Dataset Information

A novel explainable online calculator for contrast-induced AKI in diabetics: a multi-centre validation and prospective evaluation study.

ABSTRACT:

Background

In patients undergoing percutaneous coronary intervention (PCI), contrast-induced acute kidney injury (CIAKI) is a frequent complication, especially in diabetics, and is connected with severe mortality and morbidity in the short and long term. Therefore, we aimed to develop a CIAKI predictive model for diabetic patients.

Methods

3514 patients with diabetes from four hospitals were separated into three cohorts: training, internal validation, and external validation. We developed six machine learning (ML) algorithms models: random forest (RF), gradient-boosted decision trees (GBDT), logistic regression (LR), least absolute shrinkage and selection operator with LR, extreme gradient boosting trees (XGBT), and support vector machine (SVM). The area under the receiver operating characteristic curve (AUC) of ML models was compared to the prior score model, and developed a brief CIAKI prediction model for diabetes (BCPMD). We also validated BCPMD model on the prospective cohort of 172 patients from one of the hospitals. To explain the prediction model, the shapley additive explanations (SHAP) approach was used.

Results

In the six ML models, XGBT performed best in the cohort of internal (AUC: 0.816 (95% CI 0.777-0.853)) and external validation (AUC: 0.816 (95% CI 0.770-0.861)), and we determined the top 15 important predictors in XGBT model as BCPMD model variables. The features of BCPMD included acute coronary syndromes (ACS), urine protein level, diuretics, left ventricular ejection fraction (LVEF) (%), hemoglobin (g/L), congestive heart failure (CHF), stable Angina, uric acid (umol/L), preoperative diastolic blood pressure (DBP) (mmHg), contrast volumes (mL), albumin (g/L), baseline creatinine (umol/L), vessels of coronary artery disease, glucose (mmol/L) and diabetes history (yrs). Then, we validated BCPMD in the cohort of internal validation (AUC: 0.819 (95% CI 0.783-0.855)), the cohort of external validation (AUC: 0.805 (95% CI 0.755-0.850)) and the cohort of prospective validation (AUC: 0.801 (95% CI 0.688-0.887)). SHAP was constructed to provide personalized interpretation for each patient. Our model also has been developed into an online web risk calculator. MissForest was used to handle the missing values of the calculator.

Conclusion

We developed a novel risk calculator for CIAKI in diabetes based on the ML model, which can help clinicians achieve real-time prediction and explainable clinical decisions.

SUBMITTER: Ma M

PROVIDER: S-EPMC10391987 | biostudies-literature | 2023 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A novel explainable online calculator for contrast-induced AKI in diabetics: a multi-centre validation and prospective evaluation study.

Ma Mengqing M Wan Xin X Chen Yuyang Y Lu Zhichao Z Guo Danning D Kong Huiping H Pan Binbin B Zhang Hao H Chen Dawei D Xu Dongxu D Sun Dong D Lang Hong H Zhou Changgao C Li Tao T Cao Changchun C

Journal of translational medicine 20230731 1

<h4>Background</h4>In patients undergoing percutaneous coronary intervention (PCI), contrast-induced acute kidney injury (CIAKI) is a frequent complication, especially in diabetics, and is connected with severe mortality and morbidity in the short and long term. Therefore, we aimed to develop a CIAKI predictive model for diabetic patients.<h4>Methods</h4>3514 patients with diabetes from four hospitals were separated into three cohorts: training, internal validation, and external validation. We d ...[more]

