Prognostic nomograms for elderly patients with small cell lung cancer brain metastasis: a SEER population-based study with temporal external validation.

OBJECTIVE: This study aimed to pinpoint independent predictors influencing overall survival (OS) and cancer-specific survival (CSS) in elderly patients with small cell lung cancer (SCLC) brain metastasis (BM), and to create and validate nomograms for OS and CSS prediction. METHODS: Data from elderly SCLC BM patients were extracted out of the SEER database, including 1200 patients identified from 2010 and 2015 who were randomly allocated into a training set and an internal validation set at a proportion of 7:3, and 666 patients diagnosed between 2018 and 2020 as a temporal external validation set. Independent predictors for OS and CSS were determined through univariate Cox analysis, least absolute shrinkage and selection operator (LASSO) analysis, and multivariate Cox analysis sequentially. Nomograms for OS and CSS were constructed, and validated by the internal and temporal external validation sets. RESULTS: Age, N stage, chemotherapy, and liver metastasis were determined as independent predictors of OS and CSS, while radiotherapy and surgery were not. Nomograms were constructed based on these independent predictors. The results of the receiver operator characteristic (ROC) curves, the areas under the curve (AUC) and calibration curve demonstrated that the nomograms exhibited commendable discriminative ability and calibration. Moreover, decision curve analysis (DCA), net reclassification improvement (NRI), and integrated discrimination improvement (IDI) also suggested that the nomograms possessed superior clinical usefulness and predictive capability relative to the TNM system. CONCLUSIONS: Prognostic nomograms for elderly patients with SCLC BM have been developed, demonstrating good performance in terms of accuracy, reliability, and practicality.

Endoscopic Resection of Malignancies in the Upper GI Tract: A Clinical Algorithm.

Background: Malignancies in the upper gastrointestinal tract are amenable to endoscopic resection at an early stage. Achieving a curative resection is the most stringent quality criterion, but post-resection risk assessment and aftercare are also part of a comprehensive quality program. Summary: Various factors influence the achievement of curative resection. These include endoscopic assessment prior to resection using chromoendoscopy and HD technology. If resectability is possible, it is particularly important to delineate the lateral resection margins as precisely as possible before resection. Furthermore, the correct choice of resection technique depending on the lesion must be taken into account. Endoscopic submucosal dissection is the standard for esophageal squamous cell carcinoma and gastric carcinoma. In Western countries, it is becoming increasingly popular to treat Barrett's neoplasia over 2 cm in size and/or with suspected submucosal infiltration with en bloc resection instead of piece meal resection. After resection, risk assessment based on the histopathological resection determines the patient's individual risk of lymph node metastases, particularly in the case of high-risk lesions. This is categorized according to the current literature. Key Messages: This review presents clinical algorithms for endoscopic resection of esophageal SCC, Barrett's neoplasia, and gastric neoplasia. The algorithms include the pre-resection assessment of the lesion and the resection margins, the adequate resection technique for the respective lesion, as well as the post-resection risk assessment with an evidence-based recommendation for follow-up therapy and surveillance.

Comparison of multi-task ergonomic assessment methods for risk of upper extremity and low back musculoskeletal disorders.

Work-related musculoskeletal disorder of upper extremity multi-task assessment methods (Revised Strain Index [RSI], Distal Upper Extremity Tool [DUET]) and manual handling multi-task assessment methods (Revised NIOSH Lifting Equation [RNLE], Lifting Fatigue Failure Tool [LiFFT]) were compared. RSI and DUET showed a strong correlation (rs = 0.933, p < 0.001) where increasing risk factor exposure resulted in increasing outputs for both methods. RSI and DUET demonstrated fair agreement (κ = 0.299) in how the two methods classified outputs into risk categories (high, moderate or low) when assessing the same tasks. The RNLE and LiFFT showed a strong correlation (rs = 0.903, p = 0.001) where increasing risk factor exposure resulted in increasing outputs, and moderate agreement (κ = 0.574) in classifying the outputs into risk categories (high, moderate or low) when assessing the same tasks. The multi-task assessment methods provide consistent output magnitude rankings in terms of increasing exposure, however some differences exist between how different methods classify the outputs into risk categories.

