Predicting ischemic stroke patients’ prognosis changes using machine learning in a nationwide stroke registry

Objective: Stroke is a major cause of morbidity and mortality worldwide resulting from either an ischemic insult or rupture of a blood vessel in the brain. It not only leads to significant physical disability but also imposes emotional and economic burdens on patients and their families. Assessing the lipid profile in stroke patients is crucial, especially in ischemic stroke, where dyslipidemia plays a key role in atherosclerosis development. Elevated low-density lipoprotein (LDL) and triglycerides (TG) are important modifiable risk factors, and managing these can reduce the incidence and recurrence of strokes. Comprehensive lipid management should be a core component of stroke prevention and treatment strategies. Methods: This was a cross-sectional study in which 80 adult stroke patients (40 ischemic and 40 hemorrhagic strokes) were included on the basis of a predefined inclusion and exclusion criteria. The diagnosis of stroke was made on the basis of imaging and neurological examination. A comprehensive medical history, physical measurements (including body mass index), and lipid profiles (total cholesterol [TC], LDL, High-density lipoprotein [HDL], and TG) were recorded for all participants. Blood samples were taken after overnight fasting to ensure accuracy. Statistical analysis using SPSS version 21.0 compared lipid levels between the groups. An unpaired t-test and Chi-square test were used, with significance defined as a p<0.05. Results: There was a significant difference in lipid profiles between ischemic and hemorrhagic stroke patients. Ischemic stroke patients had markedly higher levels of LDL with a mean of 129.84±36.54 mg/dL compared to 106.32±26.68 mg/dL in hemorrhagic stroke patients (p=0.0015). TC levels were also significantly elevated in ischemic stroke patients, averaging 236.22±56.26 mg/dL versus 196.48±46.24 mg/dL in hemorrhagic stroke patients (p=0.0009). Total TG were higher in ischemic stroke patients (158.54±44.68 mg/dL) compared to hemorrhagic stroke patients (128.62±39.16 mg/dL, p=0.0021). HDL levels were slightly lower in ischemic stroke patients (34.54±8.26 mg/dL) compared to hemorrhagic stroke patients (38.12±9.12 mg/dL), although this difference was not statistically significant (p=0.0696). Conclusion: There were significant differences in lipid profiles between ischemic and hemorrhagic stroke patients. Ischemic stroke patients had higher levels of LDL, TC, and TG, indicating a stronger association with dyslipidemia and atherosclerosis. These findings highlight the importance of aggressive lipid management in ischemic stroke patients to reduce recurrence risk 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.

BackgroundIn ischemic stroke patients with active cancer, cryptogenic stroke has worse prognosis than stroke by conventional mechanisms. However, the individual effects of intracranial atherosclerosis (ICAS) or atrial fibrillation (AF) on the prognosis of these patients have not been studied.AimsTherefore, we aimed to investigate the effects of ICAS and AF on the prognosis of ischemic stroke patients with active cancer.MethodsWe included ischemic stroke patients with active cancer between 2010 and 2020. Early neurological deterioration (END) was defined as an increase of ≥ 1 in the motor NIHSS score, or ≥ 2 in the total NIHSS score within 72 hours of admission. Unfavorable outcomes were defined as a score of ≥ 3 on the 3-month modified Rankin Scale.ResultsIn total, 116 ischemic stroke patients with active cancer were evaluated. In multivariable analysis, ICAS was positively associated with END (adjusted odds ratio [aOR] = 4.56, 95% confidence interval [CI]: 1.52–13.70), and this association showed a quantitative relationship according to the degree of stenosis of ICAS (stenosis group: aOR = 4.24, 95% CI: 1.31–13.72; occlusion group, aOR = 5.74, 95% CI: 1.05–31.30). ICAS was also closely related to unfavorable outcomes (aOR = 6.33, 95% CI: 1.15–34.79). In contrast, AF showed no significant association with END or unfavorable outcomes. Our data showed that patients with ICAS had larger and more severe initial stroke lesions, and poorer prognosis than those without.ConclusionsICAS, but not AF, was closely associated with poor prognosis in ischemic stroke patients with active cancer.

