Development and validation of a risk prediction score for severe acute pancreatitis

Exploring the value of blood urea nitrogen-to-albumin ratio in patients with acute pancreatitis admitted to the intensive care unit: a retrospective cohort study

Background

Although blood urea nitrogen and albumin alone are well-known clinical indicators, combining them as the blood urea nitrogen-to-albumin ratio (BAR) may provide additional prognostic information because they reflect the complex interplay between renal function, nutritional status, and systemic inflammation-all of which are key factors in the pathogenesis of acute pancreatitis (AP). Therefore, the objective of this study was to investigate the relationships between BAR and short- and long-term all-cause mortality (ACM) in patients with AP and to assess the prognostic significance of the BAR in AP.

Methods

This retrospective investigation utilized information extracted from the Medical Information Mart for Intensive Care-IV (MIMIC-IV, Version 2.2) database. BAR was calculated using the BUN/ALB ratio obtained from the first measurement within 24 h of admission. R software was used to identify the optimal threshold for the BAR. The Kaplan-Meier (K-M) analysis was performed to compare mortality between the two groups. Multivariate Cox proportional hazards regression models and restricted cubic splines (RCS) were used to evaluate the association between BAR and 14-day, 28-day, 90-day, and 1-year ACM. The receiver operating characteristic curves were used to investigate the predictive ability, sensitivity, specificity, and area under the curve (AUC) of the BAR for short- and long-term mortality in AP patients. Subgroup analysis was performed to illustrate the reliability of our findings.

Results

This study comprised a total of 569 patients. The R software determined the optimal threshold for the BAR to be 16.92. The K-M analysis indicated a notable rise in ACM in patients with higher BAR (all log-rank p < 0.001). Cox proportional hazard regression models revealed independent associations between higher BAR and ACM before and after adjusting for confounding variables at days 14, 28, 90, and 1 year. The RCS analysis revealed J-shaped correlations between the BAR and short- and long-term ACM. The AUCs of the BAR for predicting ACM at days 14, 28, 90, and 1 year were 73.23, 76.14, 73.49, and 71.00%, respectively, which were superior to those of BUN, ALB, creatinine, Sequential Organ Failure Assessment, and Acute Physiology and Chronic Health Evaluation-II. Subgroup analyses revealed no significant interaction between BAR and the vast majority of subgroups.

Conclusion

This study revealed, for the first time, the unique prognostic value of BAR in ICU-managed AP patients. Higher levels of BAR were associated with higher short- and long-term ACM in ICU-managed AP patients.

Logistic regression and other statistical tools in diagnostic biomarker studies

A biomarker is a measured indicator of a variety of processes, and is often used as a clinical tool for the diagnosis of diseases. While the developmental process of biomarkers from lab to clinic is complex, initial exploratory stages often focus on characterizing the potential of biomarkers through utilizing various statistical methods that can be used to assess their discriminatory performance, establish an appropriate cut-off that transforms continuous data to apt binary responses of confirming or excluding a diagnosis, or establish a robust association when tested against confounders. This review aims to provide a gentle introduction to the most common tools found in diagnostic biomarker studies used to assess the performance of biomarkers with an emphasis on logistic regression.

Application Value of the Automated Machine Learning Model Based on Modified Computed Tomography Severity Index Combined With Serological Indicators in the Early Prediction of Severe Acute Pancreatitis

BACKGROUND AND AIMS

Machine learning (ML) algorithms are widely applied in building models of medicine due to their powerful studying and generalizing ability. To assess the value of the Modified Computed Tomography Severity Index (MCTSI) combined with serological indicators for early prediction of severe acute pancreatitis (SAP) by automated ML (AutoML).

PATIENTS AND METHODS

The clinical data, of the patients with acute pancreatitis (AP) hospitalized in Hospital 1 and hospital 2 from January 2017 to December 2021, were retrospectively analyzed. Serological indicators within 24 hours of admission were collected. MCTSI score was completed by noncontrast computed tomography within 24 hours of admission. Data from the hospital 1 were adopted for training, and data from the hospital 2 were adopted for external validation. The diagnosis of AP and SAP was based on the 2012 revised Atlanta classification of AP. Models were built using traditional logistic regression and AutoML analysis with 4 types of algorithms. The performance of models was evaluated by the receiver operating characteristic curve, the calibration curve, and the decision curve analysis based on logistic regression and decision curve analysis, feature importance, SHapley Additive exPlanation Plot, and Local Interpretable Model Agnostic Explanation based on AutoML.

RESULTS

A total of 499 patients were used to develop the models in the training data set. An independent data set of 201 patients was used to test the models. The model developed by the Deep Neural Net (DL) outperformed other models with an area under the receiver operating characteristic curve (areas under the curve) of 0.907 in the test set. Furthermore, among these AutoML models, the DL and gradient boosting machine models achieved the highest sensitivity values, both exceeding 0.800.

CONCLUSION

The AutoML model based on the MCTSI score combined with serological indicators has good predictive value for SAP in the early stage.

Predictive value of machine learning for the severity of acute pancreatitis: A systematic review and meta-analysis

Background

Predicting the severity of acute pancreatitis (AP) early poses a challenge in clinical practice. While there are well-established clinical scoring tools, their actual predictive performance remains uncertain. Various studies have explored the application of machine-learning methods for early AP prediction. However, a more comprehensive evidence-based assessment is needed to determine their predictive accuracy. Hence, this systematic review and meta-analysis aimed to evaluate the predictive accuracy of machine learning in assessing the severity of AP.

Methods

PubMed, EMBASE, Cochrane Library, and Web of Science were systematically searched until December 5, 2023. The risk of bias in eligible studies was assessed using the Prediction Model Risk of Bias Assessment Tool (PROBAST). Subgroup analyses, based on different machine learning types, were performed. Additionally, the predictive accuracy of mainstream scoring tools was summarized.

Results

This systematic review ultimately included 33 original studies. The pooled c-index in both the training and validation sets was 0.87 (95 % CI: 0.84-0.89) and 0.88 (95 % CI: 0.86-0.90), respectively. The sensitivity in the training set was 0.81 (95 % CI: 0.77-0.84), and in the validation set, it was 0.79 (95 % CI: 0.71-0.85). The specificity in the training set was 0.84 (95 % CI: 0.78-0.89), and in the validation set, it was 0.90 (95 % CI: 0.86-0.93). The primary model incorporated was logistic regression; however, its predictive accuracy was found to be inferior to that of neural networks, random forests, and xgboost. The pooled c-index of the APACHE II, BISAP, and Ranson were 0.74 (95 % CI: 0.68-0.80), 0.77 (95 % CI: 0.70-0.85), and 0.74 (95 % CI: 0.68-0.79), respectively.

Conclusions

Machine learning demonstrates excellent accuracy in predicting the severity of AP, providing a reference for updating or developing a straightforward clinical prediction tool.

Development and validation of a multimodal model in predicting severe acute pancreatitis based on radiomics and deep learning

OBJECTIVE

Aim to establish a multimodal model for predicting severe acute pancreatitis (SAP) using machine learning (ML) and deep learning (DL).

METHODS

In this multicentre retrospective study, patients diagnosed with acute pancreatitis at admission were enrolled from January 2017 to December 2021. Clinical information within 24 h and CT scans within 72 h of admission were collected. First, we trained Model α based on clinical features selected by least absolute shrinkage and selection operator analysis. Second, radiomics features were extracted from 3D-CT scans and Model β was developed on the features after dimensionality reduction using principal component analysis. Third, Model γ was trained on 2D-CT images. Lastly, a multimodal model, namely PrismSAP, was constructed based on aforementioned features in the training set. The predictive accuracy of PrismSAP was verified in the validation and internal test sets and further validated in the external test set. Model performance was evaluated using area under the curve (AUC), accuracy, sensitivity, specificity, recall, precision and F1-score.

RESULTS

A total of 1,221 eligible patients were randomly split into a training set (n = 864), a validation set (n = 209) and an internal test set (n = 148). Data of 266 patients were for external testing. In the external test set, PrismSAP performed best with the highest AUC of 0.916 (0.873-0.960) among all models [Model α: 0.709 (0.618-0.800); Model β: 0.749 (0.675-0.824); Model γ: 0.687 (0.592-0.782); MCTSI: 0.778 (0.698-0.857); RANSON: 0.642 (0.559-0.725); BISAP: 0.751 (0.668-0.833); SABP: 0.710 (0.621-0.798)].

CONCLUSION

The proposed multimodal model outperformed any single-modality models and traditional scoring systems.

The red cell distribution width-to-albumin ratio: A simple index has high predictive accuracy for clinical outcomes in patients with acute pancreatitis

BACKGROUND/OBJECTIVES

Ongoing research is seeking to identify the best prognostic marker for acute pancreatitis (AP). The purpose of this study was to investigate the role of the red blood cell distribution width-to-albumin ratio (RAR) in the prognosis of AP.

METHODS

This 18-month prospective cohort study was conducted between June 2021 and December 2022 with patients diagnosed with AP. The patients were divided into two groups: severe AP (SAP) and non-severe AP. Factors associated with SAP within the first 48 h of admission were determined. In addition, RAR values at admission and at 48 h (RAR-48th) were calculated, and their ability to predict clinical outcomes was assessed. The primary outcomes were severe disease and in-hospital mortality.

RESULTS

Fifty (13.7 %) of 365 patients had SAP. Systemic inflammatory response syndrome, blood urea nitrogen, calcium, and RAR at 48 h after admission were independent predictors of SAP. When RAR-48th was >4.35, the risk of SAP increased approximately 18-fold (OR: 18.59; 95 % CI: 8.58-40.27), whereas no patients with a RAR-48th value of <4.6 died. For in-hospital mortality, the area under the curve (AUC) value of RAR-48th was 0.960 (95 % CI: 0.931-0.989), significantly higher than the AUC values of existing scoring systems. The results of RAR-48th were comparable to those of the other scoring systems with regard to the remaining clinical outcomes.

CONCLUSIONS

RAR-48th successfully predicted clinical outcomes, particularly in-hospital mortality. Being simple and readily calculable, RAR-48th is a promising alternative to burdensome and complex scoring systems for the prediction of clinical outcomes in AP.

Incidence and prognostic role of pleural effusion in patients with acute pancreatitis: a meta-analysis

BACKGROUND

Pleural effusion (PE) is reported as a common complication in acute pancreatitis (AP), while the incidence of PE in AP varies widely among studies, and the association between PE and mortality is not clear. This study aimed to comprehensively analyze the pooled incidence of PE in patients with AP and to evaluate the influence of PE on mortality through a meta-analysis.

METHOD

Six databases (PubMed, Web of Science, EMBASE, Cochrane, Scopus, and OVID) were searched thoroughly for relevant studies. Data were extracted, and Stata SE 16.0 software was applied to compute the pooled incidence of PE and assess the association between PE and mortality, taking the risk ratio (RR) as the effect size.

RESULTS

Thirty-five articles involving 7,675 patients with AP were eventually included in this meta-analysis. The pooled incidence of PE was 34% (95% CI: 28%-39%), with significant heterogeneity among studies (I2=96.7%). Further analysis revealed that the pooled incidence of unilateral and small PE occupied 49% (95% CI: 21%-77%) and 59% (95% CI: 38%-81%) of AP patients complicated by PE, respectively. The subgroup analysis revealed that "region" and "examination method" may contribute to heterogeneity. PE may be associated with increased mortality in AP patients (RR 3.99, 95% CI: 1.73-9.2).

CONCLUSION

This study suggested that PE is a common complication with high pooled incidence and that PE may be associated with increased mortality in AP patients. More studies should be performed to validate our findings.

Acute pancreatitis: A review of diagnosis, severity prediction and prognosis assessment from imaging technology, scoring system and artificial intelligence

Acute pancreatitis (AP) is a potentially life-threatening inflammatory disease of the pancreas, with clinical management determined by the severity of the disease. Diagnosis, severity prediction, and prognosis assessment of AP typically involve the use of imaging technologies, such as computed tomography, magnetic resonance imaging, and ultrasound, and scoring systems, including Ranson, Acute Physiology and Chronic Health Evaluation II, and Bedside Index for Severity in AP scores. Computed tomography is considered the gold standard imaging modality for AP due to its high sensitivity and specificity, while magnetic resonance imaging and ultrasound can provide additional information on biliary obstruction and vascular complications. Scoring systems utilize clinical and laboratory parameters to classify AP patients into mild, moderate, or severe categories, guiding treatment decisions, such as intensive care unit admission, early enteral feeding, and antibiotic use. Despite the central role of imaging technologies and scoring systems in AP management, these methods have limitations in terms of accuracy, reproducibility, practicality and economics. Recent advancements of artificial intelligence (AI) provide new opportunities to enhance their performance by analyzing vast amounts of clinical and imaging data. AI algorithms can analyze large amounts of clinical and imaging data, identify scoring system patterns, and predict the clinical course of disease. AI-based models have shown promising results in predicting the severity and mortality of AP, but further validation and standardization are required before widespread clinical application. In addition, understanding the correlation between these three technologies will aid in developing new methods that can accurately, sensitively, and specifically be used in the diagnosis, severity prediction, and prognosis assessment of AP through complementary advantages.

Establishment and Validation of a Nomogram Prediction Model for the Severe Acute Pancreatitis

Background

Severe acute pancreatitis (SAP) can progress to lung and kidney dysfunction, and blood clotting within 48 hours of its onset, and is associated with a high mortality rate. The aim of this study was to establish a reliable diagnostic prediction model for the early stage of severe pancreatitis.

Methods

The clinical data of patients diagnosed with acute pancreatitis from October 2017 to June 2022 at the Shangluo Central Hospital were collected. The risk factors were screened by least absolute shrinkage and selection operator (LASSO) regression analysis. A novel nomogram model was then established by multivariable logistic regression analysis.

Results

The data of 436 patients with acute pancreatitis, 45 (10.3%) patients had progressed to SAP. Through univariate and LASSO regression analyses, the neutrophils (P <0.001), albumin (P < 0.001), blood glucose (P < 0.001), serum calcium (P < 0.001), serum creatinine (P < 0.001), blood urea nitrogen (P < 0.001) and procalcitonin (P = 0.005) were identified as independent predictive factors for SAP. The nomogram built on the basis of these factors predicted SAP with sensitivity of 0.733, specificity of 0.9, positive predictive value of 0.458 and negative predictive value of 0.967. Furthermore, the concordance index of the nomogram reached 0.889 (95% CI, 0.837-0.941), and the area under the curve (AUC) in receiver operating characteristic curve (ROC) analysis was significantly higher than that of the APACHEII and ABISAP scoring systems. The established model was validated by plotting the clinical decision curve analysis (DCA) and clinical impact curve (CIC).

Conclusion

We established a nomogram to predict the progression of early acute pancreatitis to SAP with high discrimination and accuracy.

Real-time prediction of organ failures in patients with acute pancreatitis using longitudinal irregular data

It is extremely important to identify patients with acute pancreatitis who are at high risk for developing persistent organ failures early in the course of the disease. Due to the irregularity of longitudinal data and the poor interpretability of complex models, many models used to identify acute pancreatitis patients with a high risk of organ failure tended to rely on simple statistical models and limited their application to the early stages of patient admission. With the success of recurrent neural networks in modeling longitudinal medical data and the development of interpretable algorithms, these problems can be well addressed. In this study, we developed a novel model named Multi-task and Time-aware Gated Recurrent Unit RNN (MT-GRU) to directly predict organ failure in patients with acute pancreatitis based on irregular medical EMR data. Our proposed end-to-end multi-task model achieved significantly better performance compared to two-stage models. In addition, our model not only provided an accurate early warning of organ failure for patients throughout their hospital stay, but also demonstrated individual and population-level important variables, allowing physicians to understand the scientific basis of the model for decision-making. By providing early warning of the risk of organ failure, our proposed model is expected to assist physicians in improving outcomes for patients with acute pancreatitis.

Relationship between Cholesterol-Related Lipids and Severe Acute Pancreatitis: From Bench to Bedside

It is well known that hypercholesterolemia in the body has pro-inflammatory effects through the formation of inflammasomes and augmentation of TLR (Toll-like receptor) signaling, which gives rise to cardiovascular disease and neurodegenerative diseases. However, the interaction between cholesterol-related lipids and acute pancreatitis (AP) has not yet been summarized before. This hinders the consensus on the existence and clinical importance of cholesterol-associated AP. This review focuses on the possible interaction between AP and cholesterol-related lipids, which include total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) and apolipoprotein (Apo) A1, from the bench to the bedside. With a higher serum level of total cholesterol, LDL-C is associated with the severity of AP, while the persistent inflammation of AP is allied with a decrease in serum levels of cholesterol-related lipids. Therefore, an interaction between cholesterol-related lipids and AP is postulated. Cholesterol-related lipids should be recommended as risk factors and early predictors for measuring the severity of AP. Cholesterol-lowering drugs may play a role in the treatment and prevention of AP with hypercholesterolemia.

The Potential Role of Model for End-Stage Liver Disease (MELD)-Sodium Score in Predicting the Severity of Acute Pancreatitis

Background and aim Acute pancreatitis is a common inflammation of the pancreas which can be severe and even potentially mortal. High rates of mortality showed the importance of immediate identification of patients at high risk and led the clinicians to refer to various scoring systems. Our aim was to investigate a clinical predictive model using the Model for End-Stage Liver Disease-Sodium (MELD-sodium) scoring system, adapting it to acute pancreatitis patients referring to the systemic inflammatory nature of the disease and potential multi-organ failures in severe form. Methods Our multicenter study was designed retrospectively. The medical records were reviewed for the period of two years. Demographics, biochemical results, MELD-sodium scores and mortality rates were analysed. Results MELD-sodium score was found to be statistically correlated with both mortality and the severity of pancreatitis (p<0.001) and significant difference between both mild and severe (p<0.001), moderate and severe groups (p<0.001). Mortality was found to be significantly higher in patients with MELD-Na score when the cut-off value was accepted as '≥11'. Conclusion We found that MELD-sodium score was significantly associated with both severity of disease and mortality rates and also significantly effective between both mild/severe and moderate/severe groups which may be a guide for future multi-center reviews with larger patient and control groups, which can define the potential role of this non-invasive and easy-to-use predictive model in acute pancreatitis patients.

Impact of multiple drug-resistant Gram-negative bacterial bacteraemia on infected pancreatic necrosis patients

Introduction

Multiple drug-resistant Gram-negative bacterial (MDR-GNB) bacteraemia poses a serious threat to patients in hospital. Infected pancreatic necrosis (IPN) patients are a vulnerable population to infectious complications during hospitalization. This study aims to evaluate the impact of MDR Gram-negative bacteraemia on IPN patients.

Methods

A case-control study was performed with data collected from 1 January 2016 to 1 July 2022 in a Chinese tertiary teaching hospital. Clinical data of the IPN patients with MDR-GNB bacteraemia were analyzed and compared to those of a matched control group without MDR-GNB bacteraemia (case-control ratio of 1:2). Comparisons were performed between with/without MDR-GNB bacteraemia and different severities of acute pancreatitis (AP). Independent predictors of overall mortality were identified via univariate and multivariate binary logistic regression analyses.

Results

MDR-GNB bacteraemia was related to a higher mortality rate (62.5% vs. 8.3%, p < 0.001). Severe AP combined with MDR-GNB bacteraemia further increased mortality up to 81.3% (p = 0.025). MDR-GNB bacteraemia (odds ratio (OR) = 8.976, 95% confidence interval (CI) = 1.805 -44.620, p = 0.007) and severe AP (OR = 9.414, 95% CI = 1.742 -50.873, p = 0.009) were independent predictors of overall mortality. MDR- Klebsiella pneumoniae was the most common causative pathogen.

Conclusion

A higher mortality rate in IPN patients was related to MDR-GNB bacteraemia and further increased in severe AP patients combined with MDR-GNB bacteraemia.

Non-linear correlation between amylase day 2 to day 1 ratio and incidence of severe acute pancreatitis

BACKGROUND

This study aimed to assess whether the amylase day 2/amylase day 1 ratio was associated with severe acute pancreatitis (SAP).

METHODS

We retrospectively enrolled 464 patients with acute pancreatitis. Serum amylase was measured on admission (day 1) and 24 h later (day 2). Univariable logistic regression with restricted cubic spline analysis, multivariable logistic analysis, and receiver operating characteristic curve analysis was used to evaluate the relationship between the amylase day 2/amylase day 1 ratio and SAP.

RESULTS

A non-linear association between the amylase day 2/amylase day 1 ratio and SAP was observed. The multivariable logistic analysis confirmed that a high amylase day 2/amylase day 1 ratio (≥0.3) was independently associated with the development of SAP (OR: 6.62). The area under the receiver operating characteristic curve (AUC) of the amylase day 2/amylase day 1 ratio, as a predictive factor for SAP, was 0.65. When amylase ratio ≥0.3 was counted as 1 point and added to the BISAP score to build a new model named the BISAPA (BISAP plus Amylase ratio) score (AUC = 0.86), it improved the diagnostic power of the original BISAP score (AUC = 0.83) for SAP. With a cut-off value of 3, the BISAPA score achieved a sensitivity of 66.0%, a specificity of 86.7%, and diagnostic accuracy of 84.48%.

CONCLUSIONS

There is a non-linear correlation between the amylase day 2/amylase day 1 ratio and the incidence of SAP. BISAPA score might also be a useful tool for the same purpose.

Antenatal risk score for prediction of shoulder dystocia with focus on fetal ultrasound data

BACKGROUND

Shoulder dystocia is one of the most threatening complications during delivery, and although it is difficult to predict, individual risk should be considered when counseling for mode of delivery.

OBJECTIVE

This study aimed to develop and validate a risk score for shoulder dystocia based on fetal ultrasound and maternal data from 15,000 deliveries.

STUDY DESIGN

Data were retrospectively obtained of deliveries in 3 tertiary centers between 2014 and 2017 for the derivation cohort and between 2018 and 2020 for the validation cohort. Inclusion criteria were singleton pregnancy, vaginal delivery in cephalic presentation at ≥37+0 weeks' gestation, and fetal biometry data available within 2 weeks of delivery. Independent predictors were determined by multivariate regression analysis in the derivation cohort, and a score was developed on the basis of the effect of the predictors.

RESULTS

The derivation cohort consisted of 7396 deliveries with a 0.91% rate of shoulder dystocia, and the validation cohort of 7965 deliveries with a 1.0% rate of shoulder dystocia. Among all women, 13.8% had diabetes mellitus, and 12.1% were obese (body mass index ≥30 kg/m2). Independent risk factors in the derivation cohort were: estimated fetal weight ≥4250 g (odds ratio, 4.27; P=.002), abdominal-head-circumference ≥2.5 cm (odds ratio, 3.96; P<.001), and diabetes mellitus (odds ratio, 2.18; P=.009). On the basis of the strength of effect, a risk score was developed: estimated fetal weight ≥4250 g=2, abdominal-head-circumference ≥2.5 cm=2, and diabetes mellitus=1. The risk score predicted shoulder dystocia with moderate discriminatory ability (area under the receiver-operating characteristic curve, 0.69; P<.001; area under the receiver-operating characteristic curve, 0.71; P<.001) and good calibration (Hosmer-Lemeshow goodness-of-fit; P=.466; P=.167) for the derivation and validation cohorts, respectively. With 1 score point, 16 shoulder dystocia cases occurred in 1764 deliveries, with 0.6% shoulder dystocia incidence and a number needed to treat with cesarean delivery to avoid 1 case of shoulder dystocia of 172 (2 points: 38/1809, 2.1%, 48; 3 points: 18/336, 5.4%, 19; 4 points: 10/96, 10.5%, 10; and 5 points: 5/20, 25%, 4); 40.8% of the shoulder dystocia cases occurred without risk factors.

CONCLUSION

The presented risk score for shoulder dystocia may act as a supplemental tool for the clinical decision-making regarding mode of delivery. According to our score model, in pregnancies with a score ≤2, meaning having solely estimated fetal weight ≥4250 g, or abdominal-head-circumference ≥2.5, or diabetes mellitus, cesarean delivery for prevention of shoulder dystocia should not be recommended because of the high number needed to treat to avoid 1 case of shoulder dystocia. Conversely, in patients with a score of ≥4 with or without diabetes mellitus, cesarean delivery may be considered. However, in 40% of the shoulder dystocia cases, no risk factors had been present.

Development and validation of a risk prediction score for the severity of acute hypertriglyceridemic pancreatitis in Chinese patients

BACKGROUND

The frequency of acute hypertriglyceridemic pancreatitis (AHTGP) is increasing worldwide. AHTGP may be associated with a more severe clinical course and greater mortality than pancreatitis caused by other causes. Early identification of patients with severe inclination is essential for clinical decision-making and improving prognosis. Therefore, we first developed and validated a risk prediction score for the severity of AHTGP in Chinese patients.

AIM

To develop and validate a risk prediction score for the severity of AHTGP in Chinese patients.

METHODS

We performed a retrospective study including 243 patients with AHTGP. Patients were randomly divided into a development cohort (n = 170) and a validation cohort (n = 73). Least absolute shrinkage and selection operator and logistic regression were used to screen 42 potential predictive variables to construct a risk score for the severity of AHTGP. We evaluated the performance of the nomogram and compared it with existing scoring systems. Last, we used the best cutoff value (88.16) for severe acute pancreatitis (SAP) to determine the risk stratification classification.

RESULTS

Age, the reduction in apolipoprotein A1 and the presence of pleural effusion were independent risk factors for SAP and were used to construct the nomogram (risk prediction score referred to as AAP). The concordance index of the nomogram in the development and validation groups was 0.930 and 0.928, respectively. Calibration plots demonstrate excellent agreement between the predicted and actual probabilities in SAP patients. The area under the curve of the nomogram (0.929) was better than those of the Bedside Index of Severity in AP (BISAP), Ranson, Acute Physiology and Chronic Health Evaluation (APACHE II), modified computed tomography severity index (MCTSI), and early achievable severity index scores (0.852, 0.825, 0.807, 0.831 and 0.807, respectively). In comparison with these scores, the integrated discrimination improvement and decision curve analysis showed improved accuracy in predicting SAP and better net benefits for clinical decisions. Receiver operating characteristic curve analysis was used to determine risk stratification classification for AHTGP by dividing patients into high-risk and low-risk groups according to the best cutoff value (88.16). The high-risk group (> 88.16) was closely related to the appearance of local and systemic complications, Ranson score ≥ 3, BISAP score ≥ 3, MCTSI score ≥ 4, APACHE II score ≥ 8, C-reactive protein level ≥ 190, and length of hospital stay.

CONCLUSION

The nomogram could help identify AHTGP patients who are likely to develop SAP at an early stage, which is of great value in guiding clinical decisions.

Automated Machine Learning for the Early Prediction of the Severity of Acute Pancreatitis in Hospitals

Background

Machine learning (ML) algorithms are widely applied in building models of medicine due to their powerful studying and generalizing ability. This study aims to explore different ML models for early identification of severe acute pancreatitis (SAP) among patients hospitalized for acute pancreatitis.

Methods

This retrospective study enrolled patients with acute pancreatitis (AP) from multiple centers. Data from the First Affiliated Hospital and Changshu No. 1 Hospital of Soochow University were adopted for training and internal validation, and data from the Second Affiliated Hospital of Soochow University were adopted for external validation from January 2017 to December 2021. The diagnosis of AP and SAP was based on the 2012 revised Atlanta classification of acute pancreatitis. Models were built using traditional logistic regression (LR) and automated machine learning (AutoML) analysis with five types of algorithms. The performance of models was evaluated by the receiver operating characteristic (ROC) curve, the calibration curve, and the decision curve analysis (DCA) based on LR and feature importance, SHapley Additive exPlanation (SHAP) Plot, and Local Interpretable Model Agnostic Explanation (LIME) based on AutoML.

Results

A total of 1,012 patients were included in this study to develop the AutoML models in the training/validation dataset. An independent dataset of 212 patients was used to test the models. The model developed by the gradient boost machine (GBM) outperformed other models with an area under the ROC curve (AUC) of 0.937 in the validation set and an AUC of 0.945 in the test set. Furthermore, the GBM model achieved the highest sensitivity value of 0.583 among these AutoML models. The model developed by eXtreme Gradient Boosting (XGBoost) achieved the highest specificity value of 0.980 and the highest accuracy of 0.958 in the test set.

Conclusions

The AutoML model based on the GBM algorithm for early prediction of SAP showed evident clinical practicability.

Usefulness of Random Forest Algorithm in Predicting Severe Acute Pancreatitis

BACKGROUND AND AIMS

This study aimed to develop an interpretable random forest model for predicting severe acute pancreatitis (SAP).

METHODS

Clinical and laboratory data of 648 patients with acute pancreatitis were retrospectively reviewed and randomly assigned to the training set and test set in a 3:1 ratio. Univariate analysis was used to select candidate predictors for the SAP. Random forest (RF) and logistic regression (LR) models were developed on the training sample. The prediction models were then applied to the test sample. The performance of the risk models was measured by calculating the area under the receiver operating characteristic (ROC) curves (AUC) and area under precision recall curve. We provide visualized interpretation by using local interpretable model-agnostic explanations (LIME).

RESULTS

The LR model was developed to predict SAP as the following function: -1.10-0.13×albumin (g/L) + 0.016 × serum creatinine (μmol/L) + 0.14 × glucose (mmol/L) + 1.63 × pleural effusion (0/1)(No/Yes). The coefficients of this formula were utilized to build a nomogram. The RF model consists of 16 variables identified by univariate analysis. It was developed and validated by a tenfold cross-validation on the training sample. Variables importance analysis suggested that blood urea nitrogen, serum creatinine, albumin, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, calcium, and glucose were the most important seven predictors of SAP. The AUCs of RF model in tenfold cross-validation of the training set and the test set was 0.89 and 0.96, respectively. Both the area under precision recall curve and the diagnostic accuracy of the RF model were higher than that of both the LR model and the BISAP score. LIME plots were used to explain individualized prediction of the RF model.

CONCLUSIONS

An interpretable RF model exhibited the highest discriminatory performance in predicting SAP. Interpretation with LIME plots could be useful for individualized prediction in a clinical setting. A nomogram consisting of albumin, serum creatinine, glucose, and pleural effusion was useful for prediction of SAP.

EASY-APP: An artificial intelligence model and application for early and easy prediction of severity in acute pancreatitis

BACKGROUND

Acute pancreatitis (AP) is a potentially severe or even fatal inflammation of the pancreas. Early identification of patients at high risk for developing a severe course of the disease is crucial for preventing organ failure and death. Most of the former predictive scores require many parameters or at least 24 h to predict the severity; therefore, the early therapeutic window is often missed.

METHODS

The early achievable severity index (EASY) is a multicentre, multinational, prospective and observational study (ISRCTN10525246). The predictions were made using machine learning models. We used the scikit-learn, xgboost and catboost Python packages for modelling. We evaluated our models using fourfold cross-validation, and the receiver operating characteristic (ROC) curve, the area under the ROC curve (AUC), and accuracy metrics were calculated on the union of the test sets of the cross-validation. The most critical factors and their contribution to the prediction were identified using a modern tool of explainable artificial intelligence called SHapley Additive exPlanations (SHAP).

RESULTS

The prediction model was based on an international cohort of 1184 patients and a validation cohort of 3543 patients. The best performing model was an XGBoost classifier with an average AUC score of 0.81 ± 0.033 and an accuracy of 89.1%, and the model improved with experience. The six most influential features were the respiratory rate, body temperature, abdominal muscular reflex, gender, age and glucose level. Using the XGBoost machine learning algorithm for prediction, the SHAP values for the explanation and the bootstrapping method to estimate confidence, we developed a free and easy-to-use web application in the Streamlit Python-based framework (http://easy-app.org/).

CONCLUSIONS

The EASY prediction score is a practical tool for identifying patients at high risk for severe AP within hours of hospital admission. The web application is available for clinicians and contributes to the improvement of the model.

Predicting the Need for Therapeutic Intervention and Mortality in Acute Pancreatitis: A Two-Center International Study Using Machine Learning

Background: Current approaches to predicting intervention needs and mortality have reached 65–85% accuracy, which falls below clinical decision-making requirements in patients with acute pancreatitis (AP). We aimed to accurately predict therapeutic intervention needs and mortality on admission, in AP patients, using machine learning (ML). Methods: Data were obtained from three databases of patients admitted with AP: one retrospective (Chengdu) and two prospective (Liverpool and Chengdu) databases. Intervention and mortality differences, as well as potential predictors, were investigated. Univariate analysis was conducted, followed by a random forest ML algorithm used in multivariate analysis, to identify predictors. The ML performance matrix was applied to evaluate the model’s performance. Results: Three datasets of 2846 patients included 25 potential clinical predictors in the univariate analysis. The top ten identified predictors were obtained by ML models, for predicting interventions and mortality, from the training dataset. The prediction of interventions includes death in non-intervention patients, validated with high accuracy (96%/98%), the area under the receiver-operating-characteristic curve (0.90/0.98), and positive likelihood ratios (22.3/69.8), respectively. The post-test probabilities in the test set were 55.4% and 71.6%, respectively, which were considerably superior to existing prognostic scores. The ML model, for predicting mortality in intervention patients, performed better or equally with prognostic scores. Conclusions: ML, using admission clinical predictors, can accurately predict therapeutic interventions and mortality in patients with AP.

Growth differentiation factor 15 is an early predictor for persistent organ failure and mortality in acute pancreatitis

OBJECTIVES

Early prediction of persistent organ failure (POF) is crucial for patients with acute pancreatitis (AP). Growth differentiation factor 15 (GDF15), also known as macrophage inhibitory cytokine 1 (MIC-1), is associated with inflammatory responses. We investigated changes in plasma GDF15 and assessed its predictive value in AP.

METHODS

The study included 290 consecutive patients with AP admitted within 36 h after symptoms onset. Clinical data obtained during hospitalization were collected. Plasma GDF15 levels were determined using enzyme-linked immunosorbent assays. The predictive value of GDF15 for POF was analyzed.

RESULTS

There were 105 mild, 111 moderately severe, and 74 severe AP patients. Plasma GDF15 peak level were measured on admission, and significantly declined on the 3rd and 7th day. Admission GDF15 predicted POF and mortality with areas under the curve (AUC) of 0.847 (95% confidence interval [CI] 0.798-0.895) and 0.934 (95% CI 0.887-0.980), respectively. Admission GDF15, Bedside Index of Severity in Acute Pancreatitis, and hematocrit were independent factors for POF by univariate and multivariate logistic regression, and the nomogram built on these variables showed good performance (optimism-corrected c-statistic = 0.921). The combined predictive model increased the POF accuracy with an AUC 0.925 (95% CI 0.894-0.956), a net reclassification improvement of 0.3024 (95% CI: 0.1482-0.4565, P < 0.001), and an integrated discrimination index of 0.11 (95% CI 0.0497-0.1703; P < 0.001).

CONCLUSIONS

Plasma GDF15 measured within 48 h of symptom onset could help predict POF and mortality in AP patients.

Early prediction of severe acute pancreatitis using machine learning

BACKGROUND

Acute pancreatitis (AP) is one of the most common causes of gastrointestinal-related hospitalizations in the United States. Severe AP (SAP) is associated with a mortality rate of nearly 30% and is distinguished from milder forms of AP. Risk stratification to identify SAP cases needing inpatient treatment is an important aspect of AP diagnosis.

METHODS

We developed machine learning algorithms to predict which patients presenting with AP would require treatment for SAP. Three models were developed using logistic regression, neural networks, and XGBoost. Models were assessed in terms of area under the receiver operating characteristic (AUROC) and compared to the Harmless Acute Pancreatitis Score (HAPS) and Bedside Index for Severity in Acute Pancreatitis (BISAP) scores for AP risk stratification.

RESULTS

61,894 patients were used to train and test the machine learning models. With an AUROC value of 0.921, the model developed using XGBoost outperformed the logistic regression and neural network-based models. The XGBoost model also achieved a higher AUROC than both HAPS and BISAP for identifying patients who would be diagnosed with SAP.

CONCLUSIONS

Machine learning may be able to improve the accuracy of AP risk stratification methods and allow for more timely treatment and initiation of interventions.