A Comparison of Globally Applied Prognostic Risk Groups and the Prevalence of Metastatic Disease on Prostate-specific Membrane Antigen Positron Emission Tomography in Patients with Newly Diagnosed Prostate Cancer.

BACKGROUND: Various risk classification systems (RCSs) are used globally to stratify newly diagnosed patients with prostate cancer (PCa) into prognostic groups. OBJECTIVE: To compare the predictive value of different prognostic subgroups (low-, intermediate-, and high-risk disease) within the RCSs for detecting metastatic disease on prostate-specific membrane antigen (PSMA) positron emission tomography (PET)/computed tomography (CT) for primary staging, and to assess whether further subdivision of subgroups would be beneficial. DESIGN, SETTING, AND PARTICIPANTS: Patients with newly diagnosed PCa, in whom PSMA-PET/CT was performed between 2017 and 2022, were studied retrospectively. Patients were stratified into risk groups based on four RCSs: European Association of Urology, National Comprehensive Cancer Network (NCCN), Cambridge Prognostic Group (CPG), and Cancer of the Prostate Risk Assessment. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: The prevalence of metastatic disease on PSMA-PET/CT was compared among the subgroups within the four RCSs. RESULTS AND LIMITATIONS: In total, 2630 men with newly diagnosed PCa were studied. Any metastatic disease was observed in 35% (931/2630) of patients. Among patients classified as having intermediate- and high-risk disease, the prevalence of metastases ranged from approximately 12% to 46%. Two RCSs further subdivided these groups. According to the NCCN, metastatic disease was observed in 5.8%, 13%, 22%, and 62% for favorable intermediate-, unfavorable intermediate-, high-, and very-high-risk PCa, respectively. Regarding the CPG, these values were 6.9%, 13%, 21%, and 60% for the corresponding risk groups. CONCLUSIONS: This study underlines the importance of nuanced risk stratification, recommending the further subdivision of intermediate- and high-risk disease given the notable variation in the prevalence of metastatic disease. PSMA-PET/CT for primary staging should be reserved for patients with unfavorable intermediate- or higher-risk disease. PATIENT SUMMARY: The use of various risk classification systems in patients with prostate cancer helps identify those at a higher risk of having metastatic disease on prostate-specific membrane antigen positron emission tomography/computed tomography for primary staging.

Reducing delivery insurance costs through risk score model for food delivery company.

In this paper, we propose a novel pricing model for delivery insurance in a food delivery company in Latin America, with the aim of reducing the high costs associated with the premium paid to the insurer. To achieve this goal, a thorough analysis was conducted to estimate the probability of losses based on delivery routes, transportation modes, and delivery drivers' profiles. A large amount of data was collected and used as a database, and various statistical models and machine learning techniques were employed to construct a comprehensive risk profile and perform risk classification. Based on the risk classification and the estimated probability associated with it, a new pricing model for delivery insurance was developed using advanced mathematical algorithms and machine learning techniques. This new pricing model took into account the pattern of loss occurrence and high and low-risk behaviors, resulting in a significant reduction of insurance costs for both the contracting company and the insurer. The proposed pricing model also allowed for greater flexibility in insurance contracting, making it more accessible and appealing to delivery drivers. The use of estimated loss probabilities and a risk score for the pricing of delivery insurance proved to be a highly effective and efficient alternative for reducing the high costs associated with insurance, while also improving the profitability and competitiveness of the food delivery company in Latin America.

Deep Learning-Based Segmentation and Risk Stratification for Gastrointestinal Stromal Tumors in Transabdominal Ultrasound Imaging.