Background:Systemic lupus erythematosus (SLE) is a chronic autoimmune disease that affects multiple organs and systems. However, in addition to physical damage, patients with SLE have an increased risk of mental...

Background: The left atrial septal pouch (LASP) is an anatomic variant of the interatrial septum, and its pathophysiological significance is uncertain (J Am Coll Cardiol Intv 2010;3:98-104). We examined the potential association between LASP and ischemic stroke subtypes in patients undergoing transesophageal echocardiography (TEE). Methods: In this cross-sectional study, we determined the prevalence of LASP ( Figure ) using a database of patients who underwent TEE at our institution between July 2008 and June 2011. Patients with ischemic strokes were evaluated by a vascular neurologist (blinded to the presence of LASP) for stroke subtyping using standard and modified TOAST (excluding ischemic stroke patients with multiple competing etiologies from the cryptogenic category) criteria. We compared the prevalence of LASP in ischemic stroke, cryptogenic stroke, and non-stroke patients. Results: The mean age of all 212 patients studied (including stroke and non-stroke patients) was 57 years. The overall prevalence of LASP was 17% (n=35). We studied 75 ischemic stroke patients. There were no significant differences in demographics or stroke risk factors comparing patients with or without LASP, except for hypertension (LASP 49% vs. no LASP 66% among all patients, p = 0.05). The prevalence of LASP among cryptogenic stroke (using standard and modified TOAST criteria) was increased compared to the prevalence among other ischemic stroke patients (26% vs. 9%, RR=1.8, 95% CI 1.1-3.1; and 30% vs. 9%, RR=2.3, 95% CI 1.2-4.1, respectively). There was no significant difference between LASP prevalence among all ischemic stroke patients (16%) and patients with no history of stroke or TIA (14%). Conclusion: In this population of relatively young patients, prevalence of LASP was increased in cryptogenic stroke compared to stroke patients of other subtypes. These findings suggest LASP may be a risk factor for stroke, which should be verified by further large-scale studies.

Regionalization, systems of care design, and quality improvement (QI) registry participation all promote the widespread dissemination of guideline-based evidence into actual practice. As a result, policy statements from the American Heart Association/American Stroke Association (AHA/ASA) advocate for the creation of regional systems of care for various time-critical diagnoses, including ST-elevation myocardial infarction (STEMI), out-of-hospital cardiac arrest resuscitation, and acute stroke.1–3 Creation of these regional networks requires multidisciplinary collaboration to implement 5 mutually reinforcing core elements4 that build each system: (1) Designation of certain hospitals with special treatment capabilities as Receiving Centers for STEMI, resuscitation, or stroke; (2) emergency medical services (EMS) destination protocols that allow for direct transport of certain patients identified by explicit triage criteria to a designated Receiving Center, thus allowing for bypass of closer hospitals if they lack the needed specialty service; (3) organized interhospital transfer and transport protocols to a Receiving Center for appropriate patients who initially self-present or are mistriaged to a Referral Hospital; (4) communication or telemedicine options to provide real-time expert consultation as needed from a Receiving Center to its associated Referral Hospitals or EMS providers; and (5) participation in a regional and/or national QI registry to track relevant process-of-care metrics and meaningful risk-adjusted clinical outcomes. Within each of the 50 states, unique challenges exist for stakeholders attempting to implement the 5 aforementioned core elements of regional or statewide systems of care. In particular, substantial variation exists with regard to the starting point for these initiatives. For example, some states already have sufficient regulatory authority within their EMS agency or state department of health (DOH) to regionalize care of time-critical diagnoses, whereas other states require new legislation to create coordinated systems. In October 2011, the Advocacy Coordinating Committee of the AHA convened a multispecialty task force to assess the …