Development an Inflammation-Related Factor-Based Model for Predicting Organ Failure in Acute Pancreatitis: A Retrospective Cohort Study

Several inflammation-related factors (IRFs) have been reported to predict organ failure of acute pancreatitis (AP) in previous clinical studies. However, there are a few shortcomings in these models. The aim of this study was to develop a new prediction model based on IRFs that could accurately identify the risk for organ failure in AP. Methods. 100 patients with their clinical information and IRF data (levels of 10 cytokines, percentages of different immune cells, and data obtained from white blood cell count) were retrospectively enrolled in this study, and 94 patients were finally selected for further analysis. Univariate and multivariate analysis were applied to evaluate the potential risk factors for the organ failure of AP. The area under the ROC curve (AUCs), sensitivity, and specificity of the relevant model were assessed to evaluate the prediction ability of IRFs. A new scoring system to predict the organ failure of AP was created based on the regression coefficient of a multivariate logistic regression model. Results. The incidence of OF in AP patients was nearly 16% (15/94) in our derivation cohort. Univariate analytic data revealed that IL6, IL8, IL10, MCP1, CD3+ CD4+ T lymphocytes, CD19+ B lymphocytes, PCT, APACHE II score, and RANSON score were potential predictors for AP organ failure, and IL6 (P = 0.038), IL8 (P = 0.043), and CD19+B lymphocytes (P = 0.045) were independent predictors according to further multivariate analysis. In addition, a preoperative scoring system (0-11 points) was constructed to predict the organ failure of AP using these three factors. The AUC of the new score system was 0.86. The optimal cut-off value of the new scoring system was 6 points. Conclusions. Our prediction model (based on IL6, IL8, and CD19+ B Lymphocyte) has satisfactory working efficiency to identify AP patients with high risk of organ failure.

Predicting postoperative surgical site infection with administrative data: a random forests algorithm

Background

Since primary data collection can be time-consuming and expensive, surgical site infections (SSIs) could ideally be monitored using routinely collected administrative data. We derived and internally validated efficient algorithms to identify SSIs within 30 days after surgery with health administrative data, using Machine Learning algorithms.

Methods

All patients enrolled in the National Surgical Quality Improvement Program from the Ottawa Hospital were linked to administrative datasets in Ontario, Canada. Machine Learning approaches, including a Random Forests algorithm and the high-performance logistic regression, were used to derive parsimonious models to predict SSI status. Finally, a risk score methodology was used to transform the final models into the risk score system. The SSI risk models were validated in the validation datasets.

Results

Of 14,351 patients, 795 (5.5%) had an SSI. First, separate predictive models were built for three distinct administrative datasets. The final model, including hospitalization diagnostic, physician diagnostic and procedure codes, demonstrated excellent discrimination (C statistics, 0.91, 95% CI, 0.90–0.92) and calibration (Hosmer-Lemeshow χ² statistics, 4.531, p = 0.402).

Conclusion

We demonstrated that health administrative data can be effectively used to identify SSIs. Machine learning algorithms have shown a high degree of accuracy in predicting postoperative SSIs and can integrate and utilize a large amount of administrative data. External validation of this model is required before it can be routinely used to identify SSIs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01369-9.

Predictors of Device-Related Thrombus Following Percutaneous Left Atrial Appendage Occlusion

BACKGROUND

Device-related thrombus (DRT) has been considered an Achilles' heel of left atrial appendage occlusion (LAAO). However, data on DRT prediction remain limited.

OBJECTIVES

This study constructed a DRT registry via a multicenter collaboration aimed to assess outcomes and predictors of DRT.

METHODS

Thirty-seven international centers contributed LAAO cases with and without DRT (device-matched and temporally related to the DRT cases). This study described the management patterns and mid-term outcomes of DRT and assessed patient and procedural predictors of DRT.

RESULTS

A total of 711 patients (237 with and 474 without DRT) were included. Follow-up duration was similar in the DRT and no-DRT groups, median 1.8 years (interquartile range: 0.9-3.0 years) versus 1.6 years (interquartile range: 1.0-2.9 years), respectively (P = 0.76). DRTs were detected between days 0 to 45, 45 to 180, 180 to 365, and >365 in 24.9%, 38.8%, 16.0%, and 20.3% of patients. DRT presence was associated with a higher risk of the composite endpoint of death, ischemic stroke, or systemic embolization (HR: 2.37; 95% CI, 1.58-3.56; P < 0.001) driven by ischemic stroke (HR: 3.49; 95% CI: 1.35-9.00; P = 0.01). At last known follow-up, 25.3% of patients had DRT. Discharge medications after LAAO did not have an impact on DRT. Multivariable analysis identified 5 DRT risk factors: hypercoagulability disorder (odds ratio [OR]: 17.50; 95% CI: 3.39-90.45), pericardial effusion (OR: 13.45; 95% CI: 1.46-123.52), renal insufficiency (OR: 4.02; 95% CI: 1.22-13.25), implantation depth >10 mm from the pulmonary vein limbus (OR: 2.41; 95% CI: 1.57-3.69), and non-paroxysmal atrial fibrillation (OR: 1.90; 95% CI: 1.22-2.97). Following conversion to risk factor points, patients with ≥2 risk points for DRT had a 2.1-fold increased risk of DRT compared with those without any risk factors.

CONCLUSIONS

DRT after LAAO is associated with ischemic events. Patient- and procedure-specific factors are associated with the risk of DRT and may aid in risk stratification of patients referred for LAAO.

The association of parameters of body composition and laboratory markers with the severity of hypertriglyceridemia-induced pancreatitis

BACKGROUND

Hypertriglyceridemia has arisen as the third leading cause of acute pancreatitis. This study aimed at exploring the association between the severity of hypertriglyceridemia-induced pancreatitis (HTGP) and computed tomography (CT)-based body composition parameters and laboratory markers.

METHODS

Laboratory and clinical parameters were collected from 242 patients with HTGP between 2017 and 2020. Severity of HTGP was evaluated by original or modified CT severity index. Body composition parameters such as area and radiodensity of muscle, subcutaneous adipose tissue and visceral adipose tissue were calculated by CT at the level of third lumbar vertebra. Parameters were compared between mild and moderately severe to severe HTGP. Uni-variate and multi-variate Logistic regression analyses were employed to assess the risk factors of the severity of HTGP.

RESULTS

Seventy patients (28.9%) presented with mild HTGP. Body mass index, waist circumference and all CT-based body composition parameters differed between male and female patients. None was associated with the severity of HTGP, neither in males nor in females. Receiver operating characteristic curves showed that areas under the curves of apolipoprotein A-I and albumin to predict the severity of HTGP were 0.786 and 0.759, respectively (all P < 0.001). Uni-variate and further multi-variate Logistic regression analysis confirmed that low serum albumin (< 35 g/L, P = 0.004, OR = 3.362, 95%CI = 1.492-8.823) and apolipoprotein A-I (< 1.1 g/L, P < 0.001, OR = 5.126, 95%CI = 2.348-11.195), as well as high C-reactive protein (> 90 mg/L, P = 0.005, OR = 3.061, 95%CI = 1.407-6.659) and lipase (P = 0.033, OR = 2.283, 95%CI = 1.070-4.873) were risk factors of moderately severe to severe HTGP. Levels of albumin, apolipoprotein A-I, C-reactive protein and lipase were also associated with the length of hospital stay (all P < 0.05). Besides, low serum albumin, low-density lipoprotein cholesterol and high radiodensity of subcutaneous adipose tissue were significant risk factors of pancreatic necrosis in patients with HTGP (all P < 0.05).

CONCLUSIONS

Low serum albumin and apolipoprotein A-I, and high C-reactive protein and lipase upon admission were associated with a more severe type of HTGP and longer hospital stay for these patients. Albumin and apolipoprotein A-I may serve as novel biomarkers for the severity of HTGP. However, none of the body composition parameters was associated with the severity of HTGP.

Clinical efficacy of duodenoscopy combined with laparoscopy in the treatment of patients with severe acute pancreatitis and pancreatic pseudocyst, and the effects on IL-6 and CRP

The study aimed to investigate the clinical efficacy of duodenoscopy combined with laparoscopy in the treatment of patients with severe acute pancreatitis (SAP) and pancreatic pseudocyst (PP), and its effects on serum inflammatory factors. Altogether 94 patients complicated with SAP and PP who were admitted to Weifang People's Hospital (Weifang, China) from September 2015 to December 2018 were included. Based on the different operation methods, 49 patients who underwent traditional laparotomy under laparoscopic surgery were included in group A, and 45 patients who underwent duodenoscopy and laparoscopy under duodenoscope to treat the drainage of nipple and pancreatic pseudocysts were included in group B. The expression levels of related serum indexes and serum stress indexes before and at 48 h after surgery, the postoperative nausea, vomiting and abdominal pain scores, as well as the clinical efficacy, perioperative related indexes, recovery and complications were compared between the two groups. The prognostic factors in both groups were assessed via Logistic univariate and multivariate analyses. C-reactive protein (CRP), tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), interleukin-β (IL-β), endotoxin and nuclear factor κB (NF-κB) were significantly lower in group B than those in group A (P<0.001). Upregulating cortisol and norepinephrine in group B was lower than that in group A (P<0.001). The total effective rate in group B was higher than that in group A (P<0.05). The perioperative related indexes, recovery, and postoperative complications in group B were better than those in group A (P<0.05). Scores of abdominal pain, nausea and vomiting in group B were markedly lower than those in group A (P<0.001). Multivariate Logistic regression analysis showed that CRP, TNF-α, IL-6, IL-β and surgical methods were independent risk factors for the prognosis of patients with SAP and PP. In conclusion, the combined treatment with duodenoscopy and laparoscopic surgery has little inflammatory and stress reaction, and it is highly safe, worthy to be popularized.

A novel severity score to predict inpatient mortality in COVID-19 patients

COVID-19 is commonly mild and self-limiting, but in a considerable portion of patients the disease is severe and fatal. Determining which patients are at high risk of severe illness or mortality is essential for appropriate clinical decision making. We propose a novel severity score specifically for COVID-19 to help predict disease severity and mortality. 4711 patients with confirmed SARS-CoV-2 infection were included. We derived a risk model using the first half of the cohort (n = 2355 patients) by logistic regression and bootstrapping methods. The discriminative power of the risk model was assessed by calculating the area under the receiver operating characteristic curves (AUC). The severity score was validated in a second half of 2356 patients. Mortality incidence was 26.4% in the derivation cohort and 22.4% in the validation cohort. A COVID-19 severity score ranging from 0 to 10, consisting of age, oxygen saturation, mean arterial pressure, blood urea nitrogen, C-Reactive protein, and the international normalized ratio was developed. A ROC curve analysis was performed in the derivation cohort achieved an AUC of 0.824 (95% CI 0.814-0.851) and an AUC of 0.798 (95% CI 0.789-0.818) in the validation cohort. Furthermore, based on the risk categorization the probability of mortality was 11.8%, 39% and 78% for patient with low (0-3), moderate (4-6) and high (7-10) COVID-19 severity score. This developed and validated novel COVID-19 severity score will aid physicians in predicting mortality during surge periods.

Sex differences in the relationship among alcohol, smoking, and Helicobacter pylori infection in asymptomatic individuals

OBJECTIVE

We aimed to investigate the relationship of Helicobacter pylori infection with alcohol and smoking.

METHODS

We conducted a cross-sectional study among participants who underwent health check-ups for H. pylori infection between January 2013 and March 2017. We subsequently investigated the relationship of H. pylori infection with alcohol and smoking.

RESULTS

A total of 7169 participants were enrolled in this study. The overall prevalence of H. pylori infection was 55.2%. Participants with H. pylori infection were more likely to be older than those without H. pylori infection. For male participants with H. pylori infection, multivariable logistic regression analysis indicated that both smoking (odds ratio (OR): 1.61; 95% confidence interval (CI): 1.41-1.83) and alcohol consumption (OR: 1.30; 95% CI: 1.10-1.52) were independently positively associated with H. pylori infection. For female participants, multivariable logistic regression analysis indicated that both smoking (OR: 0.03; 95% CI: 0.02-0.07) and alcohol consumption (OR: 0.20; 95% CI: 0.12-0.33) were inversely significantly associated with H. pylori infection after adjustment for age.

CONCLUSIONS

Smoking and alcohol consumption were risk factors for male participants but these were protective factors for female individuals with H. pylori infection.