PURPOSE: To develop a deep neural network system for the automatic segmentation and risk stratification prediction of gastrointestinal stromal tumors (GISTs). METHODS: A total of 980 ultrasound (US) images from 245 GIST patients were retrospectively collected. These images were randomly divided (6:2:2) into a training set, a validation set, and an internal test set. Additionally, 188 US images from 47 prospective GIST patients were collected to evaluate the segmentation and diagnostic performance of the model. Five deep learning-based segmentation networks, namely, UNet, FCN, DeepLabV3+, Swin Transformer, and SegNeXt, were employed, along with the ResNet 18 classification network, to select the most suitable network combination. The performance of the segmentation models was evaluated using metrics such as the intersection over union (IoU), Dice similarity coefficient (DSC), recall, and precision. The classification performance was assessed based on accuracy and the area under the receiver operating characteristic curve (AUROC). RESULTS: Among the compared models, SegNeXt-ResNet18 exhibited the best segmentation and classification performance. On the internal test set, the proposed model achieved IoU, DSC, precision, and recall values of 82.1, 90.2, 91.7, and 88.8%, respectively. The accuracy and AUC for GIST risk prediction were 87.4 and 92.0%, respectively. On the external test set, the segmentation models exhibited IoU, DSC, precision, and recall values of 81.0, 89.5, 92.8, and 86.4%, respectively. The accuracy and AUC for GIST risk prediction were 86.7 and 92.5%, respectively. CONCLUSION: This two-stage SegNeXt-ResNet18 model achieves automatic segmentation and risk stratification prediction for GISTs and demonstrates excellent segmentation and classification performance.

Graph Node Classification to Predict Autism Risk in Genes.

This study explores the genetic risk associations with autism spectrum disorder (ASD) using graph neural networks (GNNs), leveraging the Sfari dataset and protein interaction network (PIN) data. We built a gene network with genes as nodes, chromosome band location as node features, and gene interactions as edges. Graph models were employed to classify the autism risk associated with newly introduced genes (test set). Three classification tasks were undertaken to test the ability of our models: binary risk association, multi-class risk association, and syndromic gene association. We tested graph convolutional networks, Graph Sage, graph transformer, and Multi-Layer Perceptron (Baseline) architectures on this problem. The Graph Sage model consistently outperformed the other models, showcasing its utility in classifying ASD-related genes. Our ablation studies show that the chromosome band location and protein interactions contain useful information for this problem. The models achieved 85.80% accuracy on the binary risk classification, 81.68% accuracy on the multi-class risk classification, and 90.22% on the syndromic classification.

Hypertensive disorders of pregnancy among women with cardiovascular disease in Norway: A historical cohort study.

INTRODUCTION: Women with cardiovascular disease may be at increased risk of hypertensive disorders of pregnancy (HDP). We aimed to: (1) Investigate the occurrence of HDP in a cohort of pregnant women with cardiovascular disease and compare it with the occurrence in the general population. (2) Assess the association between maternal cardiovascular risk and risk of HDP. MATERIAL AND METHODS: We reviewed clinical data on a cohort of 901 pregnancies among 708 women with cardiovascular disease who were followed at the National Unit for Pregnancy and Heart Disease and gave birth at Oslo University Hospital between 2003 and 2018. The exposure under study was maternal cardiovascular risk, classified as low, moderate, or high based on a modified classification by the World Health Organization. The main outcome of interest was HDP, which included pre-eclampsia and gestational hypertension. The proportion of HDP cases in the general population in the same period was extracted from the Medical Birth Registry of Norway. We used logistic regression to estimate crude and adjusted odds ratios (OR) of HDP, with associated 95% confidence intervals (CIs), for women with moderate- and high cardiovascular risk compared to women with low risk. RESULTS: The occurrence of HDP in the study cohort was 12.1% (95% CI: 10.0%-14.4%) and varied between 8.7% (95% CI: 6.5%-11.3%) in the low-risk group, 15.7% (95% CI: 11.1%-21.4%) in the moderate-risk group, and 22.2% (95% CI: 15.1%-30.8%) in the high-risk group. By contrast, the nationwide occurrence of HDP was 5.1% (95% CI: 5.1%-5.2%). In the study cohort, the proportions of pregnancies with gestational hypertension and pre-eclampsia were similar (6.3% and 5.8%, respectively). Compared to pregnancies with low cardiovascular risk, the adjusted OR of HDP was 2.04 (95% CI: 1.21-3.44) in the moderate-risk group and 2.99 (95% CI: 1.73-5.18) in the high-risk group. CONCLUSIONS: The occurrence of hypertensive disease of pregnancy in the study cohort was more than doubled compared to the general population in Norway. The risk of HDP increased with maternal cardiovascular risk group. We recommend taking into account maternal cardiovascular risk group when assessing risk and prophylaxis of HDP.

Prior knowledge-guided multilevel graph neural network for tumor risk prediction and interpretation via multi-omics data integration.

The interrelation and complementary nature of multi-omics data can provide valuable insights into the intricate molecular mechanisms underlying diseases. However, challenges such as limited sample size, high data dimensionality and differences in omics modalities pose significant obstacles to fully harnessing the potential of these data. The prior knowledge such as gene regulatory network and pathway information harbors useful gene-gene interaction and gene functional module information. To effectively integrate multi-omics data and make full use of the prior knowledge, here, we propose a Multilevel-graph neural network (GNN): a hierarchically designed deep learning algorithm that sequentially leverages multi-omics data, gene regulatory networks and pathway information to extract features and enhance accuracy in predicting survival risk. Our method achieved better accuracy compared with existing methods. Furthermore, key factors nonlinearly associated with the tumor pathogenesis are prioritized by employing two interpretation algorithms (i.e. GNN-Explainer and IGscore) for neural networks, at gene and pathway level, respectively. The top genes and pathways exhibit strong associations with disease in survival analyses, many of which such as SEC61G and CYP27B1 are previously reported in the literature.

Adição da Troponina Ultrassensível à Avaliação de Risco Perioperatório Melhora a Capacidade Preditiva de Morte em Pacientes Submetidos à Cirurgia Não Cardíaca / Addition of High-Sensitivity Troponin to Perioperative Risk Assessment Improves the Predictive Ability of Death in Non-Cardiac Surgery Patients

Resumo Fundamento A estratificação ode risco é uma importante etapa na avaliação perioperatória. No entanto, os principais escores de risco não incorporam biomarcadores em seus conjuntos de variáveis. Objetivo Avaliar o poder incremental da troponina à estratificação de risco tradicional. Métodos Um total de 2230 pacientes admitidos na unidade de terapia intensiva após cirurgia não cardíaca foram classificados de acordo com três tipos de risco: Risco Cardiovascular (RCV), Índice de Risco Cardíaco Revisado (IRCR), e Risco Inerente da Cirurgia (RIC). O principal desfecho foi mortalidade por todas as causas. A regressão de Cox foi usada, assim como a estatística C antes e após a adição de troponina ultrassensível (pelo menos uma medida até três dias após a cirurgia). Finalmente, o índice de reclassificação líquida e a melhoria de discriminação integrada foram usadas para avaliar o poder incremental da troponina para a estratificação de risco. O nível de significância usado foi de 0,05. Resultados A idade média dos pacientes foi 63,8 anos e 55,6% eram do sexo feminino. A prevalência de lesão miocárdica após cirurgia não cardíaca (MINS) foi 9,4%. Pacientes com um RCV elevado apresentaram uma maior ocorrência de MINS (40,1% x 24,8%, p<0,001), bem como pacientes com alto RIC (21,3 x 13,9%, p=0,004) e aqueles com IRCR≥3 (3,0 x 0,7%, p=0,009). Pacientes sem MINS, independentemente do risco avaliado, apresentaram taxa de mortalidade similar. A adição de troponina à avaliação de risco melhorou a capacidade preditiva de mortalidade em 30 dias e de mortalidade em um ano em todas as avaliações de risco. Conclusão A prevalência de MINS é mais alta na população de alto risco. No entanto, sua prevalência na população de risco mais baixo não é desprezível e causa um maior risco de morte. A adição da troponina ultrassensível melhorou a capacidade preditiva da avaliação de risco em todos os grupos.

Abstract Background Risk stratification is an important step in perioperative evaluation. However, the main risk scores do not incorporate biomarkers in their set of variables. Objective Evaluate the incremental power of troponin to the usual risk stratification Methods A total of 2,230 patients admitted to the intensive care unit after non-cardiac surgery were classified according to three types of risk: cardiovascular risk (CVR), Revised Cardiac Risk Index (RCRI); and inherent risk of surgery (IRS). The main outcome was all-cause mortality. Cox regression was used as well as c-statistics before and after addition of high-sensitivity troponin (at least one measurement up to three days after surgery). Finally, net reclassification index and integrated discrimination improvement were used to assess the incremental power of troponin for risk stratification. Significance level was set at 0.05. Results Mean age of patients was 63.8 years and 55.6% were women. The prevalence of myocardial injury after non-cardiac surgery (MINS) was 9.4%. High CVR-patients had a higher occurrence of MINS (40.1 x 24.8%, p<0.001), as well as high IRS-patients (21.3 x 13.9%, p=0.004) and those with a RCRI≥3 (3.0 x 0.7%, p=0.009). Patients without MINS, regardless of the assessed risk, had similar mortality rate. The addition of troponin to the risk assessment improved the predictive ability of death at 30 days and at 1 year in all risk assessments. Conclusion The prevalence of MINS is higher in the high-risk population. However, its prevalence in lower-risk population is not negligible and causes a higher risk of death. The addition of high-sensitivity troponin increased the predictive ability of risk assessment in all groups.

Drug-drug interactions with oral anticoagulants: information consistency assessment of three commonly used online drug interactions databases in Switzerland.

Background: Toxicity or treatment failure related to drug-drug interactions (DDIs) are known to significantly affect morbidity and hospitalization rates. Despite the availability of numerous databases for DDIs identification and management, their information often differs. Oral anticoagulants are deemed at risk of DDIs and a leading cause of adverse drug events, most of which being preventable. Although many databases include DDIs involving anticoagulants, none are specialized in them. Aim and method: This study aims to compare the DDIs information content of four direct oral anticoagulants and two vitamin K antagonists in three major DDI databases used in Switzerland: Lexi-Interact, Pharmavista, and MediQ. It evaluates the consistency of DDIs information in terms of differences in severity rating systems, mechanism of interaction, extraction and documentation processes and transparency. Results: This study revealed 2'496 DDIs for the six anticoagulants, with discrepant risk classifications. Only 13.2% of DDIs were common to all three databases. Overall concordance in risk classification (high, moderate, and low risk) was slight (Fleiss' kappa = 0.131), while high-risk DDIs demonstrated a fair agreement (Fleiss' kappa = 0.398). The nature and the mechanism of the DDIs were more consistent across databases. Qualitative assessments highlighted differences in the documentation process and transparency, and similarities for availability of risk classification and references. Discussion: This study highlights the discrepancies between three commonly used DDI databases and the inconsistency in how terminology is standardised and incorporated when classifying these DDIs. It also highlights the need for the creation of specialised tools for anticoagulant-related interactions.

Whole-lesion iodine map histogram analysis in the risk classification of gastrointestinal stromal tumors: comparison with single-slice iodine concentration measurements.

PURPOSE: To evaluate and compare the diagnostic performances of whole-lesion iodine map (IM) histogram analysis and single-slice IM measurement in the risk classification of gastrointestinal stromal tumors (GISTs). METHODS: Thirty-seven patients with GISTs, including 19 with low malignant underlying GISTs (LG-GISTs) and 18 with high malignant underlying GISTs (HG-GISTs), were evaluated with dual-energy computed tomography (DECT). Whole-lesion IM histogram parameters (mean; median; minimum; maximum; standard deviation; variance; 1st, 10th, 25th, 50th, 75th, 90th, and 99th percentile; kurtosis, skewness, and entropy) were computed for each lesion. In other sessions, iodine concentrations (ICs) were derived from the IM by placing regions of interest (ROIs) on the tumor slices and normalizing them to the iodine concentration in the aorta. Both quantitative analyses were performed on the venous phase images. The diagnostic accuracies of the two methods were assessed and compared. RESULTS: The minimum, maximum, 1st, 10th, and 25th percentile of the whole-lesion IM histogram and the IC and normalized IC (NIC) of the single-slice IC measurement significantly differed between LG- and HG-GISTs (p < 0.001 - p = 0.042). The minimum value in the histogram analysis (AUC = 0.844) and the NIC in the single-slice measurement analysis (AUC = 0.886) showed the best diagnostic performances. The NIC of single-slice measurements had a diagnostic performance similar to that of the whole-lesion IM histogram analysis (p = 0.618). CONCLUSIONS: Both whole-lesion IM histogram analysis and single-slice IC measurement can differentiate LG-GISTs and HG-GISTs with similar diagnostic performances.

A decision support system to recommend appropriate therapy protocol for AML patients.

Introduction: Acute Myeloid Leukemia (AML) is one of the most aggressive hematological neoplasms, emphasizing the critical need for early detection and strategic treatment planning. The association between prompt intervention and enhanced patient survival rates underscores the pivotal role of therapy decisions. To determine the treatment protocol, specialists heavily rely on prognostic predictions that consider the response to treatment and clinical outcomes. The existing risk classification system categorizes patients into favorable, intermediate, and adverse groups, forming the basis for personalized therapeutic choices. However, accurately assessing the intermediate-risk group poses significant challenges, potentially resulting in treatment delays and deterioration of patient conditions. Methods: This study introduces a decision support system leveraging cutting-edge machine learning techniques to address these issues. The system automatically recommends tailored oncology therapy protocols based on outcome predictions. Results: The proposed approach achieved a high performance close to 0.9 in F1-Score and AUC. The model generated with gene expression data exhibited superior performance. Discussion: Our system can effectively support specialists in making well-informed decisions regarding the most suitable and safe therapy for individual patients. The proposed decision support system has the potential to not only streamline treatment initiation but also contribute to prolonged survival and improved quality of life for individuals diagnosed with AML. This marks a significant stride toward optimizing therapeutic interventions and patient outcomes.

Risk classification system using the detailed positive surgical margin status for predicting biochemical recurrence after robot-assisted radical prostatectomy.

AIM: This study aimed to evaluate the risk classification system using the detailed positive surgical margin (PSM) status to predict biochemical recurrence (BCR) after robot-assisted radical prostatectomy (RARP). METHODS: We retrospectively analyzed 427 patients who underwent RARP between January 2016 and March 2020. We investigated risk factors for BCR using univariate and multivariate Cox proportional hazard regression models. The biochemical recurrence-free survival (BRFS) rate was assessed using the Kaplan-Meier method. RESULTS: The median follow-up period was 43.4 months and 99 patients developed BCR. In the multivariate analysis, maximum PSM length > 5.0 mm and the International Society of Urological Pathology grade group (ISUP GG) at the PSM ≥3 were predictive factors for BCR in patients with a PSM. In the multivariate analysis, these factors were also independent predictive factors in the overall study population, including patients without a PSM. We classified the patients into four groups using these factors and found that the 1-year BRFS rates in the negative surgical margin (NSM) group, low-risk group (PSM and neither factor), intermediate-risk group (either factor), and high-risk group (both factors) were 94.9%, 94.5%, 83.1%, and 52.9%, respectively. The low-risk group showed similar BRFS to the NSM group (p = 0.985), while the high-risk group had significantly worse BRFS than the other groups (p < 0.001). CONCLUSION: Maximum PSM length > 5.0 mm and ISUP GG at the PSM ≥3 were independent predictive factors for BCR after RARP. Risk classification for BCR using these factors is considered to be useful and might help urologists decide on additional treatment after RARP.

The Clear Value of Coronary Artery Calcification Evaluation on Non-Gated Chest Computed Tomography for Cardiac Risk Stratification.

To enhance risk stratification in patients suspected of coronary artery disease, the assessment of coronary artery calcium (CAC) could be incorporated, especially when CAC can be readily assessed on previously performed non-gated chest computed tomography (CT). Guidelines recommend reporting on patients' extent of CAC on these non-cardiac directed exams and various studies have shown the diagnostic and prognostic value. However, this method is still little applied, and no current consensus exists in clinical practice. This review aims to point out the clinical utility of different kinds of CAC assessment on non-gated CTs. It demonstrates that these scans indeed represent a merely untapped and underestimated resource for risk stratification in patients with stable chest pain or an increased risk of cardiovascular events. To our knowledge, this is the first review to describe the clinical utility of different kinds of visual CAC evaluation on non-gated unenhanced chest CT. Various methods of CAC assessment on non-gated CT are discussed and compared in terms of diagnostic and prognostic value. Furthermore, the application of these non-gated CT scans in the general practice of cardiology is discussed. The clinical utility of coronary calcium assessed on non-gated chest CT, according to the current literature, is evident. This resource of information for cardiac risk stratification needs no specific requirements for scan protocol, and is radiation-free and cost-free. However, some gaps in research remain. In conclusion, the integration of CAC on non-gated chest CT in general cardiology should be promoted and research on this method should be encouraged.

