Bootstrap Resampling Procedure Research Articles

The diagnostic and triage accuracy of the GPT-3 artificial intelligence model: an observational study

Artificial intelligence (AI) applications in health care have been effective in many areas of medicine, but they are often trained for a single task using labelled data, making deployment and generalisability challenging. How well a general-purpose AI language model performs diagnosis and triage relative to physicians and laypeople is not well understood. We compared the predictive accuracy of Generative Pre-trained Transformer 3 (GPT-3)'s diagnostic and triage ability for 48 validated synthetic case vignettes (<50 words; sixth-grade reading level or below) of both common (eg, viral illness) and severe (eg, heart attack) conditions to a nationally representative sample of 5000 lay people from the USA who could use the internet to find the correct options and 21 practising physicians at Harvard Medical School. There were 12 vignettes for each of four triage categories: emergent, within one day, within 1 week, and self-care. The correct diagnosis and triage category (ie, ground truth) for each vignette was determined by two general internists at Harvard Medical School. For each vignette, human respondents and GPT-3 were prompted to list diagnoses in order of likelihood, and the vignette was marked as correct if the ground-truth diagnosis was in the top three of the listed diagnoses. For triage accuracy, we examined whether the human respondents' and GPT-3's selected triage was exactly correct according to the four triage categories, or matched a dichotomised triage variable (emergent or within 1 day vs within 1 week or self-care). We estimated GPT-3's diagnostic and triage confidence on a given vignette using a modified bootstrap resampling procedure, and examined how well calibrated GPT-3's confidence was by computing calibration curves and Brier scores. We also performed subgroup analysis by case acuity, and an error analysis for triage advice to characterise how its advice might affect patients using this tool to decide if they should seek medical care immediately. Among all cases, GPT-3 replied with the correct diagnosis in its top three for 88% (42/48, 95% CI 75-94) of cases, compared with 54% (2700/5000, 53-55) for lay individuals (p<0.0001) and 96% (637/666, 94-97) for physicians (p=0·012). GPT-3 triaged 70% correct (34/48, 57-82) versus 74% (3706/5000, 73-75; p=0.60) for lay individuals and 91% (608/666, 89-93%; p<0.0001) for physicians. As measured by the Brier score, GPT-3 confidence in its top prediction was reasonably well calibrated for diagnosis (Brier score=0·18) and triage (Brier score=0·22). We observed an inverse relationship between case acuity and GPT-3 accuracy (p<0·0001) with a fitted trend line of -8·33% decrease in accuracy for every level of increase in case acuity. For triage error analysis, GPT-3 deprioritised truly emergent cases in seven instances. A general-purpose AI language model without any content-specific training could perform diagnosis at levels close to, but below, physicians and better than lay individuals. We found that GPT-3's performance was inferior to physicians for triage, sometimes by a large margin, and its performance was closer to that of lay individuals. Although the diagnostic performance of GPT-3 was comparable to physicians, it was significantly better than a typical person using a search engine. The National Heart, Lung, and Blood Institute.

Using a monotonic density ratio model to increase the power of the goodness-of-fit test for logistic regression models with case-control data.

Logistic regression models are widely used in case-control data analysis, and testing the goodness-of-fit of their parametric model assumption is a fundamental research problem. In this article, we propose to enhance the power of the goodness-of-fit test by exploiting a monotonic density ratio model, in which the ratio of case and control densities is assumed to be a monotone function. We show that such a monotonic density ratio model is naturally induced by the retrospective case-control sampling design under the alternative hypothesis. The pool-adjacent-violator algorithm is adapted to solve for the constrained nonparametric maximum likelihood estimator under the alternative hypothesis. By measuring the discrepancy between this estimator and the semiparametric maximum likelihood estimator under the null hypothesis, we develop a new Kolmogorov-Smirnov-type statistic to test the goodness-of-fit for logistic regression models with case-control data. A bootstrap resampling procedure is suggested to approximate the -value of the proposed test. Simulation results show that the type I error of the proposed test is well controlled and the power improvement is substantial in many cases. Three real data applications are also included for illustration.

