Histology-specific standardized incidence ratio improves the estimation of second primary lung cancer risk.

BACKGROUND: Lung cancer (LC) survivors are at increased risk for developing a second primary cancer (SPC) compared to the general population. While this risk is particularly high for smoking-related SPCs, the published standardized incidence ratio (SIR) for lung cancer after lung cancer is unexpectedly low in countries that follow international multiple primary (IARC/IACR MP) rules when compared to the USA, where distinct rules are employed. IARC/IACR rules rely on histology-dependent documentation of SPC with the same location as the first cancer and only classify an SPC when tumors present different histology. Thus, SIR might be underestimated in cancer registries using these rules. This study aims to assess whether using histology-specific reference rates for calculating SIR improves risk estimates for second primary lung cancer (SPLC) in LC survivors. METHODS: We (i) use the distribution of histologic subtypes of LC in population-based cancer registry data of 11 regional cancer registries from Germany to present evidence that the conventional SIR metric underestimates the actual risk for SPLC in LC survivors in registries that use IARC/IACR MP rules, (ii) present updated risk estimates for SPLC in Germany using a novel method to calculate histological subtype-specific SIRs, and (iii) validate this new method using US SEER (Surveillance, Epidemiology, and End Results Program) data, where different MP rules are applied. RESULTS: The adjusted relative risk for lung cancer survivors in Germany to develop an SPLC was 2.98 (95% CI 2.53-3.49) for females and 1.15 (95% CI 1.03-1.27) for males using the novel histology-specific SIR. When using IARC/IACR MP rules, the conventional SIR underestimates the actual risk for SPLC in LC survivors by approximately 30% for both sexes. CONCLUSIONS: Our proposed histology-specific method makes the SIR metric more robust against MP rules and, thus, more suitable for cross-country comparisons.

Explore the value of carotid ultrasound radiomics nomogram in predicting ischemic stroke risk in patients with type 2 diabetes mellitus.

Background and objective: Type 2 Diabetes Mellitus (T2DM) with insulin resistance (IR) is prone to damage the vascular endothelial, leading to the formation of vulnerable carotid plaques and increasing ischemic stroke (IS) risk. The purpose of this study is to develop a nomogram model based on carotid ultrasound radiomics for predicting IS risk in T2DM patients. Methods: 198 T2DM patients were enrolled and separated into study and control groups based on IS history. After manually delineating carotid plaque region of interest (ROI) from images, radiomics features were identified and selected using the least absolute shrinkage and selection operator (LASSO) regression to calculate the radiomics score (RS). A combinatorial logistic machine learning model and nomograms were created using RS and clinical features like the triglyceride-glucose index. The three models were assessed using area under curve (AUC) and decision curve analysis (DCA). Results: Patients were divided into the training set and the testing set by the ratio of 0.7. 4 radiomics features were selected. RS and clinical variables were all statically significant in the training set and were used to create a combination model and a prediction nomogram. The combination model (radiomics + clinical nomogram) had the largest AUC in both the training set and the testing set (0.898 and 0.857), and DCA analysis showed that it had a higher overall net benefit compared to the other models. Conclusions: This study created a carotid ultrasound radiomics machine-learning-based IS risk nomogram for T2DM patients with carotid plaques. Its diagnostic performance and clinical prediction capabilities enable accurate, convenient, and customized medical care.

Verbal Probability Terms for Communicating Clinical Risk - a Systematic Review.

Verbal probability expressions such as 'likely' and 'possible' are commonly used to communicate uncertainty in diagnosis, treatment effectiveness as well as the risk of adverse events. Probability terms that are interpreted consistently can be used to standardize risk communication. A systematic review was conducted. Research studies that evaluated numeric meanings of probability terms were reviewed. Terms with consistent numeric interpretation across studies were selected and were used to construct a Visual Risk Scale. Five probability terms showed reliable interpretation by laypersons and healthcare professionals in empirical studies. 'Very Likely' was interpreted as 90% chance (range 80 to 95%); 'Likely/Probable,' 70% (60 to 80%); 'Possible,' 40% (30 to 60%); 'Unlikely,' 20% (10 to 30%); and 'Very Unlikely' with 10% chance (5% to 15%). The corresponding frequency terms were: Very Frequently, Frequently, Often, Infrequently, and Rarely, respectively. Probability terms should be presented with their corresponding numeric ranges during discussions with patients. Numeric values should be presented as X-in-100 natural frequency statements, even for low values; and not as percentages, X-in-1000, X-in-Y, odds, fractions, 1-in-X, or as number needed to treat (NNT). A Visual Risk Scale was developed for use in clinical shared decision making.