Compared with coronary artery bypass grafting (CABG) under cardiopulmonary bypass, off-pump coronary artery bypass (OPCAB) is minimally invasive and reduces the risk of intraoperative blood transfusion and acute kidney injury. Nonetheless, OPCAB-related complications still pose a threat. Machine learning technology can analyze a large number of clinical data, establish risk prediction models and help clinicians make early and correct clinical decisions. Risk prediction models are available for mortality and morbidity after cardiac surgery, but they are not specific to OPCAB. This study aimed to develop a predictive model of severe complications after OPCAB, based on machine learning. Anesthesia records of OPCAB from the General Hospital of the Northern Theater Command (Shenyang, China) collected between January 1, 2019, and June 15, 2020, were analyzed. The endpoint of the study was the occurrence of serious complications after OPCAB (postoperative unplanned intra-aortic balloon pump, secondary surgery and death). The features entered into the models were as follows: intraoperative ventricular fibrillation, number of saphenous vein grafts, nerve block (NeB), venous oxygen saturation (SvO2), skin incision-bypass time, and hypertension. A total of 8 machine learning algorithms were tested: logistic regression analysis (LRA), k-nearest neighbor (KNN), naïve Bayes (NB), support vector machine (SVM), random forest (RF), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), and categorical features gradient boosting (CatBoost). Among the 506 patients found in the records, 27 met the endpoint. The highest area under the curve (AUC) value was achieved with the XGBoost model (AUC = 0.94), and the lowest with the SVM model (AUC = 0.75). The highest and lowest accuracy were observed with the XGBoost and NB models, respectively, while the highest and lowest precision were achieved using the SVM and NB models, respectively. Based on the receiver operating characteristic (ROC) curves, the XGBoost model was selected as the most useful in this study. This study suggests using the XGBoost model to predict the risk of complications after OPCAB.

ABSTRACTObjective: Microalbuminuria could be detected in patients with acute stroke. However, the association between microalbuminuria and the severity of ischemic stroke has not been systematically investigated. This study aimed to systematically explore the prevalence of microalbuminuria in ischemic stroke patients, and the association of microalbuminuria with the severity of ischemic stroke, as well as the prognostic value of microalbuminuria in cerebral infarction patients.Methods: 160 ischemic stroke patients and 54 controls were enrolled and clinical characteristics were recorded. Severity of stroke was assessed by NIHSS score at admission, and outcome was measured using mRS score. The concentration of urinary microalbumin was collected for each participant. Multiple linear regression analysis was performed to evaluate the relationship between microalbuminuria and the severity of ischemic stroke, and logistic regression analysis was employed to identify the prognostic value of microalbuminuria in ischemic stroke patients.Results: The incidence of microalbuminuria in ischemic stroke patients was 36.88%. The concentration of urinary microalbumin increased with the increasing of cerebral infarction size, and was independently correlated with NIHSS score at admission and mRS score at 3 months after onset. In multivariate logistic regression analyses, microalbuminuria was one of the independent risk factors for poor prognosis of cerebral infarction patients.Conclusions: MAU was found in approximately one-third of patients with acute ischemic stroke. It was correlated with the severity of cerebral infarction at admission and clinical outcomes at 3 months after onset and could be used as a potential indicator of poor prognosis in ischemic stroke patients.

ObjectivesA radiomics-based explainable eXtreme Gradient Boosting (XGBoost) model was developed to predict central cervical lymph node metastasis (CCLNM) in patients with papillary thyroid carcinoma (PTC), including positive and negative effects.MethodsA total of 587 PTC patients admitted at Binzhou Medical University Hospital from 2017 to 2021 were analyzed retrospectively. The patients were randomized into the training and test cohorts with an 8:2 ratio. Radiomics features were extracted from ultrasound images of the primary PTC lesions. The minimum redundancy maximum relevance algorithm and the least absolute shrinkage and selection operator regression were used to select CCLNM positively-related features and radiomics scores were constructed. Clinical features, ultrasound features, and radiomics score were screened out by the Boruta algorithm, and the XGBoost model was constructed from these characteristics. SHapley Additive exPlanations (SHAP) was used for individualized and visualized interpretation. SHAP addressed the cognitive opacity of machine learning models.ResultsEleven radiomics features were used to calculate the radiomics score. Five critical elements were used to build the XGBoost model: capsular invasion, radiomics score, diameter, age, and calcification. The area under the curve was 91.53% and 90.88% in the training and test cohorts, respectively. SHAP plots showed the influence of each parameter on the XGBoost model, including positive (i.e., capsular invasion, radiomics score, diameter, and calcification) and negative (i.e., age) impacts. The XGBoost model outperformed the radiologist, increasing the AUC by 44%.ConclusionsThe radiomics-based XGBoost model predicted CCLNM in PTC patients. Visual interpretation using SHAP made the model an effective tool for preoperative guidance of clinical procedures, including positive and negative impacts.

