Early prediction of acute necrotizing pancreatitis by artificial intelligence: a prospective cohort-analysis of 2387 cases.
Ontology highlight
ABSTRACT:
SUBMITTER: Kiss S
PROVIDER: S-EPMC9098474 | biostudies-literature | 2022 May
REPOSITORIES: biostudies-literature
Publications
Early prediction of acute necrotizing pancreatitis by artificial intelligence: a prospective cohort-analysis of 2387 cases.
Kiss Szabolcs S Pintér József J Molontay Roland R Nagy Marcell M Farkas Nelli N Sipos Zoltán Z Fehérvári Péter P Pecze László L Földi Mária M Vincze Áron Á Takács Tamás T Czakó László L Izbéki Ferenc F Halász Adrienn A Boros Eszter E Hamvas József J Varga Márta M Mickevicius Artautas A Faluhelyi Nándor N Farkas Orsolya O Váncsa Szilárd S Nagy Rita R Bunduc Stefania S Hegyi Péter Jenő PJ Márta Katalin K Borka Katalin K Doros Attila A Hosszúfalusi Nóra N Zubek László L Erőss Bálint B Molnár Zsolt Z Párniczky Andrea A Hegyi Péter P Szentesi Andrea A
Scientific reports 20220512 1
Pancreatic necrosis is a consistent prognostic factor in acute pancreatitis (AP). However, the clinical scores currently in use are either too complicated or require data that are unavailable on admission or lack sufficient predictive value. We therefore aimed to develop a tool to aid in necrosis prediction. The XGBoost machine learning algorithm processed data from 2387 patients with AP. The confidence of the model was estimated by a bootstrapping method and interpreted via the 10th and the 90t ...[more]
Similar Datasets
Project description:BackgroundAcute 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.MethodsThe 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).ResultsThe 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/).ConclusionsThe 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.
| S-EPMC9162438 | biostudies-literature
Project description:BackgroundAlthough several biomarkers exist for patients with heart failure (HF), their use in routine clinical practice is often constrained by high costs and limited availability.ObjectiveWe examined the utility of an artificial intelligence (AI) algorithm that analyzes printed electrocardiograms (ECGs) for outcome prediction in patients with acute HF.MethodsWe retrospectively analyzed prospectively collected data of patients with acute HF at two tertiary centers in Korea. Baseline ECGs were analyzed using a deep-learning system called Quantitative ECG (QCG), which was trained to detect several urgent clinical conditions, including shock, cardiac arrest, and reduced left ventricular ejection fraction (LVEF).ResultsAmong the 1254 patients enrolled, in-hospital cardiac death occurred in 53 (4.2%) patients, and the QCG score for critical events (QCG-Critical) was significantly higher in these patients than in survivors (mean 0.57, SD 0.23 vs mean 0.29, SD 0.20; P<.001). The QCG-Critical score was an independent predictor of in-hospital cardiac death after adjustment for age, sex, comorbidities, HF etiology/type, atrial fibrillation, and QRS widening (adjusted odds ratio [OR] 1.68, 95% CI 1.47-1.92 per 0.1 increase; P<.001), and remained a significant predictor after additional adjustments for echocardiographic LVEF and N-terminal prohormone of brain natriuretic peptide level (adjusted OR 1.59, 95% CI 1.36-1.87 per 0.1 increase; P<.001). During long-term follow-up, patients with higher QCG-Critical scores (>0.5) had higher mortality rates than those with low QCG-Critical scores (<0.25) (adjusted hazard ratio 2.69, 95% CI 2.14-3.38; P<.001).ConclusionsPredicting outcomes in patients with acute HF using the QCG-Critical score is feasible, indicating that this AI-based ECG score may be a novel biomarker for these patients.Trial registrationClinicalTrials.gov NCT01389843; https://clinicaltrials.gov/study/NCT01389843.
