Meyfroidt, Geert

Purpose of review Acute kidney injury (AKI) frequently complicates hospital admission, especially in the ICU or after major surgery, and is associated with high morbidity and mortality. The risk of developing AKI depends on the presence of preexisting comorbidities and the cause of the current disease. Besides, many other parameters affect the kidney function, such as the state of other vital organs, the host response, and the initiated treatment. Advancements in the field of informatics have led to the opportunity to store and utilize the patient-related data to train and validate models to detect specific patterns and, as such, predict disease states or outcomes. Recent findings Machine-learning techniques have also been applied to predict AKI, as well as the patients’ outcomes related to their AKI, such as mortality or the need for kidney replacement therapy. Several models have recently been developed, but only a few of them have been validated in external cohorts. Summary In this article, we provide an overview of the machine-learning prediction models for AKI and its outcomes in critically ill patients and individuals undergoing major surgery. We also discuss the pitfalls and the opportunities related to the implementation of these models in clinical practices.

Recently, during critical illness, cortisol metabolism was found to be reduced. We hypothesize that such reduced cortisol breakdown may suppress pulsatile ACTH and cortisol secretion via feedback inhibition. To test this hypothesis, nocturnal ACTH and cortisol secretory profiles were constructed by deconvolution analysis from plasma concentration time series in 40 matched critically ill patients and eight healthy controls, excluding diseases or drugs that affect the hypothalamic-pituitary-adrenal axis. Blood was sampled every 10 min between 2100 and 0600 to quantify plasma concentrations of ACTH and (free) cortisol. Approximate entropy, an estimation of process irregularity, cross-approximate entropy, a measure of ACTH-cortisol asynchrony, and ACTH-cortisol dose-response relationships were calculated. Total and free plasma cortisol concentrations were higher at all times in patients than in controls (all P < 0.04). Pulsatile cortisol secretion was 54% lower in patients than in controls ( P = 0.005), explained by reduced cortisol burst mass ( P = 0.03), whereas cortisol pulse frequency ( P = 0.35) and nonpulsatile cortisol secretion ( P = 0.80) were unaltered. Pulsatile ACTH secretion was 31% lower in patients than in controls ( P = 0.03), again explained by a lower ACTH burst mass ( P = 0.02), whereas ACTH pulse frequency ( P = 0.50) and nonpulsatile ACTH secretion ( P = 0.80) were unchanged. ACTH-cortisol dose response estimates were similar in patients and controls. ACTH and cortisol approximate entropy were higher in patients ( P ≤ 0.03), as was ACTH-cortisol cross-approximate entropy ( P ≤ 0.001). We conclude that hypercortisolism during critical illness coincided with suppressed pulsatile ACTH and cortisol secretion and a normal ACTH-cortisol dose response. Increased irregularity and asynchrony of the ACTH and cortisol time series supported non-ACTH-dependent mechanisms driving hypercortisolism during critical illness.