Category:Ankle Arthritis; AnkleIntroduction/Purpose:Ankle arthrodesis and total ankle replacement are the most commonly performed procedures for surgical management of ankle arthritis. Arthrodesis provides effective pain relief but the rate of complications after arthrodesis is higher as it is more commonly performed in patients with comorbidities that preclude ankle replacement. Accurately risk- stratifying patients who undergo ankle arthrodesis would be of great utility, given the significant cost and morbidity associated with developing major perioperative complications. There is a paucity of accurate prediction models that can be used to pre- operatively risk-stratify patients for ankle arthrodesis. We aim to develop a machine learning (ML) algorithm for prediction of major perioperative complication after ankle arthrodesis as well as compare its performance against traditional predictive models based on logistic regression.Methods:This is a retrospective cohort study of adult patients who underwent ankle arthrodesis at any non-federal California hospital between 2015 and 2017. The primary outcome was readmission within 30 days or major perioperative complication - venous thromboembolism within 30 days, myocardial infarction within 7 days, pneumonia within 7 days, systemic infection within 7 days, surgical site bleeding within 90 days, and wound complications within 90 days. We build ML and logistic regression models that span different classes of modeling approaches: XGBoost, AdaBoost, Gradient Boosting, and Random Forest. Discrimination and calibration were assessed using area under the receiver operating characteristic curve (AUROC) and Brier score, respectively. We utilize a partial dependence function to measure the importance of an individual feature by assessing the average effect in predicted risks when its value is altered. We rank the contribution of the included variables to the prediction of adverse outcomes.Results:A total of 1,084 patients met inclusion criteria for this study. There were 131 major complications or readmission (12.1%). The optimized XGBoost algorithm demonstrates higher discrimination (AUROC: 0.707 + 0.052) compared to LR (0.691 + 0.055). The receiver operating characteristic curves for the XGBoost and logistic regression models are visualized in Figure 1. XGBoost also outperforms the three other ML models. This model was well calibrated (Brier score: 0.103 + 0.001). The variables most important for the XGBoost model include diabetes, chronic kidney disease, implant complication, and major fracture. Five of the ten most important features for XGBoost were markedly less important for the traditional logistic regression model: male sex, prior hip fracture, cardiorespiratory failure, acute renal failure, and dialysis status.Conclusion:We report a ML algorithm for prediction of major perioperative complications after ankle arthrodesis. The optimized XGBoost model is well-calibrated and demonstrates superior risk prediction to logistic regression. This tool may identify and address potentially modifiable risk factors, helping to accurately risk-stratify patients and decrease likelihood of major complications. Notably, the predictors most important for XGBoost are different from those for logistic regression. This suggests that the superior discriminative capability of ML methods stems from their ability to capture complex non-linear relationships between variables that logistic regression is unable to detect.

The purpose of this study was to cluster long-term multifaceted functional recovery patterns and to establish prediction models for functional outcome in first-time stroke patients using unsupervised machine learning. This study is an interim analysis of the dataset from the Korean Stroke Cohort for Functioning and Rehabilitation (KOSCO), a long-term, prospective, multicenter cohort study of first-time stroke patients. The KOSCO screened 10,636 first-time stroke patients admitted to nine representative hospitals in Korea during a three-year recruitment period, and 7,858 patients agreed to enroll. Early clinical and demographic features of stroke patients and six multifaceted functional assessment scores measured from 7 days to 24 months after stroke onset were used as input variables. K-means clustering analysis was performed, and prediction models were generated and validated using machine learning. A total of 5,534 stroke patients (4,388 ischemic and 1,146 hemorrhagic; mean age 63·31 ± 12·86; 3,253 [58.78%] male) completed functional assessments 24 months after stroke onset. Through K-means clustering, ischemic stroke (IS) patients were clustered into five groups and hemorrhagic stroke (HS) patients into four groups. Each cluster had distinct clinical characteristics and functional recovery patterns. The final prediction models for IS and HS patients achieved relatively high prediction accuracies of 0.926 and 0.887, respectively. The longitudinal, multi-dimensional, functional assessment data of first-time stroke patients were successfully clustered, and the prediction models showed relatively good accuracies. Early identification and prediction of long-term functional outcomes will help clinicians develop customized treatment strategies.