Developing a risk framework for assembly construction based on stakeholder theory and structural equation modelling.

Occupational injuries in the construction industry have plagued many countries, and many cases have shown that accidents often occur because of a combination of project participants. Assembled construction (AC) projects have received extensive attention from Chinese scholars as a future trend, but few studies have explored the interrelationships and potential risks of various stakeholders in depth. This study fills this research gap by proposing a multi-stakeholder AC risk framework. The study surveyed 396 stakeholders, then analyzed the collected data and created a risk framework based on Structural Equation Modelling (SEM) and the CRITIC weighting method. The results revealed that factors like "regular supervision is a formality," "blindly approving the wrong safety measures," and "failure to organize effective safety education and training." are vital risks in AC of China. Finally, the study validates the risk factors and the framework with 180 real-life cases, which shows that the proposed framework is theoretically grounded and realistic. The study also suggests multi-level strategies such as introducing AI-based automated risk monitoring, improving the adaptability of normative provisions to technological advances, and advancing the culture of project communities of interest to ensure AC's safe practices.

External validation of a multi-biomarker-based score for predicting risk of cardiovascular disease in patients with rheumatoid arthritis.

BACKGROUND: A multi-biomarker disease activity (MBDA)-based cardiovascular disease (CVD) risk score was developed and internally validated in a Medicare cohort to predict 3-year risk for myocardial infarction (MI), stroke or CVD death in patients with rheumatoid arthritis (RA). It combines the MBDA score, leptin, MMP-3, TNF-R1, age and four clinical variables. We are now externally validating it in a younger RA cohort. METHODS: Claims data from a private aggregator were linked to MBDA test data to create a cohort of RA patients ≥18 years old. A univariable Cox proportional hazards regression model was fit using the MBDA-based CVD risk score as sole predictor of time-to-a-CVD event (hospitalized MI or stroke). Hazard ratio (HR) estimate was determined for all patients and for clinically relevant subgroups. A multivariable Cox model evaluated whether the MBDA-based CVD risk score adds predictive information to clinical data. RESULTS: 49,028 RA patients (340 CVD events) were studied. Mean age was 52.3 years; 18.3% were male. HR for predicting 3-year risk of a CVD event by the MBDA-based CVD risk score in the full cohort was 3.99 (95% CI: 3.51-4.49, p = 5.0×10-95). HR were also significant for subgroups based on age, comorbidities, disease activity, and drug use. In a multivariable model, the MBDA-based CVD risk score added significant information to hypertension, diabetes, tobacco use, history of CVD, age, sex and CRP (HR = 2.27, p = 1.7×10-7). CONCLUSION: The MBDA-based CVD risk score has been externally validated in an RA cohort that is younger than and independent of the Medicare cohort that was used for development and internal validation.

Assessing the effectiveness of marine nature-based solutions with climate risk assessments.

Prospective risks from climate change impacts in ocean and coastal systems are urging the implementation of nature-based solutions (NBS). These are climate-resilient strategies to maintain biodiversity and the delivery of ecosystem services, contributing to the adaptation of social-ecological systems and the mitigation of climate-related impacts. However, the effectiveness of measures like marine restoration or conservation is not exempt from the impacts of climate change, and the degree to which they can sustain biodiversity and ecosystem services remains unknown. Such uncertainty, together with the slow pace of implementation, causes decision-makers and societies to demand a better understanding of NBS effects. To address this gap, in this study, we use the risk mitigation capacity of marine NBS as a proxy for their effectiveness while providing a toolset for the implementation of the method. The method considers environmental data and relies on expert elicitation, allowing us to go beyond current practice to evaluate the effectiveness of NBS in reducing habitat or species risks under different future socio-political and climate-change scenarios. As a result, we present a ready-to-use tool, and supporting materials, for the implementation of the Climate Risk Assessment method and an illustrative example considering the application of the NBS "nature-inclusive harvesting" in two shellfisheries. The method works as a rapid assessment that guarantees comparability across sites and species due to its low data or resource demand, so it can be widely incorporated to adaptation policies across the marine realm.