The American Thyroid Association risk classification of papillary thyroid cancer according to presurgery cytology.

OBJECTIVE: To compare the American Thyroid Association (ATA) risk staging of histologically proven papillary thyroid cancer (PTC) in patients who received a presurgery cytologic result of either indeterminate thyroid nodules (ITNs, Bethesda III/IV) or suspicious for malignancy/malignant (TIR 4/5, Bethesda V/VI). METHODS: Clinical, ultrasonographic, cytological data from patients with histologically diagnosed PTC were retrospectively collected. RESULTS: Patients were stratified according to the preoperative fine-needle aspiration cytology into 2 groups: 51 ITNs (TIR3A/3B) and 118 suspicious/malignant (TIR 4/5). Male/female ratio, age, and presurgery TSH level were similar between the 2 groups. At ultrasound, TIR 4/5 nodules were significantly more frequently hypoechoic (P = .037), with irregular margins (P = .041), and with microcalcifications (P = .020) and were more frequently classified as high-risk according to the European Thyroid Imaging and Reporting Data System (EU-TIRADS; P = .021). At histology, the follicular PTC subtype was significantly more prevalent among ITNs while classical PTC subtype was more frequent in TIR 4/5 group (P = .002). In TIR 4/5 group, a higher rate of focal vascular invasion (P < .001) and neck lymph node metastasis (P = .028) was observed. Intermediate-risk category according to ATA was significantly more frequent in TIR 4/5 group while low-risk category was more frequently found among ITNs (P = .021), with a higher number of patients receiving radioiodine in TIR 4/5 group (P = .002). At multivariate logistic regression, having a TIR 4/5 cytology was associated with a significant risk of having a higher ATA risk classification as compared to ITN (OR 4.6 [95% CI 1.523-14.007], P = .007), independently from presurgery findings (nodule size at ultrasound, sex, age, and EU-TIRADS score). CONCLUSIONS: Papillary thyroid cancers recorded among ITNs are likely less aggressive and are generally assessed as at lower risk according to ATA classification.

CT-Based Radiomics Analysis of Different Machine Learning Models for Discriminating the Risk Stratification of Pheochromocytoma and Paraganglioma: A Multicenter Study.

RATIONALE AND OBJECTIVES: Using different machine learning models CT-based radiomics to integrate clinical radiological features to discriminating the risk stratification of pheochromocytoma/paragangliomas (PPGLs). MATERIALS AND METHODS: The present study included 201 patients with PPGLs from three hospitals (training set: n = 125; external validation set: n = 45; external test set: n = 31). Patients were divided into low-risk and high-risk groups using a staging system for adrenal pheochromocytoma and paraganglioma (GAPP). We extracted and selected CT radiomics features, and built radiomics models using support vector machines (SVM), k-nearest neighbors, random forests, and multilayer perceptrons. Using receiver operating characteristic curve analysis to select the optimal radiomics model, a combined model was built using the output of the optimal radiomics model and clinical radiological features, and its accuracy and clinical applicability were evaluated using calibration curves and clinical decision curve analysis (DCA). RESULTS: Finally, 13 radiomics features were selected to construct machine learning models. In the radiomics model, the SVM model demonstrated higher accuracy and stability, with an AUC value of 0.915 in the training set, 0.846 in external validation set, and 0.857 in external test set. Combining the outputs of SVM models with two clinical radiological features, a combined model constructed has demonstrated optimal risk stratification ability for PPGLs with an AUC of 0.926 for the training set, 0.883 for the external validation set, and 0.899 for the external test set. The calibration curve and DCA show good calibration accuracy and clinical effectiveness for the combined model. CONCLUSION: Combined model that integrates radiomics and clinical radiological features can discriminate the risk stratification of PPGLs.