How variable are the classic ERP effects during sentence processing? A systematic resampling analysis of the N400 and P600 effects

Although event-related potential (ERP) research on language processing has capitalized on key, theoretically influential components such as the N400 and P600, their measurement properties—especially the variability in their temporal and spatial parameters—have rarely been examined. The current study examined the measurement properties of the N400 and P600 effects elicited by semantic and syntactic anomalies, respectively, during sentence processing. We used a bootstrap resampling procedure to randomly draw many thousands of resamples varying in sample size and stimulus count from a larger sample of 187 participants and 40 stimulus sentences of each type per condition. Our resampling investigation focused on three issues: (a) statistical power; (b) variability in the magnitudes of the effects; and (c) variability in the temporal and spatial profiles of the effects. At the level of grand averages, the N400 and P600 effects were both robust and substantial. However, across resamples, there was a high degree of variability in effect magnitudes, onset times, and scalp distributions, which may be greater than is currently appreciated in the literature, especially for the P600 effects. These results provide a useful basis for designing future studies using these two well-established ERP components. At the same time, the results also highlight challenges that need to be addressed in future research (e.g., how best to analyze the ERP data without engaging in such questionable research practices as p-hacking).

A nomogram prediction model for mild cognitive impairment in non-dialysis outpatient patients with chronic kidney disease

Background The high prevalence of mild cognitive impairment (MCI) in non-dialysis individuals with chronic kidney disease (CKD) impacts their prognosis and quality of life. Objective This study aims to investigate the variables associated with MCI in non-dialysis outpatient patients with CKD and to construct and verify a nomogram prediction model. Methods 416 participants selected from two hospitals in Chengdu, between January 2023 and June 2023. They were categorized into two groups: the MCI group (n = 210) and the non-MCI (n = 206). Univariate and multivariate binary logistic regression analyses were employed to identify independent influences (candidate predictor variables). Subsequently, regression models was constructed, and a nomogram was drawn. The restricted cubic spline diagram was drawn to further analyze the relationship between the continuous numerical variables and MCI. Internally validated using a bootstrap resampling procedure. Results Among 416 patients, 210 (50.9%) had MCI. Logistic regression analysis revealed that age, educational level, occupational status, use of smartphones, sleep disorder, and hemoglobin were independent influencing factors of MCI (all p<.05). The model’s area under the curve was 0.926,95% CI (0.902, 0.951), which was a good discriminatory measure; the Calibration curve, the Hosmer–Lemeshow test, and the Clinical Decision Curve suggested that the model had good calibration and clinical benefit. Internal validation results showed the consistency index was 0.926, 95%CI (0.925, 0.927). Conclusion The nomogram prediction model demonstrates good performance and can be used for early screening and prediction of MCI in non-dialysis patients with CKD. It provides valuable reference for medical staff to formulate corresponding intervention strategies.

The mediating role of coping styles in the relationship between fear of COVID-19 and mental health problems: a cross-sectional study among nurses

Fear of being infected by coronavirus disease 2019 (COVID-19) could trigger mental health problems among nurses at the frontline. In such a situation, coping strategies are needed to deal with the imminent threat. The purpose of this study was to test the mediating effects of coping on relationships of fear of COVID-19 with anxiety, depression and post-traumatic syndrome among nurses who were in contact with COVID-19 patients. A cross-sectional and correlational research design was used to recruit a sample of 278 nurses who treated COVID-19 patients in four government referral hospitals in Indonesia. A bootstrap resampling procedure was used to test the significance of the total and specific indirect effects of coping on relationships of Fear of COVID-19 with anxiety, depression and post-traumatic syndrome. The nurses reported moderate levels of fear of COVID-19, considerable anxiety and depression, and a moderate level of coping. We found coping to be significantly negatively correlated with the reported levels of anxiety, depression and post-traumatic syndrome (p < 0.001). Coping mediated relationships of fear of COVID-19 on depression, anxiety and post-traumatic syndrome after controlling for relevant confounders for each dependent variable. This shows that enacting coping mechanisms is important to achieve an adaptive effect on nurses' mental health. Proper assessments and interventions should be tailored and implemented for nurses who have contact with COVID-19 patients to facilitate their use of coping strategies when needed in stressful situations.

Construction, validation, and visualization of a web-based nomogram to predict overall survival in small-cell lung cancer patients with brain metastasis.