The Predictive Value of the Inflammatory Prognostic Index for Detecting No-Reflow in ST-Elevation Myocardial Infarction Patients. / O Valor Preditivo do Índice Prognóstico Inflamatório para Detecção de No-Reflow em Pacientes com Infarto do Miocárdio com Supradesnivelamento do Segmento ST.

BACKGROUND: No-reflow (NR) is characterized by an acute reduction in coronary flow that is not accompanied by coronary spasm, thrombosis, or dissection. Inflammatory prognostic index (IPI) is a novel marker that was reported to have a prognostic role in cancer patients and is calculated by neutrophil/lymphocyte ratio (NLR) multiplied by C-reactive protein/albumin ratio. OBJECTIVE: We aimed to investigate the relationship between IPI and NR in ST-segment elevation myocardial infarction (STEMI) patients undergoing primary percutaneous coronary intervention (pPCI). METHODS: A total of 1541 patients were enrolled in this study (178 with NR and 1363 with reflow). Lasso panelized shrinkage was used for variable selection. A nomogram was created based on IPI for detecting the risk of NR development. Internal validation with Bootstrap resampling was used for model reproducibility. A two-sided p-value <0.05 was accepted as a significance level for statistical analyses. RESULTS: IPI was higher in patients with NR than in patients with reflow. IPI was non-linearly associated with NR. IPI had a higher discriminative ability than the systemic immune-inflammation index, NLR, and CRP/albumin ratio. Adding IPI to the baseline multivariable logistic regression model improved the discrimination and net-clinical benefit effect of the model for detecting NR patients, and IPI was the most prominent variable in the full model. A nomogram was created based on IPI to predict the risk of NR. Bootstrap internal validation of nomogram showed a good calibration and discrimination ability. CONCLUSION: This is the first study that shows the association of IPI with NR in STEMI patients who undergo pPCI.

FUNDAMENTO: O no-reflow (NR) é caracterizado por uma redução aguda no fluxo coronário que não é acompanhada por espasmo coronário, trombose ou dissecção. O índice prognóstico inflamatório (IPI) é um novo marcador que foi relatado como tendo um papel prognóstico em pacientes com câncer e é calculado pela razão neutrófilos/linfócitos (NLR) multiplicada pela razão proteína C reativa/albumina. OBJETIVO: Nosso objetivo foi investigar a relação entre IPI e NR em pacientes com infarto do miocárdio com supradesnivelamento do segmento ST (IAMCSST) submetidos a intervenção coronária percutânea primária (ICPp). MÉTODOS: Um total de 1.541 pacientes foram incluídos neste estudo (178 com NR e 1.363 com refluxo). A regressão penalizada LASSO (Least Absolute Shrinkage and Select Operator) foi usada para seleção de variáveis. Foi criado um nomograma baseado no IPI para detecção do risco de desenvolvimento de NR. A validação interna com reamostragem Bootstrap foi utilizada para reprodutibilidade do modelo. Um valor de p bilateral <0,05 foi aceito como nível de significância para análises estatísticas. RESULTADOS: O IPI foi maior em pacientes com NR do que em pacientes com refluxo. O IPI esteve associado de forma não linear com a NR. O IPI apresentou maior capacidade discriminativa do que o índice de imunoinflamação sistêmica, NLR e relação PCR/albumina. A adição do IPI ao modelo de regressão logística multivariável de base melhorou a discriminação e o efeito do benefício clínico líquido do modelo para detecção de pacientes com NR, e o IPI foi a variável mais proeminente no modelo completo. Foi criado um nomograma baseado no IPI para prever o risco de NR. A validação interna do nomograma Bootstrap mostrou uma boa capacidade de calibração e discriminação. CONCLUSÃO: Este é o primeiro estudo que mostra a associação de IPI com NR em pacientes com IAMCSST submetidos a ICPp.

Is having a moderate or low history, electrocardiogram, age, risk factors, troponin risk score a handicap for long-term mortality?

