Modeling the Error of the Medtronic Paradigm Veo Enlite Glucose Sensor

Abstract

Continuous glucose monitors (CGMs) are prone to inaccuracy due to time lags, sensor drift, calibration errors, and measurement noise. The aim of this study is to derive the model of the error of the second generation Medtronic Paradigm Veo Enlite (ENL) sensor and compare it with the Dexcom SEVEN PLUS (7P), G4 PLATINUM (G4P), and advanced G4 for Artificial Pancreas studies (G4AP) systems. An enhanced methodology to a previously employed technique was utilized to dissect the sensor error into several components. The dataset used included 37 inpatient sessions in 10 subjects with type 1 diabetes (T1D), in which CGMs were worn in parallel and blood glucose (BG) samples were analyzed every 15 ± 5 min Calibration error and sensor drift of the ENL sensor was best described by a linear relationship related to the gain and offset. The mean time lag estimated by the model is 9.4 ± 6.5 min. The overall average mean absolute relative difference (MARD) of the ENL sensor was 11.68 ± 5.07% Calibration error had the highest contribution to total error in the ENL sensor. This was also reported in the 7P, G4P, and G4AP. The model of the ENL sensor error will be useful to test the in silico performance of CGM-based applications, i.e., the artificial pancreas, employing this kind of sensor.

Kalman Smoothing for Objective and Automatic Preprocessing of Glucose Data

“…It finds parameter values in real FGM data sets that are plausible and comparable to those found by other groups for other CGM systems [32], [33] showing that simultaneous estimation of these parameters can be done as part of the smoothing. This is an important preprocessing step when using CGM data, as using biased data could influence downstream processing.…”

“…For instance in metabolism model parameter estimation, where bias correction and knowledge of the plasma-interstitial fluid time constant is needed to prevent CGM device-specific dynamics affecting the estimation of person-specific parameters. Our SMBG and FGM data does not allow us to conclude whether the biases we observed originate from the SMBG or the FGM measurements, but we assume the latter, since bias/calibration error is commonly found to be the largest error in CGM systems [32], [33], whereas SMBG measurement errors have been found to be uncorrelated in time [17]. It should be noted that glucose data sets with both CGM and frequent SMBG measurements as those analysed here, rarely occur outside research settings.…”

“…Others argue that inferring the error distribution is confounded by modelling inaccuracies in the plasma-ISF dynamics and/or calibration errors, [31]. As stated in [32], not all CGM systems are equal. Along the lines of the SMBG measurement noise modeling of Sec.…”

“…(15) and (16) a useful default when no more information is available, and if more detailed information about the CGM errors are available for the data to be smoothed, e.g. as given in [28], [32], [33], these models can be used instead.…”

Glycemic control in the intensive care unit: A control systems perspective

“…More specifically, it is possible to know the metabolic inputs of exogenous insulin and nutrition given to support the patient, but only the BG level is measured. In particular, this BG measurement can be intermittent in 1-4 hourly intervals, but with low error [65][66][67][68][69], or effectively continuous at intervals of 1-5 minutes using continuous glucose monitors (CGMs), but with increased error and drift [70][71][72][73][74][75][76]. Hence, automation must account for either intermittent control with high quality measurements of patient response, or more continuous feedback control with lower quality measurements, both of which will impact the shape and structure of control.…”