Brain metastasis (BM) is an aggressive complication with an extremely poor prognosis in patients with small-cell lung cancer (SCLC). A well-constructed prognostic model could help in providing timely survival consultation or optimizing treatments. We analyzed clinical data from SCLC patients between 2000 and 2018 based on the Surveillance, Epidemiology, and End Results (SEER) database. We identified significant prognostic factors and integrated them using a multivariable Cox regression approach. Internal validation of the model was performed through a bootstrap resampling procedure. Model performance was evaluated based on the area under the curve (AUC) and calibration curve. A total of 2,454 SCLC patients' clinical data was collected from the database. It was determined that seven clinical parameters were associated with prognosis in SCLC patients with BM. A satisfactory level of discrimination was achieved by the predictive model, with 6-, 12-, and 18-month AUC values of 0.726, 0.707, and 0.737 in the training cohort; and 0.759, 0.742, and 0.744 in the validation cohort. As measured by survival rate probabilities, the calibration curve agreed well with actual observations. Furthermore, prognostic scores were found to significantly alter the survival curves of different risk groups. We then deployed the prognostic model onto a website server so that users can access it easily. In this study, a nomogram and a web-based predictor were developed to predict overall survival in SCLC patients with BM. It may assist physicians in making informed clinical decisions and determining the best treatment plan for each patient.

Seismic fragility analysis of containment structure subjected to near fault ground motions

This study presents seismic fragility analysis of the containment structure subjected to near fault ground motion. Prior to perform seismic fragility analysis, 20 near fault ground motion records with pulse characteristics are selected PEER database. Nonlinear dynamic analysis of the containment structure is performed automatically using Matlab scripts and Batch files. To quantify the global damage behavior of the containment structure subjected to near fault ground motion, energy-based global damage index is proposed. The evolution of global damage behavior and tensile damage of the containment structure are analyzed in depth. Goodness of fit for two widely used fragility fitting methods is compared. To quantify the uncertainty in fragility function parameters, Bootstrap resampling procedure and analytical statistical inference method are adopted to estimate the confidence interval of the fragility function parameters. Results indicated that the bottom of containment structure suffered from severe tensile damage and it is most vulnerable location of the containment structure. With the increase of peak ground motion intensity level, tensile damage increases rapidly and extends to higher elevations of the containment structure. The relative error for fragility function parameter x˜m is not more than 4%, and the error of fragility function parameter β˜R is more than 12%. In general, analytical statistical method provides good estimation for the confidence interval of fragility function parameter x˜m, while overestimate the confidence interval of fragility function parameter β˜R.

A web-based calculator combining Geriatric Nutritional Risk Index (GNRI) and Tilburg Frailty Indicator (TFI) predicts postoperative complications among young elderly patients with gastric cancer.

Nutritional status and frailty are significant indicators reflecting physiological reserve. We sought to establish and validate a web-based calculator containing the Geriatric Nutritional Risk Index (GNRI) and the Tilburg Frailty Indicator (TFI) together with general clinical information to predict total complications among elderly patients with gastric cancer. This was a prospective cohort study of 582 elderly patients with gastric cancer in a tertiary hospital in China. Nutritional status and frailty were assessed by the GNRI and the TFI, respectively. The nomogram was built and further converted into a web-based calculator. The receiver operating characteristic analysis was performed to evaluate the discrimination of the nomogram. Calibration was assessed using the calibration curve and Hosmer-Lemeshow test via the bootstrap resampling procedure. The decision curve analyses (DCAs) were employed to quantify the net benefits of a certain threshold probability for assessing the clinical values. The GNRI (odds ratio [OR],0.921; 95% confidence interval [CI],0.895-0.949; P < 0.001), the TFI (OR,1.243; 95% CI,1.113-1.386; P < 0.001), surgical approach (OR,1.913; 95% CI,1.073-3.408; P=0.028) and comorbidity (OR=1.599, 95%CI=1.028-2.486, P=0.037) were independently associated with total complications. The nomogram demonstrated good discrimination (area under the receiver operating characteristic curve: training cohort,0.735; validation cohort,0.777) and calibration (P=0.135). The DCA curves of the nomogram also showed good positive net benefits. The web-based calculator incorporating the GNRI, the TFI, surgical approach, and comorbidity could successfully predict total complications among elderly patients with gastric cancer with good accuracy in a convenient manner. Geriatr Gerontol Int 2023; 23: 205-212.

