Introduction
There is evidence that SARS-CoV2 has a particular affinity for kidney tissue and is often
associated with kidney failure.
Methods
We assessed whether proteinuria can be predictive of kidney failure, the development of
chronic kidney disease, and mortality in 37 critically ill COVID-19 patients. We used machine
learning (ML) methods as decision trees and cut-off points created by the OneR package to
add new aspects, even in smaller cohorts.
Results
Among a total of 37 patients, 24 suffered higher-grade renal failure, 20 of whom required kidney
replacement therapy. More than 40% of patients remained on hemodialysis after intensive
care unit discharge or died (27%). Due to frequent anuria proteinuria measured in two-thirds of
the patients, it was not predictive for the investigated endpoints; albuminuria was higher in
patients with AKI 3, but the difference was not significant. ML found cut-off points of >31.4 kg/
m2 for BMI and >69 years for age, constructed decision trees with great accuracy, and identified
highly predictive variables for outcome and remaining chronic kidney disease.
Conclusions
Different ML methods and their clinical application, especially decision trees, can provide
valuable support for clinical decisions. Presence of proteinuria was not predictive of CKD or
AKI and should be confirmed in a larger cohort.
Continuous glucose monitoring (CGM) improves treatment with lower blood glucose levels
and less patient effort. In combination with continuous insulin application, glycemic control
improves and hypoglycemic episodes should decrease. Direct feedback of CGM to continuous
subcutaneous insulin application, using an algorithm is called a closed-loop (CL)
artificial pancreas system. Commercial devices stop insulin application by predicting hypoglycemic
blood glucose levels through direct interaction between the sensor and pump. The
prediction is usually made for about 30 minutes and insulin delivery is restarted at the previous
level if a rise in blood glucose is predicted within the next 30 minutes (hybrid closed loop
system, HCL this is known as a predictive low glucose suspend system (PLGS)). In a fully
CL system, sensor and pump communicate permanently with each other. Hybrid closedloop
(HCL) systems, which require the user to estimate the meal size and provide a meal
insulin basis, are commercially available in Germany at the moment. These systems result
in fewer hyperglycemic and hypoglycemic episodes with improved glucose control. Open
source initiatives have provided support by building do-it-yourself CL (DIYCL) devices for
automated insulin application since 2014, and are used by a tech-savvy subgroup of
patients. The first commercial hybrid CL system has been available in Germany since September
2019. We surveyed 1054 patients to determine which devices are currently used,
which features would be in demand by potential users, and the benefits of DIYCL systems.
9.7% of these used a DIYCL system, while 50% would most likely trust these systems but
more than 85% of the patients would use a commercial closed loop system, if available. The
DIYCL users had a better glucose control regarding their time in range (TIR) and glycated
hemoglobin (HbA1c).