| S-EPMC11255523 | biostudies-literature
Project description:To assess the association between the clinical parameters within 48 hours of admission and the occurrence of infected pancreatic necrosis (IPN) during the late phase of necrotizing pancreatitis (NP).All patients were divided into 2 groups, the IPN and non-IPN groups. The clinical data were retrospectively analyzed. Univariate and multivariate logistic regression analyses were performed to evaluate the relationship between clinical parameters and IPN secondary to NP. The performance of each independent variable was plotted by the receiver-operating characteristic (ROC) curve. Consequently, the cut-off level of each independent variable with its sensitivity and specificity was calculated.A total of 215 patients were enrolled in our study. Among them, 87 (40.5%) patients developed IPNs after a median of 13.5 (9.5-23.0) days from admission. Multivariate analysis indicated that the level of hematocrit (HCT) from 40% to 50% (P=.012, odds ratio (OR) = 2.407), HCT over 50% (P < .009, OR = 6.794), blood urea nitrogen (BUN) (P = .040, OR = 1.894), C-reactive protein (CRP) (P = .043, OR = 1.837), and procalcitonin (PCT) (P = .002, OR = 2.559) were independent risk factors of IPN secondary to NP. The ROC cures revealed that the area under the ROC (AUC) of the maximum level of HCT, BUN, CRP, and PCT within 48 hours of admission was 0.687, 0.620, 0.630, and 0.674 respectively. Furthermore, the combination of these 4 individual parameters contributes to a more preferable AUC of 0.789 with a sensitivity of 67.8% and specificity of 77.3%.The maximum levels of PCT, CRP, HCT, and BUN within 48 hours of admission are independent factors of IPN and their combination might accurately predict the occurrence of IPN secondary to NP.
| S-EPMC5627815 | biostudies-literature
Project description:BackgroundAcute necrotizing pancreatitis (ANP) can result in a significant healthcare burden. The present study aimed to develop a new scoring system to accurately and promptly identify patients with a high likelihood of mortality to determine the need for aggressive measures.MethodsWe retrospectively analyzed patients diagnosed with ANP using the National Inpatient Sample (NIS). The mortality in ANP during admission (MANP-A) scoring system was derived using multivariate Cox regression analysis and validated using receiver operating characteristic (ROC) curves in a validation cohort.ResultsA total of 22,980 hospitalizations were identified in the derivation cohort. There was a predominance of males (65%) and white race (73%). Five variables showed significant association with mortality and were selected for developing the MANP-A scoring system: age ≥60 years; acute renal failure/kidney injury; sepsis with shock; vasopressor use; and disseminated intravascular coagulation. The MANP-A score has a maximum of 5 points and the cutoff for predicting mortality was set at 2 points. The area under the curve (AUC) using the ROC curve of the derivation cohort was 0.9195, 95% confidence interval [CI] 0.8838-0.9551 (P<0.001) for 7- and 0.8954, 95%CI 0.8723-0.9185 (P<0.001) for 30-day periods. The AUC of the Validation Cohort was 0.9204, 95%CI 0.8937-0.9469 (P<0.001) for 7- and 0.9059, 95%CI 0.8893-0.9223 (P<0.001) for 30-day periods.ConclusionWe propose a simple and objective score for predicting ANP inpatient mortality at 7- and 30-day intervals with high validity.
| S-EPMC9399582 | biostudies-literature
Project description:Infected pancreatic necrosis (IPN) is a significant complication of acute necrotizing pancreatitis (ANP). Early identification of patients at high risk of IPN would enable appropriate treatment, but there is a lack of valid tools. This study aimed to assess the performance of the Pancreatitis Activity Scoring System (PASS) and its modifications (by removing or reducing the weight of opioid usage) in predicting IPN in a cohort of predicted severe ANP patients. Data was prospectively collected in the TRACE trial (2017-2020) involving 16 sites across China. The predictive performance of PASS, modified PASS (mPASS), and conventional indices were assessed by the area under the receiver operating characteristic curve (AUC), Hosmer-Lemeshow Ĉ-test, Brier score, and Fagan's nomogram. Multivariate logistic regression analysis (MLRA) was used to define the relationship between the best-performing PASS/mPASS model and IPN. A total of 508 subjects were enrolled (median age, 43 years; 62.8% males) in the original trial, and 122 developed IPN (24%) within 90 days after randomization. Compared with non-IPN patients, the scores of PASS and its modified models were significantly higher in the IPN patients (all p < 0.001). Among the PASS and its modifications, mPASS-4 had the largest AUC, the lowest Brier score, and good calibration. The mPASS-4 model demonstrated an AUC of 0.752 in predicting IPN (the optimal cut-off for the mPASS-4 was 292.5) and outperformed the conventional indices. The MLRA results showed that mPASS-4 >292.5 was an independent risk factor of IPN (OR: 3.6, 95% CI: 2.1-6.3). The PASS and its modifications during the first week of ANP onset predict the development of IPN, with mPASS-4 performing best. The mPASS-4 model simplifies the original PASS, increasing the likelihood of clinical implementation.