OBJECTIVE: History, electrocardiogram, age, risk factors, troponin risk score and troponin level follow-up are used to safely discharge low-risk patients with suspected non-ST elevation acute coronary syndrome from the emergency department for a 1-month period. We aimed to comprehensively investigate the 6-month mortality of patients with the history, electrocardiogram, age, risk factors, troponin risk score. METHODS: A total of 949 non-ST elevation acute coronary syndrome patients admitted to the emergency department from 01.01.2019 to 01.10.2019 were included in this retrospective study. History, electrocardiogram, age, risk factors, troponin scores of all patients were calculated by two emergency clinicians and a cardiologist. We compared the 6-month mortality of the groups. RESULTS: The mean age of the patients was 67.9 (56.4-79) years; 57.3% were male and 42.7% were female. Six-month mortality was significantly lower in the high-risk history, electrocardiogram, age, risk factors, troponin score group than in the low- and moderate-risk groups: 11/80 (12.1%), 58/206 (22%), and 150/444 (25.3%), respectively (p=0.019). CONCLUSION: Patients with high history, electrocardiogram, age, risk factors, troponin risk scores are generally treated with coronary angioplasty as soon as possible. We found that the mortality rate of this group of patients was lower in the long term compared with others. Efforts are also needed to reduce the mortality of moderate and low-risk patients. Further studies are needed on the factors affecting the 6-month mortality of moderate and low-risk acute coronary syndrome patients.

A proposed simplified definition of metabolic syndrome in children and adolescents: a global perspective.

Metabolic syndrome (MetS) is becoming prevalent in the pediatric population. The existing pediatric MetS definitions (e.g., the International Diabetes Federation (IDF) definition and the modified National Cholesterol Education Program (NCEP) definition) involve complex cut-offs, precluding fast risk assessment in clinical practice.We proposed a simplified definition for assessing MetS risk in youths aged 6-17 years, and compared its performance with two existing widely used pediatric definitions (the IDF definition, and the NCEP definition) in 10 pediatric populations from 9 countries globally (n = 19,426) using the receiver operating characteristic (ROC) curve analyses. In general, the total MetS prevalence of 6.2% based on the simplified definition was roughly halfway between that of 4.2% and 7.7% estimated from the IDF and NCEP definitions, respectively. The ROC curve analyses showed a good agreement between the simplified definition and two existing definitions: the total area under the curve (95% confidence interval) of the proposed simplified definition for identifying MetS risk achieved 0.91 (0.89-0.92) and 0.79 (0.78-0.81) when using the IDF or NCEP definition as the gold standard, respectively.The proposed simplified definition may be useful for pediatricians to quickly identify MetS risk and cardiometabolic risk factors (CMRFs) clustering in clinical practice, and allow direct comparison of pediatric MetS prevalence across different populations, facilitating consistent pediatric MetS risk monitoring and the development of evidence-based pediatric MetS prevention strategies globally.

A first trimester prediction model and nomogram for gestational diabetes mellitus based on maternal clinical risk factors in a resource-poor setting.

BACKGROUND: The implementation of universal screening for Gestational Diabetes Mellitus (GDM) is challenged by several factors key amongst which is limited resources, hence the continued reliance on risk factor-based screening. Effective identification of high-risk women early in pregnancy may enable preventive intervention. This study aimed at developing a GDM prediction model based on maternal clinical risk factors that are easily assessable in the first trimester of pregnancy in a population of Nigerian women. METHODS: This was a multi-hospital prospective observational cohort study of 253 consecutively selected pregnant women from which maternal clinical data was collected at 8-12 weeks gestational age. Diagnosis of GDM was made via a one-step 75-gram Oral Glucose Tolerance Test (OGTT) at 24-28 weeks of gestation. A GDM prediction model and nomogram based on selected maternal clinical risk factors was developed using multiple logistic regression analysis, and its performance was assessed by Receiver Operator Curve (ROC) analysis. Data analysis was carried out using Statistical Package for Social Sciences (SPSS) version 25 and Python programming language (version 3.0). RESULTS: Increasing maternal age, higher body mass index (BMI), a family history of diabetes mellitus in first-degree relative and previous history of foetal macrosomia were the major predictors of GDM. The model equation was: LogitP = 6.358 - 0.066 × Age - 0.075 × First trimester BMI - 1.879 × First-degree relative with diabetes mellitus - 0.522 × History of foetal macrosomia. It had an area under the receiver operator characteristic (ROC) curve (AUC) of 0.814 (95% CI: 0.751-0.877; p-value < 0.001), and at a predicted probability threshold of 0.745, it had a sensitivity of 79.2% and specificity of 74.5%. CONCLUSION: This first trimester prediction model reliably identifies women at high risk for GDM development in the first trimester, and the nomogram enhances its practical applicability, contributing to improved clinical outcomes in the study population.