The Influence of Mobile App and Media, Towards Entomophagy Awareness and Acceptability

The speedy development of technology in the use of smartphones and their mobile applications (apps) have become momentous and influential in today’s consumer perceptions. Accordingly, the media has in recent years advocated the nutritional, environmental, and economic benefits of edible insects. However, there is still a lacklustre aggressive multipronged approach to filling the awareness and acceptability gap of entomophagy through mobile apps and media. Therefore, this study aimed to assess the contribution of a developed and designed Mobile App in comparison with Other-Media Platforms in creating awareness and lack of acceptability. A total of 200 questionnaires were purposively distributed, 100 to each group and retained a 92% response rate. PLS-SEM was employed to simultaneously explain and predict the structural relationships among the (Insefood Mobile App and Other Media Platforms) and (Awareness and Acceptability). The reliability and validity of the measurement model were assessed using the four recommended procedural remedies; Factor loadings, Average Variance Extracted (AVE), Composite Reliability (CR), and Discriminant Validity. The structural model assessment and bootstrap resampling procedure was conducted to ascertain the statistical significance of path coefficients after running the PLS algorithm. The findings indicated high coefficients of determination for Insefood App, about 80% on awareness and 83% on acceptability, while Other Media Platforms’ coefficient of determination was 28% on awareness and 35% on acceptability. Predictive relevance of Insefood values 0.562 on acceptability and 0.535 on awareness, and other media platforms recorded 0.061 on acceptability and 0.081 on awareness at (p˂ 0.05). Thus, Insefood Mobile App contributed highly to awareness and acceptability than Other-Media Platforms. Affirmed the higher efficiency of the Insefood Mobile App in driving awareness and acceptability at a larger scale than Other Media Platforms. This informs the need to Engage in robust and more technological yet, vigorous insect-product marketing.

A dynamic risk factor assessment for myocardial infarction and cardiac arrest in patients undergoing pancreatectomy

To identify pancreatectomy specific risk factors for myocardial infarction and cardiac arrest (MICA) and to assess whether addition of new information obtained during the hospitalization changes these risk factors. Analysis was performed on elective pancreatectomy data from the ACS-NSQIP database (2014-2019). Risk factors were grouped into pre-operative, intra-operative, and postoperative phases. Factors were selected using a bootstrap resampling procedure to determine MICA association. Independent significance was assessed by logistic regression. In the first 30 days post-op, 650 of 39779 patients (1.88%) developed MICA. Some of the surgery specific, intra- and post-operative factors that were identified are: delayed gastric emptying (OR: 2.61; 95% CI: 2.12-3.21), total pancreatectomy (OR: 2.16; 95% CI: 1.29-3.42), pancreatic fistula (OR: 1.54; 95% CI: 1.25-1.90), post-operative transfusion (OR: 1.28; 95% CI: 1.03-1.58), and open approach (OR: 1.36; 95% CI: 1.05-1.77). Adding new variables improved statistical model performance and the c-statistic improved from 0.69 to 0.76 in the final analysis. Surgery specific, intra-, and post-operative factors were associated with MICA. Addition of new information during the hospital course changed risk factors and the statistical prediction of MICA risk improved.

Development and validation of prediction models for neurocognitive disorders in adult patients admitted to the ICU with sleep disturbance.

