OBJECTIVES: This study aims to improve the early detection of cardiac surgery-associated acute kidney injury using artificial intelligence-based algorithms. METHODS: Data from consecutive patients undergoing cardiac surgery between 2008 and 2018 in our institution served as the source for artificial intelligence-based modelling. Cardiac surgery-associated acute kidney injury was defined according to the Kidney Disease Improving Global Outcomes criteria. Different machine learning algorithms were trained and validated to detect cardiac surgery-associated acute kidney injury within 12 h after surgery. Demographic characteristics, comorbidities, preoperative cardiac status and intra- and postoperative variables including creatinine and haemoglobin values were retrieved for analysis. RESULTS: From 7507 patients analysed, 1699 patients (22.6%) developed cardiac surgery-associated acute kidney injury. The ultimate detection model, ‘Detect-A(K)I’, recognizes cardiac surgery-associated acute kidney injury within 12 h with an area under the curve of 88.0%, sensitivity of 78.0%, specificity of 78.9% and accuracy of 82.1%. The optimal parameter set includes serial changes of creatinine and haemoglobin, operative emergency, bleeding-associated variables, cardiac ischaemic time and cardiac function-associated variables, age, diuretics and active infection, chronic obstructive lung and peripheral vascular disease. CONCLUSIONS: The ‘Detect-A(K)I’ model successfully detects cardiac surgery-associated acute kidney injury within 12 h after surgery with the best discriminatory characteristics reported so far.

Background: Cardiac surgery in patients with infective endocarditis (IE) is still associated with high mortality and morbidity; an already present inflammation might further be aggravated due to a cardiopulmonary bypass-induced dysregulated immune response. Intraoperative hemoadsorption therapy may attenuate this septic response. Our objective was therefore to assess the efficacy of intraoperative hemoadsorption in active left-sided native- and prosthetic infective endocarditis. Methods: Consecutive high-risk patients with active left-sided infective endocarditis were enrolled between January 2015 and April 2021. Patients with intraoperative hemoadsorption (Cytosorbents, Princeton, NJ, USA) were compared to patients without hemoadsorption (control). Endpoints were the incidence of postoperative sepsis, sepsis-associated death and in-hospital mortality. Predictors for sepsis-associated mortality and in-hospital mortality were analysed by multivariable logistic regression. Results: A total of 202 patients were included, 135 with active left-sided native and 67 with prosthetic valve infective endocarditis. Ninety-nine patients received intraoperative hemoadsorption and 103 patients did not. Ninety-nine propensity-matched pairs were selected for final analyses. Postoperative sepsis and sepsis-related mortality was reduced in the hemoadsorption group (22.2% vs. 39.4%, p = 0.014 and 8.1% vs. 22.2%, p = 0.01, respectively). In-hospital mortality tended to be lower in the hemoadsorption group (14.1% vs. 26.3%, p = 0.052). Key predictors for sepsis-associated mortality and in-hospital mortality were preoperative inotropic support, lactate-levels 24 h after surgery, C-reactive protein levels on postoperative day 1, chest tube output, cumulative inotropes and white blood cell counts on postoperative day 2, and new onset of dialysis. Multivariate regression analysis revealed intraoperative hemoadsorption to be associated with lower sepsis-associated (OR 0.09, 95% CI 0.013–0.62, p = 0.014) as well as in-hospital mortality (OR 0.069, 95% CI 0.006–0.795, p = 0.032). Conclusions: Intraoperative hemoadsorption holds promise to reduce sepsis and sepsis-associated mortality after cardiac surgery for active left-sided native and prosthetic valve infective endocarditis.

In the absence of a systematic approach to epidemiological modeling in Slovenia, various isolated mathematical epidemiological models emerged shortly after the outbreak of the COVID-19 epidemic. We present an epidemiological model adapted to the COVID-19 situation in Slovenia. The standard SEIR model was extended to distinguish between age groups, symptomatic or asymptomatic disease progression, and vaccinated or unvaccinated populations. Evaluation of the model forecasts for 2021 showed the expected behavior of epidemiological modeling: our model adequately predicts the situation up to 4 weeks in advance; the changes in epidemiologic dynamics due to the emergence of a new viral variant in the population or the introduction of new interventions cannot be predicted by the model, but when the new situation is incorporated into the model, the forecasts are again reliable. Comparison with ensemble forecasts for 2022 within the European Covid-19 Forecast Hub showed better performance of our model, which can be explained by a model architecture better adapted to the situation in Slovenia, in particular a refined structure for vaccination, and better parameter tuning enabled by the more comprehensive data for Slovenia. Our model proved to be flexible, agile, and, despite the limitations of its compartmental structure, heterogeneous enough to provide reasonable and prompt short-term forecasts and possible scenarios for various public health strategies. The model has been fully operational on a daily basis since April 2020, served as one of the models for decision-making during the COVID-19 epidemic in Slovenia, and is part of the European Covid-19 Forecast Hub.

An edit-distance model that can be used for the approximate matching of contiguous and noncontiguous timed strings is presented. The model extends the concept of the weighted string-edit distance by introducing timed edit operations and by making the edit costs time dependent. Special attention is paid to the timed null symbols that are associated with the timed insertions and deletions. The usefulness of the presented model is demonstrated on the classification of phone-recognition errors using the TIMIT speech database.