Aortic geometry and long-term outcome in patients with a repaired coarctation.

OBJECTIVE: This study aims to compare aortic morphology between repaired coarctation patients and controls, and to identify aortic morphological risk factors for hypertension and cardiovascular events (CVEs) in coarctation patients. METHODS: Repaired coarctation patients with computed tomography angiography (CTA) or magnetic resonance angiography (MRA) were included, followed-up and compared with sex-matched and age-matched controls. Three-dimensional aortic shape was reconstructed using patients' CTA or MRA, or four-dimensional flow cardiovascular magnetic resonance in controls, and advanced geometrical characteristics were calculated and visualised using statistical shape modelling. In patients, we examined the association of geometrical characteristics with (1) baseline hypertension, using multivariable logistic regression; and (2) cardiovascular events (CVE, composite of aortic complications, coronary artery disease, ventricular arrhythmias, heart failure hospitalisation, stroke, transient ischaemic attacks and cardiovascular death), using multivariable Cox regression. The least absolute shrinkage and selection operator (LASSO) method selected the most informative multivariable model. RESULTS: Sixty-five repaired coarctation patients (23 years (IQR 19-38)) were included, of which 44 (68%) patients were hypertensive at baseline. After a median follow-up of 8.7 years (IQR 4.8-15.4), 27 CVEs occurred in 20 patients. Aortic arch dimensions were smaller in patients compared with controls (diameter p<0.001, wall surface area p=0.026, volume p=0.007). Patients had more aortic arch torsion (p<0.001) and a higher curvature (p<0.001). No geometrical characteristics were associated with hypertension. LASSO selected left ventricular mass, male sex, tortuosity and age for the multivariable model. Left ventricular mass (p=0.014) was independently associated with CVE, and aortic tortuosity showed a trend towards significance (p=0.070). CONCLUSION: Repaired coarctation patients have a smaller aortic arch and a more tortuous course of the aorta compared with controls. Besides left ventricular mass index, geometrical features might be of importance in long-term risk assessment in coarctation patients.

Deep learning for risk stratification of thymoma pathological subtypes based on preoperative CT images.

OBJECTIVES: This study aims to develop an innovative, deep model for thymoma risk stratification using preoperative CT images. Current algorithms predominantly focus on radiomic features or 2D deep features and require manual tumor segmentation by radiologists, limiting their practical applicability. METHODS: The deep model was trained and tested on a dataset comprising CT images from 147 patients (82 female; mean age, 54 years ± 10) who underwent surgical resection and received subsequent pathological confirmation. The eligible participants were divided into a training cohort (117 patients) and a testing cohort (30 patients) based on the CT scan time. The model consists of two stages: 3D tumor segmentation and risk stratification. The radiomic model and deep model (2D) were constructed for comparative analysis. Model performance was evaluated through dice coefficient, area under the curve (AUC), and accuracy. RESULTS: In both the training and testing cohorts, the deep model demonstrated better performance in differentiating thymoma risk, boasting AUCs of 0.998 and 0.893 respectively. This was compared to the radiomic model (AUCs of 0.773 and 0.769) and deep model (2D) (AUCs of 0.981 and 0.760). Notably, the deep model was capable of simultaneously identifying lesions, segmenting the region of interest (ROI), and differentiating the risk of thymoma on arterial phase CT images. Its diagnostic prowess outperformed that of the baseline model. CONCLUSIONS: The deep model has the potential to serve as an innovative decision-making tool, assisting on clinical prognosis evaluation and the discernment of suitable treatments for different thymoma pathological subtypes. KEY POINTS: • This study incorporated both tumor segmentation and risk stratification. • The deep model, using clinical and 3D deep features, effectively predicted thymoma risk. • The deep model improved AUCs by 16.1pt and 17.5pt compared to radiomic model and deep model (2D) respectively.

Towards proactive palliative care in oncology: developing an explainable EHR-based machine learning model for mortality risk prediction.