BackgroundNeurocognitive disorders (NCDs) and sleep disturbance are highly prevalent in the perioperative period and intensive care unit (ICU). There has been a lack of individualized evaluation tools designed for the high‐risk NCDs in critically ill patients with sleep disturbance.ObjectivesThe aim of this study was to develop and validate prediction models for NCDs among adult patients with sleep disturbance.MethodsThe R software was used to analyze the dataset of adult patients admitted to the ICU with sleep disturbance, who were diagnosed following the codes of the International Classification of Diseases, 9th Revision (ICD‐9) and 10th Revision (ICD‐10) using the MIMIC‐IV database. We used logistic regression and LASSO analyses to identify important risk factors associated with NCDs and develop nomograms for NCDs predictions. We measured the performances of the nomograms using the bootstrap resampling procedure, sensitivity, specificity of the receiver operating characteristic (ROC), area under the ROC curves (AUC), and decision curve analysis (DCA).ResultsThe prediction models shared the 10 risk factors (age, gender, midazolam, morphine, glucose, diabetes diseases, potassium, international normalized ratio, partial thromboplastin time, and respiratory rate). Cardiovascular diseases were included in the logistic regression, the sensitivity was 74.1%, and specificity was 64.6%. When platelet and Glasgow Coma Score (GCS) were included and cardiovascular diseases were removed in the LASSO prediction model, the sensitivity was 86.1% and specificity was 82.8%. Discriminative abilities of the logistic prediction and LASSO prediction models for NCDs in the validation set were evaluated as the AUC scores, which were 0.730 (95% CI 0.716–0.743) and 0.920 (95% CI 0.912–0.927). Net benefits of the prediction models were observed at threshold probabilities of 0.567 and 0.914.ConclusionsThe LASSO prediction model showed better performance than the logistic prediction model and should be preferred for nomogram‐assisted decisions on clinical risk management of NCDs among adult patients with sleep disturbance in the ICU.

Development of a Prognostic Nomogram and Nomogram Software Application Tool to Predict Overall Survival and Disease-Free Survival After Curative-Intent Gastrectomy for Gastric Cancer.

We sought to derive and validate a prediction model of survival and recurrence among Western patients undergoing resection of gastric cancer. Patients who underwent curative-intent surgery for gastric cancer at seven US institutions and a major Italian center from 2000 to 2020 were included. Variables included in the multivariable Cox models were identified using an automated model selection procedure based on an algorithm. Best models were selected using the Bayesian information criterion (BIC). The performance of the models was internally cross-validated via the bootstrap resampling procedure. Discrimination was evaluated using the Harrell's Concordance Index and accuracy was evaluated using calibration plots. Nomograms were made available as online tools. Overall, 895 patients met inclusion criteria. Age (hazard ratio [HR] 1.47, 95% confidence interval [CI] 1.17-1.84), presence of preoperative comorbidities (HR 1.66, 95% CI 1.14-2.41), lymph node ratio (LNR; HR 1.72, 95% CI 1.42-2.01), and lymphovascular invasion (HR 1.81, 95% CI 1.33-2.45) were associated with overall survival (OS; all p<0.01), whereas tumor location (HR 1.93, 95% CI 1.23-3.02), T category (Tis-T1 vs. T3: HR 0.31, 95% CI 0.14-0.66), LNR (HR 1.82, 95% CI 1.45-2.28), and lymphovascular invasion (HR 1.49; 95% CI 1.01-2.22) were associated with disease-free survival (DFS; all p<0.05) The models demonstrated good discrimination on internal validation relative to OS (C-index 0.70) and DFS (C-index 0.74). A web-based nomograms to predict OS and DFS among gastric cancer patients following resection demonstrated good accuracy and discrimination and good performance on internal validation.

Drinking water, sanitation, and hygiene (WASH) situation in primary schools of Pakistan: the impact of WASH-related interventions and policy on children school performance.

The United Nation's Sustainable Development Goals include the target of ensuring access to water and sanitation and hygiene (WASH) for all; however, very few studies have assessed comprehensive school WASH service in Pakistan. The purpose of this study was to identify WASH services in primary schools of Pakistan, and to assess how recent WASH interventions and policies are associated with the school's academic performance. A representative cross-sectional study was conducted in primary schools in the Sindh province of Pakistan. Structured observations and interviews were done to ascertain the schools' WASH conditions. The primary exposures of interest were the implementation of previous WASH interventions and National WASH policy in the school and the WASH coverage. Outcomes of interest included WASH conditions and school performance. The structural equation modeling (SEM) using a bootstrap resampling procedure was employed to characterize how WASH exposures were associated with WASH conditions and school performance. Data were collected from 425 schools. The Basic WASH facilities coverage in the primary schools of Sindh remains overall low according to WHO WASH service ladder criteria. Also, inconsistency in all three inclusive domains of WASH (availability, accessibility, and functionality) facilities were found. The school performance was significantly associated (P<0.001) with the presence of WASH interventions and/or WASH policy, while WASH policy and/or recent WASH intervention at the school were not associated with overall water quality. Our assessment unveiled several WASH gaps that exist, including high heavy metal and fecal contamination. Adoption of national WASH policy and financing of evidence-based WASH interventions are recommended in primary schools to improve educational outcomes.