This study develops predictive models for Chinese female patients with VL utilizing machine learning techniques. The aim is to create an effective model that can assist in clinical diagnosis and treatment of vaginal relaxation, thereby enhancing women’s pelvic floor health. In total, 1184 women with VL have been randomly selected and categorized into groups using the finger measurement method. Among them, there are 383 cases of mild VL, 405 cases of moderate VL, and 396 cases of severe VL. Concurrently, 396 healthy women without VL who underwent routine health examinations have been chosen at random and assigned to the non-VL group. Based on 1580 cases, we have established LightGBM, Random Forest, XGBoost, and AdaBoost models based on training dataset using 5-fold cross-validation and GridSearch, and analyzed the performance of the models on the hold-out test dataset. The confusion matrix, precision, recall, F1 score, overall accuracy, and ROC curve of the models on the hold-out test dataset are compared. The overall accuracy of LightGBM model, RF model, XGBoost model, and AdaBoost model are 0.8987, 0.8987, 0.8987, and 0.8457, respectively. The average AUC of LightGBM model is 0.976, the one of RF model is 0.9763, the one of XGBoost model is 0.9775, and the one of AdaBoost model is 0.928. The XGBoost model has the more comprehensive and reasonable performance among the four prediction models, which can accurately distinguish between healthy, mild VL, as well as moderate VL and severe VL, which can assist doctors in diagnosing persons’ conditions more accurately, devising personalized treatment plans, avoiding unnecessary surgeries, reducing persons’ psychological stress, improving patient compliance and treatment outcomes, thus enhancing overall treatment results.

The antioxidant properties of Ainsliaea acerifolia, a wild edible plant, were examined by ultrasonic-assisted ethanol extraction methods. The primary objective was to optimize the extraction conditions and accurately predict antioxidant activities using advanced machine learning models. The extraction conditions were optimized using Response Surface Methodology (RSM). Various parameters, including temperature, extraction time, and ethanol concentration, were adjusted to maximize antioxidant activity. The optimal conditions identified were a temperature of 68 °C, an extraction time of 86 min, and an ethanol concentration of 57%. Under these conditions, the extracts exhibited the highest antioxidant activity. To enhance the predictive accuracy of antioxidant activity, an XGBoost (XGB) model was employed. The XGB model performance was evaluated and compared with the RSM model. The XGB model achieved an R² value of 94.71%, significantly outperforming the RSM model by 12.8%. This highlights the superiority of the XGB model in predicting antioxidant activities based on the given extraction parameters. Additionally, the study developed a graphical user interface (GUI). This GUI allows researchers and industry experts to input extraction conditions and obtain quick, accurate predictions of antioxidant activity.

A long-standing problem at the frontier of biomechanical studies is to develop fast methods capable of estimating material properties from clinical data. In this paper, we have studied three surrogate models based on machine learning (ML) methods for fast parameter estimation of left ventricular (LV) myocardium. We use three ML methods named K-nearest neighbour (KNN), XGBoost and multi-layer perceptron (MLP) to emulate the relationships between pressure and volume strains during the diastolic filling. Firstly, to train the surrogate models, a forward finite-element simulator of LV diastolic filling is used. Then the training data are projected in a low-dimensional parametrized space. Next, three ML models are trained to learn the relationships of pressure–volume and pressure–strain. Finally, an inverse parameter estimation problem is formulated by using those trained surrogate models. Our results show that the three ML models can learn the relationships of pressure–volume and pressure–strain very well, and the parameter inference using the surrogate models can be carried out in minutes. Estimated parameters from both the XGBoost and MLP models have much less uncertainties compared with the KNN model. Our results further suggest that the XGBoost model is better for predicting the LV diastolic dynamics and estimating passive parameters than other two surrogate models. Further studies are warranted to investigate how XGBoost can be used for emulating cardiac pump function in a multi-physics and multi-scale framework.

Development and validation of interpretable machine learning model for pre-treatment predicting the response to targeted and immune therapy in hepatocellular carcinoma.