PMID: 37525240

Similar Datasets

Project description:BackgroundThe Iowa Rectal Surgery Risk Calculator estimates risk for proctectomy procedures. The Iowa Calculator performed well on NSQIP 2010-2011 training and 2005-2009 validation datasets, but was not prospectively validated and did not include low anterior resections. This study sought to demonstrate validity on new independent data, to update the calculator to include low anterior resection, and to compare performance to other risk assessment tools.MethodsNon-emergent ACS-NSQIP proctectomy and low anterior resection data from 2010 to 2015 (n = 65,683) were included. The Iowa Calculator generated risk estimates for 30-day morbidity using 2012-2015 data. An Updated Calculator used 2010-2011 training data to include low anterior resection, with validation on 2012-2015 data. NSQIP data provided NSQIP Morbidity Model predictions and a custom web-script collected ACS-NSQIP Online Surgical Risk Calculator predictions for all patients.ResultsProctectomy morbidity (not including low anterior resection) decreased from 40.4% in 2010-2011 to 37.0% in 2012-2015. Low anterior resection had lower morbidity (22.4% in 2012-15). The Iowa Calculator demonstrated good discrimination and calibration using 2012-2015 data (C-statistic 0.676, deviance + 9.2%). After including low anterior resection, the Updated Iowa Calculator performed well during training (c-statistic 0.696, deviance 0%) and validation (C-statistic 0.706, deviance + 7.9%). The Updated Iowa Calculator had significantly better discrimination and calibration than morbidity predictions from the ACS Online Calculator (C-statistic 0.693, P < 0.001, deviance - 28.1%) and NSQIP General/Vascular Surgery Model (C-statistic 0.703, P < 0.05, deviance - 40.8%).ConclusionWhen applied to new independent data, the Iowa Calculator supplies accurate risk estimates. The Updated Iowa Calculator includes low anterior resection, and both are prospectively validated. Risk estimation by the Iowa Calculators was superior to ACS-provided risk tools.

Project description:BackgroundIdiopathic pulmonary fibrosis (IPF) is characterized by a poor prognosis, with a progressive decline in lung function and considerable variability in the disease's natural history. Besides lung transplantation (LTx), the only available treatments are anti-fibrosing drugs, which have shown to slow down the disease course. Therefore, predicting the prognosis is of pivotal importance to avoid treatment delays, which may be fatal for patients with a high risk of progression. Previous studies showed that a multi-dimensional approach is practical and effective in the development of a reliable prognostic score for IPF. In the RIsk Stratification scorE (RISE), physiological parameters, an objective measure of patient-reported dyspnea and exercise capacity are combined to capture different domains of the complex pathophysiology of IPF.MethodsThis is an observational, multi-centre, prospective cohort study, designed to reflect common clinical practice in IPF. A development cohort and a validation cohort will be included. Patients newly diagnosed with IPF based on the ATS/ERS criteria and multi-disciplinary discussion will be included in the study. A panel of chest radiologists and lung pathologists will further assess eligibility. At the first visit (time of diagnosis), and every 4-months, MRC dyspnea score, pulmonary function tests (FEV1, FVC and DLCO), and 6-min walking distance will be recorded. Patients will be prospectively followed for 3 years. Comorbidities will be considered. The radiographic extent of fibrosis on HRCT will be recalculated at a 2-year interval. RISE, Gender-Age-Physiology, CPI and Mortality Risk Scoring System will be calculated at 4-month intervals. Longitudinal changes of each variable considered will be assessed. The primary endpoint is 3-year LTx-free survival from the time of diagnosis. Secondary endpoints include several, clinically-relevant information to ensure reproducibility of results across a wide range of disease severity and in concomitance of associated pulmonary hypertension or emphysema.DiscussionThe objective of this study is to validate RISE as a simple, straightforward, inexpensive and reproducible tool to guide clinical decision making in IPF, and potentially as an endpoint for future clinical trials.Trial registrationU.S National Library of Medicine Clinicaltrials.gov, trial n. NCT02632123 "Validation of the risk stratification score in idiopathic pulmonary fibrosis". Date of registration: December 16th, 2015.