BACKGROUND: Ex-ante identification of the last year in life facilitates a proactive palliative approach. Machine learning models trained on electronic health records (EHR) demonstrate promising performance in cancer prognostication. However, gaps in literature include incomplete reporting of model performance, inadequate alignment of model formulation with implementation use-case, and insufficient explainability hindering trust and adoption in clinical settings. Hence, we aim to develop an explainable machine learning EHR-based model that prompts palliative care processes by predicting for 365-day mortality risk among patients with advanced cancer within an outpatient setting. METHODS: Our cohort consisted of 5,926 adults diagnosed with Stage 3 or 4 solid organ cancer between July 1, 2017, and June 30, 2020 and receiving ambulatory cancer care within a tertiary center. The classification problem was modelled using Extreme Gradient Boosting (XGBoost) and aligned to our envisioned use-case: "Given a prediction point that corresponds to an outpatient cancer encounter, predict for mortality within 365-days from prediction point, using EHR data up to 365-days prior." The model was trained with 75% of the dataset (n = 39,416 outpatient encounters) and validated on a 25% hold-out dataset (n = 13,122 outpatient encounters). To explain model outputs, we used Shapley Additive Explanations (SHAP) values. Clinical characteristics, laboratory tests and treatment data were used to train the model. Performance was evaluated using area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC), while model calibration was assessed using the Brier score. RESULTS: In total, 17,149 of the 52,538 prediction points (32.6%) had a mortality event within the 365-day prediction window. The model demonstrated an AUROC of 0.861 (95% CI 0.856-0.867) and AUPRC of 0.771. The Brier score was 0.147, indicating slight overestimations of mortality risk. Explanatory diagrams utilizing SHAP values allowed visualization of feature impacts on predictions at both the global and individual levels. CONCLUSION: Our machine learning model demonstrated good discrimination and precision-recall in predicting 365-day mortality risk among individuals with advanced cancer. It has the potential to provide personalized mortality predictions and facilitate earlier integration of palliative care.

A perspective on genetic and polygenic risk scores-advances and limitations and overview of associated tools.

Polygenetic Risk Scores are used to evaluate an individual's vulnerability to developing specific diseases or conditions based on their genetic composition, by taking into account numerous genetic variations. This article provides an overview of the concept of Polygenic Risk Scores (PRS). We elucidate the historical advancements of PRS, their advantages and shortcomings in comparison with other predictive methods, and discuss their conceptual limitations in light of the complexity of biological systems. Furthermore, we provide a survey of published tools for computing PRS and associated resources. The various tools and software packages are categorized based on their technical utility for users or prospective developers. Understanding the array of available tools and their limitations is crucial for accurately assessing and predicting disease risks, facilitating early interventions, and guiding personalized healthcare decisions. Additionally, we also identify potential new avenues for future bioinformatic analyzes and advancements related to PRS.

Dynamic risk assessment of a coal slurry preparation system based on the structure-variable Dynamic Bayesian Network.

In order to strengthen the safety management of coal slurry preparation systems, a dynamic risk assessment method was established by using the bow-tie (BT) model and the Structure-variable Dynamic Bayesian Network (SVDBN). First, the BT model was transformed into a static Bayesian network (BN) model of the failure of a coal slurry preparation system by using the bow-tie model and the structural similarity of the Bayesian cognitive science, based on the SVDBN recursive reasoning algorithm. The risk factors of the coal slurry preparation system were deduced using the Python language in two ways, and at the same time, preventive measures were put forward according to the weak links. In order to verify the accuracy and feasibility of this method, the simulation results were compared with those obtained using GeNIe software. The reasoning results of the two methods were very similar. Without considering maintenance factors, the failure rate of the coal slurry preparation system gradually increases with increasing time. When considering maintenance factors, the reliability of the coal slurry preparation system will gradually be maintained at a certain threshold, and the maintenance factors will increase the reliability of the system. The proposed method can provide a theoretical basis for the risk assessment and safety management of coal slurry preparation systems.

Stomach Cancer Prediction Model (SCoPM): An approach to risk stratification in a diverse U.S. population.