External validation of the rCAST for patients after in-hospital cardiac arrest: a multicenter retrospective observational study.

No established predictive or risk classification tool exists for the neurological outcomes of post-cardiac arrest syndrome (PCAS) in patients with in-hospital cardiac arrest (IHCA). This study aimed to investigate whether the revised post-cardiac arrest syndrome for therapeutic hypothermia score (rCAST), which was developed to estimate the prognosis of PCAS patients with out-of-hospital cardiac arrest (OHCA), was applicable to patients with IHCA. A retrospective, multicenter observational study of 140 consecutive adult IHCA patients admitted to three intensive care units. The area under the receiver operating characteristic curves (AUCs) of the rCAST for poor neurological outcome and mortality at 30 days were 0.88 (0.82-0.93) and 0.83 (0.76-0.89), respectively. The sensitivity and specificity of the risk classification according to rCAST for poor neurological outcomes were 0.90 (0.83-0.96) and 0.67 (0.55-0.79) for the low, 0.63 (0.54-0.74) and 0.67 (0.55-0.79) for the moderate, and 0.27 (0.17-0.37) and 1.00 (1.00-1.00) for the high-severity grades. All 22 patients classified with a high-severity grade showed poor neurological outcomes. The rCAST showed excellent predictive accuracy for neurological prognosis in patients with PCAS after IHCA. The rCAST may be useful as a risk classification tool for PCAS after IHCA.

MRI-based prostate cancer classification using 3D efficient capsule network.

BACKGROUND: Prostate cancer (PCa) is the most common cancer in men and the second leading cause of male cancer-related death. Gleason score (GS) is the primary driver of PCa risk-stratification and medical decision-making, but can only be assessed at present via biopsy under anesthesia. Magnetic resonance imaging (MRI) is a promising non-invasive method to further characterize PCa, providing additional anatomical and functional information. Meanwhile, the diagnostic power of MRI is limited by qualitative or, at best, semi-quantitative interpretation criteria, leading to inter-reader variability. PURPOSES: Computer-aided diagnosis employing quantitative MRI analysis has yielded promising results in non-invasive prediction of GS. However, convolutional neural networks (CNNs) do not implicitly impose a frame of reference to the objects. Thus, CNNs do not encode the positional information properly, limiting method robustness against simple image variations such as flipping, scaling, or rotation. Capsule network (CapsNet) has been proposed to address this limitation and achieves promising results in this domain. In this study, we develop a 3D Efficient CapsNet to stratify GS-derived PCa risk using T2-weighted (T2W) MRI images. METHODS: In our method, we used 3D CNN modules to extract spatial features and primary capsule layers to encode vector features. We then propose to integrate fully-connected capsule layers (FC Caps) to create a deeper hierarchy for PCa grading prediction. FC Caps comprises a secondary capsule layer which routes active primary capsules and a final capsule layer which outputs PCa risk. To account for data imbalance, we propose a novel dynamic weighted margin loss. We evaluate our method on a public PCa T2W MRI dataset from the Cancer Imaging Archive containing data from 976 patients. RESULTS: Two groups of experiments were performed: (1) we first identified high-risk disease by classifying low + medium risk versus high risk; (2) we then stratified disease in one-versus-one fashion: low versus high risk, medium versus high risk, and low versus medium risk. Five-fold cross validation was performed. Our model achieved an area under receiver operating characteristic curve (AUC) of 0.83 and 0.64 F1-score for low versus high grade, 0.79 AUC and 0.75 F1-score for low + medium versus high grade, 0.75 AUC and 0.69 F1-score for medium versus high grade and 0.59 AUC and 0.57 F1-score for low versus medium grade. Our method outperformed state-of-the-art radiomics-based classification and deep learning methods with the highest metrics for each experiment. Our divide-and-conquer strategy achieved weighted Cohen's Kappa score of 0.41, suggesting moderate agreement with ground truth PCa risks. CONCLUSIONS: In this study, we proposed a novel 3D Efficient CapsNet for PCa risk stratification and demonstrated its feasibility. This developed tool provided a non-invasive approach to assess PCa risk from T2W MR images, which might have potential to personalize the treatment of PCa and reduce the number of unnecessary biopsies.