Predicting the risk of acute kidney injury in patients with acute pancreatitis complicated by sepsis using a stacked ensemble machine learning model: a retrospective study based on the MIMIC database

Objective
This study developed and validated a stacked ensemble machine learning model to predict the risk of acute kidney injury in patients with acute pancreatitis complicated by sepsis.

Design
A retrospective study based on patient data from public databases.

Participants
This study analysed 1295 patients with acute pancreatitis complicated by septicaemia from the US Intensive Care Database.

Methods
From the MIMIC database, data of patients with acute pancreatitis and sepsis were obtained to construct machine learning models, which were internally and externally validated. The Boruta algorithm was used to select variables. Then, eight machine learning algorithms were used to construct prediction models for acute kidney injury (AKI) occurrence in intensive care unit (ICU) patients. A new stacked ensemble model was developed using the Stacking ensemble method. Model evaluation was performed using area under the receiver operating characteristic curve (AUC), precision-recall (PR) curve, accuracy, recall and F1 score. The Shapley additive explanation (SHAP) method was used to explain the models.

Main outcome measures
AKI in patients with acute pancreatitis complicated by sepsis.

Results
The final study included 1295 patients with acute pancreatitis complicated by sepsis, among whom 893 cases (68.9%) developed acute kidney injury. We established eight base models, including Logit, SVM, CatBoost, RF, XGBoost, LightGBM, AdaBoost and MLP, as well as a stacked ensemble model called Multimodel. Among all models, Multimodel had an AUC value of 0.853 (95% CI: 0.792 to 0.896) in the internal validation dataset and 0.802 (95% CI: 0.732 to 0.861) in the external validation dataset. This model demonstrated the best predictive performance in terms of discrimination and clinical application.

Conclusion
The stack ensemble model developed by us achieved AUC values of 0.853 and 0.802 in internal and external validation cohorts respectively and also demonstrated excellent performance in other metrics. It serves as a reliable tool for predicting AKI in patients with acute pancreatitis complicated by sepsis.

Leggi
Febbraio 2025

Association between adjuvant radiotherapy in adults with gastric cancer and risk of second primary malignancy: a retrospective cohort study using the Surveillance, Epidemiology and End Results database

Objectives
This study aims to assess the association between adjuvant radiotherapy and the development of second primary malignancies (SPMs) and identify its determinants in patients who have undergone surgical treatment for gastric cancer.

Design
Retrospective cohort study using the Surveillance, Epidemiology and End Results (SEER) database.

Setting
Cohorts (18 registries, 2000–2018, from SEER) were screened for any malignancy that developed after sufficient latency from diagnosis of surgically treated non-metastatic gastric cancer.

Participants
24 777 surgically treated gastric cancer cases were included in the cohort. Among them, 6128 patients underwent adjuvant radiotherapy.

Outcome measures
The cumulative incidence of SPMs was estimated using Fine and Gray’s competing risk model and the radiotherapy-correlated risks were calculated using Poisson regression analysis.

Results
Among patients with sufficient latency, there was no significant association between radiotherapy and the risk of developing second primary solid malignancies (relative risk=1.05, 95% CI 0.83 to 1.33) or haematological malignancies (relative risk=1.17, 95% CI 0.62 to 2.11). Interestingly, radiotherapy was associated with a reduced cumulative incidence of second lung and bronchus cancer compared with no radiotherapy, with a 15-year incidence of 1.4%–3.17% (p

Leggi
Febbraio 2025

Examining the relationship between incidence and mortality for commonly diagnosed cancers in the USA: an observational study using population-based SEER database

Objective
Incidence and mortality are fundamental epidemiologic measures of cancer burden, yet few studies have examined individual cancers to determine how these measures correlate across place. We assessed the relationship between incidence and mortality for commonly diagnosed cancers in the USA.

Design
Population-based observational study of US counties.

Setting and participants
The Surveillance, Epidemiology and End Results (SEER) database was used to obtain incidence (2000–2016) and mortality (2002–2018) data for the 12 most commonly diagnosed non-haematologic cancers.

Outcome measures
County-level correlation between cancer incidence and mortality. Cancers were grouped into terciles based on the population-weighted correlation coefficient (r). We also examined the 10 year risk of death, both from the diagnosed cancer and other causes.

Results
County-level incidence and mortality were strongly correlated in some cancers, yet uncorrelated in others. Cancers in the high-correlation tercile (r range: 0.96 to 0.78) included lung, stomach, liver and pancreas. For patients with these cancers, the risk of death from the diagnosed cancer was >4-times the risk of death from other causes. The moderate-correlation tercile (r: 0.75 to 0.58) included cancers of the colon, bladder, kidney and uterus. There was little or no relationship between incidence and mortality for cancers in the low-correlation tercile (r: 0.33 to –0.10): melanoma, prostate, breast and thyroid. The risk of death from the diagnosed cancer for these patients was either lower or no different than their risk of death from other causes.

Conclusions
For some cancers in the USA, the fundamental epidemiologic measure of disease frequency—incidence—now has little relationship with cancer death (mortality). Low correlations are most likely explained by differences in diagnostic practice leading to variable amounts of cancer overdiagnosis between different US counties.

Leggi
Febbraio 2025