| S-EPMC9892470 | biostudies-literature
Project description:BackgroundAcute pancreatitis (AP) is a potentially life-threatening condition characterized by inflammation of the pancreas, which can lead to complications such as pancreatic necrosis. The modified computed tomography severity index (MCTSI) is a widely used tool for assessing the severity of AP, particularly the extent of pancreatic necrosis. The accurate and timely assessment of the necrosis volume is crucial in guiding treatment decisions and improving patient outcomes. However, the current diagnostic process relies heavily on the manual interpretation of computed tomography (CT) scans, which can be subjective and prone to variability among clinicians. This study aimed to develop a deep-learning network model to assist clinicians in diagnosing the volume ratio of pancreatic necrosis based on the MCTSI for AP.MethodsThe datasets comprised retrospectively collected plain and contrast-enhanced CT scans from 144 patients (6 with scores of 0 points, 42 with scores of 2 points, and 65 with scores of 4 points) and the National Institutes of Health contrast-enhanced CT scans from 45 patients with scores of 0 points. An improved fully convolutional neural networks for volumetric medical image segmentation (V-Net) model was developed to segment the pancreatic volume (i.e., the whole pancreas, necrotic pancreatic tissue, and non-necrotic pancreatic tissue) and to quantify the split volume ratios. The improved strategy included three stages of body up- and down-sampling adapted to the task of segmentation in AP, and the selection of objects, loss function, and smoothing coefficients. The model interpretations were compared with those of clinicians with different levels of experience. The reference standard was manually segmented by a pancreatic radiologist. Accuracy, macro recall, and macro specificity were employed to compare the diagnostic efficacy of the model and the clinicians.ResultsIn total, 144 patients (mean age: 44±13 years; 40 females, 104 males) were included in the study. Optimal training results were obtained using the necrotic pancreatic tissue and whole pancreas as the input objects, and combining dice loss and 500 smoothing coefficients as the loss function for training. The dice coefficient for the whole pancreas was 0.811 and that for the necrotic pancreatic tissue was 0.761. The performance of the artificial intelligence model and clinicians were compared. The accuracy, macro recall, and macro specificity of the improved V-net were 0.854, 0.850 and 0.923, respectively, which were all significantly higher than those of the senior and junior clinicians (P<0.05).ConclusionsOur proposed model could improve the effectiveness of clinicians in diagnosing pancreatic necrosis volume ratios in clinical settings.
| S-EPMC11744103 | biostudies-literature
Project description:ObjectiveThe aim of this study was to analyse the clinical characteristics of acute pancreatitis (AP) in a prospectively collected, large, multicentre cohort and to validate the major recommendations in the IAP/APA evidence-based guidelines for the management of AP.DesignEighty-six different clinical parameters were collected using an electronic clinical research form designed by the Hungarian Pancreatic Study Group.Patients600 adult patients diagnosed with AP were prospectively enrolled from 17 Hungarian centres over a two-year period from 1 January 2013.Main resultsWith respect to aetiology, biliary and alcoholic pancreatitis represented the two most common forms of AP. The prevalence of biliary AP was higher in women, whereas alcoholic AP was more common in men. Hyperlipidaemia was a risk factor for severity, lack of serum enzyme elevation posed a risk for severe AP, and lack of abdominal pain at admission demonstrated a risk for mortality. Abdominal tenderness developed in all the patients with severe AP, while lack of abdominal tenderness was a favourable sign for mortality. Importantly, lung injury at admission was associated with mortality. With regard to laboratory parameters, white blood cell count and CRP were the two most sensitive indicators for severe AP. The most common local complication was peripancreatic fluid, whereas the most common distant organ failure in severe AP was lung injury. Deviation from the recommendations in the IAP/APA evidence-based guidelines on fluid replacement, enteral nutrition and timing of interventions increased severity and mortality.ConclusionsAnalysis of a large, nationwide, prospective cohort of AP cases allowed for the identification of important determinants of severity and mortality. Evidence-based guidelines should be observed rigorously to improve outcomes in AP.
