Search

Frequent assessment of the severity of illness for hospitalized patients is essential in clinical settings to prevent outcomes such as in-hospital mortality and unplanned ICU admission. Classical severity scores have been developed typically using relatively few patient features, especially for intensive care. Recently, deep learning-based models demonstrated better individualized risk assessments compared to classic risk scores such as SOFA and NEWS, thanks to the use of aggregated and more heterogeneous data sources for dynamic risk prediction. We investigated to what extent deep learning methods can capture patterns of longitudinal change in health status using time-stamped data from electronic health records. We used medical history data, biochemical measurements, and the clinical notes from all patients admitted to non-intensive care units in 12 hospitals in Denmarks Capital Region and Region Zealand during 2011-2016. Data from a total of 852,620 patients and 2,241,849 admissions were used to predict the composite outcome of unplanned ICU transfer and in-hospital death at different time points after admission to general departments. We subsequently examined feature interpretations of the models. The best model used all data modalities with an assessment rate of 6 hours and a prediction window of 14 days, with an AUPRC of 0.287 and AUROC of 0.898. These performances are comparable to the current state of the art and make the model suitable for further prospective validation as a risk assessment tool in a clinical setting.

Background. Patients admitted to the intensive care unit often have low serum levels of magnesium, phosphate, and zinc. We aim to describe the prevalence, management, characteristics, and outcomes of patients with low levels of these minerals in a large cohort of Danish ICU patients. Methods. In this multicenter cohort, we will include all adult ICU patients admitted to one of 10 ICUs during a six years period. Patients will be excluded if the ICU admission is post-operative after planned surgery, length of stay is < 24 hours, they are transferred from ICUs not participating in the study, or are without measurements of neither magnesium, phosphate nor zinc. First, we will do a descriptive assessment of the prevalence of low serum levels and management of supplementation according to different patient characteristics. Second, we will conduct a cohort study using prospectively registered outcome data, to assess associations between serum levels of magnesium, phosphate, and zinc respectively, and each of two principal outcomes: time to successful extubation and incident tachyarrhythmia. We will use a joint modelling approach for estimating the strengths of associations, combining time to event data and longitudinal data from repeated measurements of serum levels. Discussion. This study will provide information from a very large cohort of Danish ICU patients. Strengths of the study will be the large number of patients, the interconnection of data from several registers, and for the association analysis the use of a statistical approach making the optimal use of the repeated measures of serum levels. Limitations will be the risk of bias from unmeasured or unregistered confounding factors, and the lack of randomization in the observational design, which will prevent us from interpreting possible associations between serum levels and outcomes as definite effects on the outcome.