A Fully Integrated Closed‐Loop System Based on Mesoporous Microneedles‐Iontophoresis for Diabetes Treatment

Abstract A closed‐loop system that can mini‐invasively track blood glucose and intelligently treat diabetes is in great demand for modern medicine, yet it remains challenging to realize. Microneedles technologies have recently emerged as powerful tools for transdermal applications with inherent painlessness and biosafety. In this work, for the first time to the authors' knowledge, a fully integrated wearable closed‐loop system (IWCS) based on mini‐invasive microneedle platform is developed for in situ diabetic sensing and treatment. The IWCS consists of three connected modules: 1) a mesoporous microneedle‐reverse iontophoretic glucose sensor; 2) a flexible printed circuit board as integrated and control; and 3) a microneedle‐iontophoretic insulin delivery component. As the key component, mesoporous microneedles enable the painless penetration of stratum corneum, implementing subcutaneous substance exchange. The coupling with iontophoresis significantly enhances glucose extraction and insulin delivery and enables electrical control. This IWCS is demonstrated to accurately monitor glucose fluctuations, and responsively deliver insulin to regulate hyperglycemia in diabetic rat model. The painless microneedles and wearable design endows this IWCS as a highly promising platform to improve the therapies of diabetic patients.

the basic glucose level, while iontophoresis was coupled to MMN to facilitate a bolus delivery to minimize glucose fluctuation. 5) The IWCS was compactly designed, with satisfying miniaturization and flexibility to be worn on arms of human adults. 6) The system was intelligent and automatic. By Bluetooth communication with the smartphone, the glucose sensing data could be stored, displayed and tracked. The smartphone could also analyze data and trigger delivery. 7) By analyzing and co-relating the continuous data of glucose fluctuations with daily activities, diabetes managements could be quantitatively performed.         The diffusion coefficient of small molecules in water is generally in the range of 1e -9 -1e -11 [m^2/s]. 2 The results using different diffusion coefficients (ranging from 3e -10 to 9e -10 [m^2/s])were also investigated below.

Diffusion coefficient of insulin in water.
The diffusion coefficient of glucose in water is generally in the range of 3e -10 to 9e -10 [m^2/s]. Considering that the diameter of an insulin molecule is roughly 3-folds larger than a glucose molecule, the Diffusion coefficient of insulin is assumed to be 3folds lower than the glucose. The results using different diffusion coefficients (ranging from 1e -10 to 2.5e -10 [m^2/s]) were also investigated below. Other porosities (0.3, 0.4, 0.6) were also investigated.
Zg 0 Electric charge of glucose.
Glucose molecule is uncharged in water. During iontophoresis, Na + and Clunderneath skin migrate toward the electrodes by the applied potentials. The iontophoresis of glucose is achieved by convective transport of glucose along with the ions. The electric charge of Na + is +1. The glucose moved to the cathode along with the Na + , with a convective ratio assumed to be 1. The results using different convective ratio (ranging from 1 to 1/16) were also investigated below.

Zi -2
The net negative charge of insulin in PBS solution (pH=7.4) is about -2. 5,6 Cg0 10 mM Initial concentration of glucose in the interstitial space.

Ci0 4 mg/ml
Initial concentration of insulin in the delivery chamber.
Ii 0.5 [mA] Iontophoresis current for insulin delivery. The results using different iontophoresis currents were also investigated.
Iri 0.5 [mA] Reverse iontophoresis current for glucose extraction. The results using different iontophoresis currents were also investigated.       The uses of more calibration points could further improve the sensing accuracies, but induce more pain and inconvenience due to frequent BG measurements. One-point calibration could produce results closely satisfying the clinical requirement of error <15%. The uses of more calibration points could further improve the sensing accuracy, but induced more pains and inconvenience due to frequent BG measurements.                  At the beginning of the program, user need to connect IWCS's Bluetooth manually.
Application received and resolved the stream of data that transmitted in real time from the IWCS. Since IWCS did not filter the analog signal, it was necessary to filter the data before calculating the glucose concentration value to eliminate noise interference. The glucose concentration signal was a slowly changing signal, close to a direct-current signal. Therefore, the Butterworth low-pass filter with the cut-off frequency of 1 Hz was adopted to filter the glucose concentration signal. The output data of the digital filter was converted to glucose    The error of all the RIMN sensor-measured glucose signal was below 40%, with an average error of 17.5 ± 13.9%. In the clarke's error grid analysis, 90% of data located in region A.
(b) Calibration using two BG measurements (at time points t=0 and 180 min). The error of all the RIMN sensor-measured glucose signal was below 25%, with an average error of 9.5 ± 8.3%. In the clarke error grid analysis, 89% of data located in region A. (c) Calibration using four BG measurements (at time points t=0, 80, 180, and 280 min). The error of all the RIMN sensor-measured glucose signal was below 17%, with an average error of 4.13 ± 5.18%. In the clarke error grid analysis, 100% of healthy group data located in region A. (d) Statistical analysis of the average error using one, two, and four-points calibrations. Onepoint calibration could produce results closely satisfying the clinical requirement of error <15%. The uses of more calibration points could further improve the sensing accuracy, but induced more pains and inconvenience due to frequent BG measurements.