Mediation of Coping Strategies among Patients with Advanced Cancer

The purpose of this study was to test the mediating effects of coping on relationships of psychological distress and stress with anxiety, depression, and quality of life. A cross-sectional and correlational research study was used to recruit a sample of 440 patients with advanced cancer in Indonesia. A bootstrap resampling procedure was used to test the significance of the total and specific indirect effects of coping. Data analysis showed that problem-focused coping (PFC) mediated relationships of psychological distress and stress on depression, anxiety and functional well-being. PFC also mediated the relationship between stress and social well-being. Emotional-focused coping (EFC) mediated the relationship of stress with physical and emotional well-being. EFC also mediated the relationships between psychological distress and physical well-being. Thus, proper assessments and interventions should be tailored and implemented for patients in order to facilitate their use of coping strategies when needed in stressful situations.

The Activation of Prothrombin Seems to Play an Earlier Role than the Complement System in the Progression of Colorectal Cancer: A Mass Spectrometry Evaluation.

Colorectal cancer (CRC) is the second cause of death in men and the third in women. This work deals with the study of the low molecular weight protein fraction of sera from patients who underwent surgery for CRC and who were followed for several years thereafter. MALDI-TOF MS was used to identify serum peptidome profiles of healthy controls, non-metastatic CRC patients and metastatic CRC patients. A multiple regression model was applied to signals preliminarily selected by SAM analysis to take into account the age and gender differences between the groups. We found that, while a signal m/z 2021.08, corresponding to the C3f fragment of the complement system, appears significantly increased only in serum from metastatic CRC patients, a m/z 1561.72 signal, identified as a prothrombin fragment, has a significantly increased abundance in serum from non-metastatic patients as well. The findings were also validated by a bootstrap resampling procedure. The present results provide the basis for further studies on large cohorts of patients in order to confirm C3f and prothrombin as potential serum biomarkers. Thus, new and non-invasive tests might be developed to improve the classification of colorectal cancer.

Monthly drought prediction based on ensemble models

Drought is a natural hazard, which is a result of a prolonged shortage of precipitation, high temperature and change in the weather pattern. Drought harms society, the economy and the natural environment, but it is difficult to identify and characterize. Many areas of Pakistan have suffered severe droughts during the last three decades due to changes in the weather pattern. A drought analysis with the incorporation of climate information has not yet been undertaken in this study region. Here, we propose an ensemble approach for monthly drought prediction and to define and examine wet/dry events. Initially, the drought events were identified by the short term Standardized Precipitation Index (SPI-3). Drought is predicted based on three ensemble models i.e., Equal Ensemble Drought Prediction (EEDP), Weighted Ensemble Drought Prediction (WEDP) and the Conditional Ensemble Drought Prediction (CEDP) model. Besides, two weighting procedures are used for distributing weights in the WEDP model, such as Traditional Weighting (TW) and the Weighted Bootstrap Resampling (WBR) procedure. Four copula families (i.e., Frank, Clayton, Gumbel and Joe) are used to explain the dependency relation between climate indices and precipitation in the CEDP model. Among all four copula families, the Joe copula has been found suitable for most of the times. The CEDP model provides better results in terms of accuracy and uncertainty as compared to other ensemble models for all meteorological stations. The performance of the CEDP model indicates that the climate indices are correlated with a weather pattern of four meteorological stations. Moreover, the percentage occurrence of extreme drought events that have appeared in the Multan, Bahawalpur, Barkhan and Khanpur are 1.44%, 0.57%, 2.59% and 1.71%, respectively, whereas the percentage occurrence of extremely wet events are 2.3%, 1.72%, 0.86% and 2.86%, respectively. The understanding of drought pattern by including climate information can contribute to the knowledge of future agriculture and water resource management.

Tu1487 DERIVATION AND EXTERNAL VALIDATION OF A PREDICTION MODEL (BEST-J SCORE) OF BLEEDING AFTER ENDOSCOPIC SUBMUCOSAL DISSECTION FOR EARLY GASTRIC CANCER

Bleeding after endoscopic submucosal dissection (ESD) for early gastric cancer (EGC) is a frequent adverse event (4.1%–8.5%). However, no prediction models have been reported for bleeding after ESD for EGC. In this study, we developed and externally validated a clinically useful prediction model (BEST-J score) for bleeding after ESD for EGC. We retrospectively enrolled patients who underwent ESD for EGC between 2013 and 2016. Patients in the derivation (n =8,291) and validation cohorts (n =2,029) were recruited from 25 and 8 institutions, respectively (Figure). This study followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis reporting guidelines. In the derivation stage, we evaluated the following 23 candidates for predictive factors of bleeding in the multivariate logistic regression analysis: age ≥75 years, sex, comorbidities (e.g., liver cirrhosis), each type of antithrombotic therapy, perioperative management of antithrombotic agents (e.g., interruption of antithrombotic agents), the tumor characteristics (e.g., tumor differentiation), procedure (e.g., ESD procedure time >120 min.), and endoscopic gastric atrophy. Then, weighted points proportional to β regression coefficient values were assigned to significant predictors of bleeding and a prediction model was established. External validation of the model was conducted to analyze discrimination and calibration. Bleeding after ESD was shown in 4.7% of the derivation cohort. Among 23 candidate predictors of bleeding, 9 variables were eligible for a prediction model: 4 points each for warfarin and direct oral anticoagulant, 3 points for chronic kidney disease with hemodialysis, 2 points each for thienopyridine and aspirin, and 1 point each for cilostazol, a tumor size >30 mm, lower-third in tumor location, and presence of multiple tumors. This prediction model showed moderately good discrimination, with a c-statistic of 0.71 (95% confidence interval [CI], 0.68–0.74) and this was well calibrated (p =0.689). This model was then internally validated by bootstrap resampling procedure, which gave a c-statistic of 0.71 (95% CI, 0.68–0.73). When the total score was categorized as low- (0–1 points), intermediate- (2–3 points), high- (4–5 points), or very high-risk (≥6 points) for bleeding after ESD for EGC, a significantly increasing trend of risk from low- to very high-risk groups was observed (p <0.001, Cochran–Armitage trend test)(Table). In the external validation cohort, the model showed moderately good discrimination, with a c-statistic of 0.70 (95% CI, 0.64–0.76), and good calibration (p =0.168). In this nationwide multicenter study, we derived a prediction model (BEST-J score) for bleeding after ESD for EGC. This model may be a good clinical decision-making support tool for ESD in patients with EGC.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

Arsenic Trioxide Abrogate MN1 Mediated RA-Resistance in Acute Promyelocytic Leukemia

Arsenic Trioxide Abrogate MN1 Mediated RA-Resistance in Acute Promyelocytic Leukemia

Molecular-Based Score Inspired on Metabolic Signature Improves Prognostic Stratification for Myelodysplastic Syndrome

Background : The recent efforts to uncover the molecular heterogeneity of myelodysplastic syndromes (MDS), mainly by new sequencing technologies, allow the comprehensive identification of driver mutations and/or altered gene expression recurrently found in a recognizable fraction of patients. Ongoing efforts are being made to clarify the impact of molecular changes on clinical phenotype and prognosis, as well as their role in the pathogenesis of MDS. Refining risk stratification allows the proposition of risk-adapted therapy and may shed light in biology of MDS. Aims: Based on the gene expression of selected metabolic targets, we aimed to design a score system that improves MDS overall survival prediction. Patients and methods: Clinical, mutations and transcriptomic data from CD34+ cells from 159 MDS patients and 17 healthy volunteers freely available at Gene Expression Omnibus (GEO/NCBI: GSE58831) were used in the present work. Forty-one genes related to metabolic processes, previously demonstrated as deregulated among diverse neoplastic conditions, were ranked and asked for differential expression and prognostic impact. Each gene was dichotomized according to Receiving-Operating Curve (ROC) and Cox Proportional-Hazard Model was used for multivariate analysis using gender, age and IPSS-R as cofounders. Genes independently associated with overall survival (OS) were selected to compose the Molecular-Based Score (MBS) and integer weight of each one was defined according Hazard Ratio (HR). Survival curves were constructed using Kaplan-Meyer method and compared with Log-Rank Test. ROC c-statistic was used to measure the predictive function of MBS. Prediction accuracy of MBS was cross-validated by a nonparametric bootstrap procedure with 1,000 resamplings of the original cohort allowing replacement and also estimated their respective 95% confidence interval (95% CI) computing the bias-corrected and accelerated bootstrap interval. Results: Among selected genes, 18 were differentially expressed between CD34+ cells from MDS and healthy volunteers. Fifteen genes predict OS in univariate analysis, of which ACLY (HR: 0.48; 95%CI: 0.24 - 0.96; P=0.04), ANPEP (HR: 2.16; 95%CI: 1.08 - 4.31; P=0.02), PANK1 (HR: 0.43; 95%CI: 0.19 - 0.98; P=0.04), PKM (HR: 2.01; 95%CI: 1.02 - 3.93; P=0.04) and SLC25A5 (HR: 0.52; 95%CI: 0.27 - 0.99; P=0.05) were independently associated with OS. Higher expression of ANPEP and PKM, as well as lower expression of ACLY, PANK1 and SLC25A5 were considered to integer high risk being attributed weight 2 for each condition. MBS varied from 0 to 10 (median=2) and was calculated as: MBS Low-Risk =0 (MBS-LR; n=28); MBS Intermediate-Risk=2 and 4 (MBS-IR; n=90) and High-Risk: ≥6 (MBS-HR; n=48). The modeled MBS showed a ROC c-statistic of 0.699 (95%CI: 0.603 - 0.794) and HR=3.05 (95%CI: 1.81 - 5.05; P&lt;0.001) and efficiently identified patients with different risk: MBS-LR (3-year OS: 100%; median time [MT]: not reached); MBS-IR (3-year OS: 66% [95%CI: 53% - 83%]; MT: 52.2 months [95%CI: 34.2 - 71.5]) and MBS-HR (3-year OS: 32% [95%CI: 17% - 61%]; MT: 23.5 months [95%CI: 11.6 - 35.4]). Using Cox Proportional Hazard Model for multivariate analysis, the proposed MBS (HR:2.5 [95%CI: 1.2-4.96]; P=0.008) was independently associated with OS using gender (HR:0.48 [95%CI: 0.21-1.06]; P=0.07), age (HR:1.03 [95%CI: 1.001-1.07]; P=0.02) and IPSS-R (HR:0.98 [95%CI: 0.69-1.36]; P=0.88) as confounders. The bootstrap resampling procedure validated the MBS and demonstrated the stability of its prediction (3-year OS for MBS-LR: 100% [95%CI: 100-100%]; MBS-IR: 66% [95%CI: 49-80%]; MBS-HR: 32% [95%CI: 13-54%]; P&lt;0.001). Conclusions: The proposed Molecular-Based Score (MBS) independently predict OS with superior efficacy in comparison to the most clinically relevant prognostic factors, and reinforce the therapeutic opportunity of metabolic processes in MDS. Figure Disclosures No relevant conflicts of interest to declare.

Combining gene mutation with gene expression analysis improves outcome prediction in acute promyelocytic leukemia

AbstractBy combining the analysis of mutations with aberrant expression of genes previously related to poorer prognosis in both acute promyelocytic leukemia (APL) and acute myeloid leukemia, we arrived at an integrative score in APL (ISAPL) and demonstrated its relationship with clinical outcomes of patients treated with all-trans retinoic acid (ATRA) in combination with anthracycline-based chemotherapy. Based on fms-like tyrosine kinase-3–internal tandem duplication mutational status; the ΔNp73/TAp73 expression ratio; and ID1, BAALC, ERG, and KMT2E gene expression levels, we modeled ISAPL in 159 patients (median ISAPL score, 3; range, 0-10). ISAPL modeling identified 2 distinct groups of patients, with significant differences in early mortality (P < .001), remission (P = .004), overall survival (P < .001), cumulative incidence of relapse (P = .028), disease-free survival (P = .03), and event-free survival (P < .001). These data were internally validated by using a bootstrap resampling procedure. At least for patients treated with ATRA and anthracycline-based chemotherapy, ISAPL modeling may identify those who need to be treated differently to maximize their chances for a cure.