BACKGROUND AND AIMS: Population-based screening for gastric cancer (GC) in low prevalence nations is not recommended. The objective of this study was to develop a risk-prediction model to identify high-risk patients who could potentially benefit from targeted screening in a racial/ethnically diverse regional US population. METHODS: We performed a retrospective cohort study from Kaiser Permanente Southern California from January 2008-June 2018 among individuals age ≥50 years. Patients with prior GC or follow-up <30 days were excluded. Censoring occurred at GC, death, age 85 years, disenrollment, end of 5-year follow-up, or study conclusion. Cross-validated LASSO regression models were developed to identify the strongest of 20 candidate predictors (clinical, demographic, and laboratory parameters). Records from 12 of the medical service areas were used for training/initial validation while records from a separate medical service area were used for testing. RESULTS: 1,844,643 individuals formed the study cohort (1,555,392 training and validation, 289,251 testing). Mean age was 61.9 years with 53.3% female. GC incidence was 2.1 (95% CI 2.0-2.2) cases per 10,000 person-years (pyr). Higher incidence was seen with family history: 4.8/10,000 pyr, history of gastric ulcer: 5.3/10,000 pyr, H. pylori: 3.6/10,000 pyr and anemia: 5.3/10,000 pyr. The final model included age, gender, race/ethnicity, smoking, proton-pump inhibitor, family history of gastric cancer, history of gastric ulcer, H. pylori infection, and baseline hemoglobin. The means and standard deviations (SD) of c-index in validation and testing datasets were 0.75 (SD 0.03) and 0.76 (SD 0.02), respectively. CONCLUSIONS: This prediction model may serve as an aid for pre-endoscopic assessment of GC risk for identification of a high-risk population that could benefit from targeted screening.

Early warning of systemic risk in stock market based on EEMD-LSTM.

With the increasing importance of the stock market, it is of great practical significance to accurately describe the systemic risk of the stock market and conduct more accurate early warning research on it. However, the existing research on the systemic risk of the stock market lacks multi-dimensional factors, and there is still room for improvement in the forecasting model. Therefore, to further measure the systemic risk profile of the Chinese stock market, establish a risk early warning system suitable for the Chinese stock market, and improve the risk management awareness of investors and regulators. This paper proposes a combination model of EEMD-LSTM, which can describe the complex nonlinear interaction. Firstly, 35 stock market systemic risk indicators are selected from the perspectives of macroeconomic operation, market cross-contagion and the stock market itself to build a comprehensive indicator system that conforms to the reality of China. Furthermore, based on TEI@I complex system methodology, an EEMD-LSTM model is proposed. The EEMD method is adopted to decompose the composite index sequence into intrinsic mode function components (IMF) of different scales and one trend term. Then the LSTM algorithm is used to predicted and model the decomposed sub-sequences. Finally, the forecast result of the composite index is obtained through integration. The empirical results show that the stock market systemic risk index constructed in this paper can effectively identify important risk events within the sample period. In addition, compared with the benchmark model, the EEMD-LSTM model constructed in this paper shows a stronger early warning ability for systemic financial risks in the stock market.

Machine learning to predict the occurrence of thyroid nodules: towards a quantitative approach for judicious utilization of thyroid ultrasonography.

Introduction: Ultrasound is instrumental in the early detection of thyroid nodules, which is crucial for appropriate management and favorable outcomes. However, there is a lack of clinical guidelines for the judicious use of thyroid ultrasonography in routine screening. Machine learning (ML) has been increasingly used on big data to predict clinical outcomes. This study aims to leverage the ML approach in assessing the risk of thyroid nodules based on common clinical features. Methods: Data were sourced from a Chinese cohort undergoing routine physical examinations including thyroid ultrasonography between 2013 and 2023. Models were established to predict the 3-year risk of thyroid nodules based on patients' baseline characteristics and laboratory tests. Four ML algorithms, including logistic regression, random forest, extreme gradient boosting, and light gradient boosting machine, were trained and tested using fivefold cross-validation. The importance of each feature was measured by the permutation score. A nomogram was established to facilitate risk assessment in the clinical settings. Results: The final dataset comprised 4,386 eligible subjects. Thyroid nodules were detected in 54.8% (n=2,404) individuals within the 3-year observation period. All ML models significantly outperformed the baseline regression model, successfully predicting the occurrence of thyroid nodules in approximately two-thirds of individuals. Age, high-density lipoprotein, fasting blood glucose and creatinine levels exhibited the highest impact on the outcome in these models. The nomogram showed consistency and validity, providing greater net benefits for clinical decision-making than other strategies. Conclusion: This study demonstrates the viability of an ML-based approach in predicting the occurrence of thyroid nodules. The findings highlight the potential of ML models in identifying high-risk individuals for personalized screening, thereby guiding the judicious use of ultrasound in this context.