Project description:BackgroundDue to methodological shortcomings the available post-registration data on the adverse events (AEs) occurring in interferon beta-1a (INFb-1a)-treated patients fail to adequately validate phase III data and only partially inform on safety in daily practice. We assessed AEs in relapsing remitting multiple sclerosis (RRMS) patients treated with intramuscular (IM) INFb-1a in daily practice using data quality assurance measures similar to those in phase III trials.MethodsA prospective, International Conference on Harmonization (ICH) - Good Clinical Practice (GCP)-based, clinical research organization (CRO)-supported study in 36 practices in the Netherlands, Belgium, the United Kingdom and Luxembourg. During 24 months after start of IM INFb-1a treatment 275 RRMS patients were assessed for AEs' severity (mild, moderate, severe) and relationship to treatment (not, unlikely, likely, definite). Data were compared with those reported in the pivotal phase III trial.Findings75.3% of the patients experienced one or more AEs that were likely or definitely related to INFb-1a. Of all AEs 40.5% were likely or definitely treatment-related; 68.5% of these were mild, and 3% severe. 6.6% of the patients discontinued treatment because of an AE. Compared to the pivotal phase III trial, we found statistically significantly lower incidences for most of the common AEs: headache, muscle ache, fatigue, fever, chills, nausea. One patient died following two cerebral vascular events in study month 22, both AEs were assessed as not related to INFb-1a.ConclusionThree out of four RRMS patients treated with IM INFb-1a in daily practice experience treatment-related AEs, most of these being mild. Our data externally validate the favorable phase III safety profile of IM INFb-1a and suggest that the real-life incidence of treatment-related AEs is less than reported in the pivotal phase III trial. Larger studies are needed to detect rare, potentially hazardous AEs of IM INFb-1a.

Project description:ObjectivesTo describe (1) infant feeding practices during initial hospitalisation and up to 6 months corrected age (CA) in infants born late preterm with mothers intending to breastfeed, (2) the impact of early feeding practices on hospital length of stay and (3) maternal and infant factors associated with duration of breastfeeding.MethodsWe conducted a prospective cohort study of infants born at 34+0 to 36+6 weeks gestational age during 2018-2020. Families were followed up until the infant reached 6 months of age (corrected for prematurity). Feeding practices during the birth hospitalisation, length of initial hospital stay, and the prevalence of exclusive or any breastfeeding at 6 weeks, 3 months, and 6 months CA were examined. Associations between maternal and infant characteristics and breastfeeding at 6 weeks, 3 months and 6 months CA were assessed using multivariable logistic regression models.Results270 infants were enrolled, of these, 30% were multiple births. Overall, 78% of infants received only breastmilk as their first feed, and 83% received formula during the hospitalisation. Seventy-four per cent of infants were exclusively breastfed at discharge, 41% at 6 weeks CA, 35% at 3 months CA, and 29% at 6 months CA. The corresponding combined exclusive and partial breastfeeding rates (any breastfeeding) were 72%, 64%, and 53% of babies at 6 weeks CA, 3 months CA, and 6 months CA, respectively. The mean duration of hospitalisation was 2.9 days longer (95% confidence interval (CI) 0.31, 5.43 days) in infants who received any formula compared with those receiving only breastmilk (adjusted for GA, maternal age, multiple birth, site, and neonatal intensive care unit admission). In multivariable models, receipt of formula as the first milk feed was associated with a reduction in exclusive breastfeeding at 6 weeks CA (odds ratio = 0.22; 95% CI 0.09 to 0.53) and intention to breastfeed >6 months with an increase (odds ratio = 4.98; 95% CI 2.39 to 10.40). Intention to breastfeed >6 months remained an important predictor of exclusive breastfeeding at 3 and 6 months CA.ConclusionsOur study demonstrates that long-term exclusive breastfeeding rates were low in a cohort of women intending to provide breastmilk to their late preterm infants, with approximately half providing any breastmilk at 6 months CA. Formula as the first milk feed and intention to breastfeed >6 months were significant predictors of breastfeeding duration. Improving breastfeeding outcomes may require strategies to support early lactation and a better understanding of the ongoing support needs of this population.

Dataset Information

A novel explainable online calculator for contrast-induced AKI in diabetics: a multi-centre validation and prospective evaluation study.

Background

Methods

Results

Conclusion

Publications

A novel explainable online calculator for contrast-induced AKI in diabetics: a multi-centre validation and prospective evaluation study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets