
RESULTS: A total of 245 questionnaires were distributed, and 243 valid questionnaires were recovered, in which 143 cases had good QOL, with an average score of (71.86 ± 10.83) points, and 100 cases had poor QOL, with an average score of (40.03 ± 5.95) points. Statistical differences were observed in age, education level, monthly family average income, payment methods of medical expenses, albumin, 24-hour urinary protein quantification (24 h UPro) and serum uric acid (SUA) in both groups (p < 0.05), whereas no statistical difference was found in gender, body mass index (BMI) and marital status (p > 0.05). The multiple logistic regression analysis showed that age (X1), monthly family average income (X2), payment methods of medical expenses (X3), albumin (X4), 24 h UPro (X5) and SUA (X6) were risk factors for the QOL of patients with PNS, with Y = -12.105 + 0.130X1 + 0.457X2 + 0.448X3 + -0.161X4 + 0.823X5 + 0.025X6 as the regression prediction model. The results of ROC curve showed that the area under the curve (AUC) was 0.987 with standard error of 0.005 (p < 0.001), and 95% CI was 0.976-0.998.
CONCLUSIONS: Age, monthly family average income, payment methods of medical expenses, albumin, 24 h UPro and SUA are risk factors that affect the QOL of patients with PNS, and the construction of prediction model has good evaluation value and can provide a reference for clinical practice.

OBJECTIVE: To establish a computed tomography (CT)-based scale to evaluate the resectability of locally advanced thyroid cancer.
METHODS: This twin-centre retrospective study included 95 locally advanced thyroid cancer patients from the 1st centre as the training cohort and 31 patients from the 2nd centre as the testing cohort, who were categorised into the resectable and unresectable groups. Three radiologists scored the CT scans of each patient by evaluating the extension to the recurrent laryngeal nerve (RLN), trachea, oesophagus, artery, vein, soft tissue, and larynx. A 14-score scale (including all comprised structures) and a 12-score scale (excluding larynx) were developed. Receiver-operating characteristic (ROC) analysis was used to evaluate the performance of the scales. Stratified fivefold cross-validation and external verification were used to validate the scale.
RESULTS: In the training cohort, compromised RLN (p < 0.001), trachea (p = 0.001), oesophagus (p = 0.002), artery (p < 0.001), vein (p = 0.005), and soft tissue (p < 0.001) were predictors for unresectability, while compromised larynx (p = 0.283) was not. The 12-score scale (AUC = 0.882, 95%CI: 0.812-0.952) was not inferior to the 14-score scale (AUC = 0.891, 95%CI: 0.823-0.960). In subgroup analysis, the AUCs of the 12-score scale were 0.826 for treatment-naïve patients and 0.976 for patients with prior surgery. The 12-score scale was further validated with a fivefold cross-validation analysis, with an overall accuracy of 78.9-89.4%. Finally, external validation using the testing cohort showed an AUC of 0.875.
CONCLUSIONS: The researchers built a CT-based 12-score scale to evaluate the resectability of locally advanced thyroid cancer. Validation with a larger sample size is required to confirm the efficacy of the scale.
CONCLUSIONS: This 12-score CT scale would help clinicians evaluate the resectability of locally advanced thyroid cancer.
CONCLUSIONS: • The researchers built a 12-score CT scale (including recurrent laryngeal nerve, trachea, oesophagus, artery, vein, and soft tissue) to evaluate the resectability of locally advanced thyroid cancer. • This scale has the potential to help clinicians make treatment plans for locally advanced thyroid cancer.

UNASSIGNED: Maternal glycemia is associated with the risk of cesarean delivery (CD); therefore, our study aims to developed a prediction model based on glucose indicators in the second trimester to earlier identify the risk of CD.
UNASSIGNED: This was a nested case-control study, and data were collected from the 5th Central Hospital of Tianjin (training set) and Changzhou Second People\'s Hospital (testing set) from 2020 to 2021. Variables with significant difference in training set were incorporated to develop the random forest model. Model performance was assessed by calculating the area under the curve (AUC) and Komogorov-Smirnoff (KS), as well as accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
UNASSIGNED: A total of 504 eligible women were enrolled; of these, 169 underwent CD. Pre-pregnancy body mass index (BMI), first pregnancy, history of full-term birth, history of livebirth, 1 h plasma glucose (1hPG), glycosylated hemoglobin (HbA1c), fasting plasma glucose (FPG), and 2 h plasma glucose (2hPG) were used to develop the model. The model showed a good performance, with an AUC of 0.852 [95% confidence interval (CI): 0.809-0.895]. The pre-pregnancy BMI, 1hPG, 2hPG, HbA1c, and FPG were identifies as the more significant predictors. External validation confirmed the good performance of our model, with an AUC of 0.734 (95%CI: 0.664-0.804).
UNASSIGNED: Our model based on glucose indicators in the second trimester performed well to predict the risk of CD, which may reach the earlier identification of CD risk and may be beneficial to make interventions in time to decrease the risk of CD.

This study aims to establish a simple and non-invasive risk prediction model for hyperuricemia in Chinese adults based on modifiable risk factors. In 2020-2021, the baseline survey of the Beijing Health Management Cohort (BHMC) was conducted in Beijing city among the health examination population. Diverse life-style risk factors including dietary patterns and habits, cigarette smoking, alcohol intake, sleep duration and cell-phone use were collected. We developed hyperuricemia prediction models using three machine-learning techniques, namely logistic regression (LR), random forest (RF), and XGBoost. Performances in discrimination, calibration, and clinical applicability of the three methods were compared. Decision curve analysis (DCA) was used to assess the model\'s clinical usefulness. A total of 74 050 people were included in the study, of whom 55 537 (75%) were randomly selected into the training set and the other 18 513 (25%) were in the validation set. The prevalence of HUA was 38.43% in men and 13.29% in women. The XGBoost model has better performance than the LR and RF models. The area under the curve (AUC) (95% CI) in the training set for the LR, RF and XGBoost models were 0.754 (0.750-0.757), 0.844 (0.841-0.846) and 0.854 (0.851-0.856), respectively. The XGBoost model had a higher classification accuracy of 0.774 than the logistic (0.592) and RF (0.767) models. The AUC (95% CI) values in the validation set for the LR, RF and XGBoost models were 0.758 (0.749-0.765), 0.809 (0.802-0.816) and 0.820 (0.813-0.827), respectively. As demonstrated by the DCA curves, all the three models could bring net benefits within the appropriate threshold probability. XGBoost had better discrimination and accuracy. Various modifiable risk factors included in the model were helpful in facilitating the easy identification and life-style interventions of the HUA high-risk population.

To develop a machine learning model and nomogram to predict the probability of persistent virus shedding (PVS) in hospitalized patients with coronavirus disease 2019 (COVID-19), the clinical symptoms and signs, laboratory parameters, cytokines, and immune cell data of 429 patients with nonsevere COVID-19 were retrospectively reviewed. Two models were developed using the Akaike information criterion (AIC). The performance of these two models was analyzed and compared by the receiver operating characteristic (ROC) curve, calibration curve, net reclassification index (NRI), and integrated discrimination improvement (IDI). The final model included the following independent predictors of PVS: sex, C-reactive protein (CRP) level, interleukin-6 (IL-6) level, the neutrophil-lymphocyte ratio (NLR), monocyte count (MC), albumin (ALB) level, and serum potassium level. The model performed well in both the internal validation (corrected C-statistic = 0.748, corrected Brier score = 0.201) and external validation datasets (corrected C-statistic = 0.793, corrected Brier score = 0.190). The internal calibration was very good (corrected slope = 0.910). The model developed in this study showed high discriminant performance in predicting PVS in nonsevere COVID-19 patients. Because of the availability and accessibility of the model, the nomogram designed in this study could provide a useful prognostic tool for clinicians and medical decision-makers.

The risk factors of complications after SWL are not well characterized. Therefore, based on a large prospective cohort, we aimed to develop and validate a nomogram for predicting major complications after extracorporeal shockwave lithotripsy (SWL) in patients with ureteral stones. The development cohort included 1522 patients with ureteral stones who underwent SWL between June 2020 and August 2021 in our hospital. Five hundred and fifty-three patients with ureteral stones participated in the validation cohort from September 2020 to April 2022. The data were prospectively recorded. Backward stepwise selection was applied using the likelihood ratio test with Akaike\'s information criterion as the stopping rule. The efficacy of this predictive model was assessed concerning its clinical usefulness, calibration, and discrimination. Finally, 7.2% (110/1522) of patients in the development cohort and 8.7% (48/553) of those in the validation cohort suffered from major complications. We identified five predictive factors for major complications: age, gender, stone size, Hounsfield unit of stone, and hydronephrosis. This model showed good discrimination with an area under the receiver operating characteristic curves of 0.885 (0.872-0.940) and good calibration (P = 0.139). The decision curve analysis showed that the model was clinically valuable. In this large prospective cohort, we found that older age, female gender, higher Hounsfield unit, size, and grade of hydronephrosis were risk predictors of major complications after SWL. This nomogram will be helpful in preoperative risk stratification to provide individualized treatment recommendations for each patient. Furthermore, early identification and appropriate management of high-risk patients may decrease postoperative morbidity.

暂无摘要。

The purpose of this study was to assess and predict risk factors for death within 30 days after orthotopic liver transplant and to develop a nomogram to predict mortality after liver transplant.
We retrospectively studied 185 patients who underwent orthotopic livertransplant at Sichuan Provincial People\'s Hospital from January 1, 2010, to December 31, 2018. Multivariable logistic regression analyses were used to identify independent risk factors. A nomogram model was developed to predict mortality after liver transplant. The performance of the prediction model was assessed and validated by receiver operating characteristic curve and bootstrap methods (1000 replications).
Multivariable logistic regression analyses revealed that tracheal extubation time, postoperative infection, and intraperitoneal hemorrhage posttransplant were independentrisk factors for mortality after liver transplant. The receiver operating characteristic curve of the nomogram prediction model was 0.896 (96% CI, 0.803-0.989), and the mean absolute error of internal validation by bootstrap (1000 replications) was 0.019 (n = 184). These results showed that the nomogram model had an excellent prediction accuracy.
A nomogram model can provide clinicians with an individualized risk assessment of perioperative mortality in liver transplant recipients.