| S-EPMC5087847 | biostudies-literature
Project description:IntroductionPain is the most common symptom in acute pancreatitis (AP) and is among the diagnostic criteria. Therefore, we aimed to characterize acute abdominal pain in AP.MethodsThe Hungarian Pancreatic Study Group prospectively collected multicentre clinical data on 1435 adult AP patients between 2012 and 2017. Pain was characterized by its intensity (mild or intense), duration prior to admission (hours), localization (nine regions of the abdomen) and type (sharp, dull or cramping).Results97.3% of patients (n = 1394) had pain on admission. Of the initial population with acute abdominal pain, 727 patients answered questions about pain intensity, 1148 about pain type, 1134 about pain localization and 1202 about pain duration. Pain was mostly intense (70%, n = 511/727), characterized by cramping (61%, n = 705/1148), mostly starting less than 24 h prior to admission (56.7%, n = 682/1202). Interestingly, 50.9% of the patients (n = 577/1134) had atypical pain, which means pain other than epigastric or belt-like upper abdominal pain. We observed a higher proportion of peripancreatic fluid collection (19.5% vs. 11.0%; p = 0.009) and oedematous pancreas (8.4% vs. 3.1%; p = 0.016) with intense pain. Sharp pain was associated with AP severity (OR = 2.481 95% CI: 1.550-3.969) and increased mortality (OR = 2.263, 95% CI: 1.199-4.059) compared to other types. Longstanding pain (>72 h) on admission was not associated with outcomes. Pain characteristics showed little association with the patient's baseline characteristics.ConclusionA comprehensive patient interview should include questions about pain characteristics, including pain type. Patients with sharp and intense pain might need special monitoring and tailored pain management.SignificanceAcute abdominal pain is the leading presenting symptom in acute pancreatitis; however, we currently lack specific guidelines for pain assessment and management. In our cohort analysis, intense and sharp pain on admission was associated with higher odds for severe AP and several systemic and local complications. Therefore, a comprehensive patient interview should include questions about pain characteristics and patients with intense and sharp pain might need closer monitoring.
| S-EPMC9299627 | biostudies-literature
Project description:Necrosis of pancreatic parenchyma or extrapancreatic tissues is present in 10%-20% of patients with acute pancreatitis, defining the necrotizing presentation frequently associated with high morbidity and mortality rates. During the initial phase of acute necrotizing pancreatitis the most important pillars of medical treatment are fluid resuscitation, early enteral nutrition, endoscopic retrograde colangiopancreatography if associated cholangitis and intensive care unit support. When infection of pancreatic or extrapancreatic necrosis occurs, surgical approach constitutes the most accepted therapeutic option. In this context, we have recently assited to changes in time for surgery (delaying the indication if possible to around 4 wk to deal with "walled-off" necrosis) and type of access for necrosectomy: from a classical open approach (with closure over large-bore drains for continued postoperative lavage or semiopen techniques with scheduled relaparotomies), trends have changed to a "step-up" philosophy with initial percutaneous drainage and posterior minimally invasive or endoscopic access to the retroperitoneal cavity for necrosectomy if no improvement has been previously achieved. These approaches are progressively gaining popularity and morbidity and mortality rates have decreased significantly. Therefore, a staged, multidisciplinary, step-up approach with minimally invasive or endoscopic access for necrosectomy is widely accepted nowadays for management of pancreatic necrosis.
| S-EPMC4266831 | biostudies-literature
Project description:BackgroundAbdominal free fluid is frequently encountered on cross-sectional imaging for acute pancreatitis and may be a sign of increased severity and complications. This study examines the ability of free fluid to predict necrotizing pancreatitis and other adverse outcomes.MethodsWe conducted a single-center retrospective study of patients with acute pancreatitis and multiple cross-sectional imaging studies. Patients were divided into those who demonstrated free fluid on initial imaging and those without free fluid. The primary outcome was developing necrotizing pancreatitis. Logistic regression analysis assessed the performance of several predictors.ResultsA total of 245 acute pancreatitis patients were included. Pancreatic necrosis occurred more frequently in the free fluid group (31.3 vs. 1.3%, P<0.001). The free fluid group also had higher rates of transient organ failure (17.7 vs. 3.4%, P<0.001), persistent organ failure (17.7 vs. 2.0%, P<0.001), in-hospital mortality (7.3 vs. 1.3%, P=0.016), length of stay (16.2 vs. 5.5 days, P<0.001), and intensive care unit admission (30.2 vs. 4.7%, P<0.001). On multivariate logistic regression, free fluid was the strongest predictor (adjusted odds ratio 17.11, 95% confidence interval 3.68-79.65; P<0.001) for necrotizing pancreatitis, with an excellent performance (area under the curve 0.92). When neither fluid on initial imaging nor persistent systemic inflammatory response syndrome was present, the negative predictive value for developing pancreatic necrosis was 100%.ConclusionsFree fluid in acute pancreatitis is a strong predictor for necrotizing pancreatitis, organ failure and mortality, and outperformed current predictors. Patients who lacked both free fluid on imaging and persistent systemic inflammatory response syndrome are at low risk for adverse outcomes and may be considered for early discharge.
| S-EPMC8596223 | biostudies-literature