Evolvable hardware-based digital filter to handle patient and sensor variability in continuous glucose sensor measurements of artificial pancreas

Project facts

Project promoter
Complutense University of Madrid
Project Number:
Target groups
Researchers or scientists
Initial project cost:
Final project cost:
From EEA Grants:
€ 4,028
The project is carried out in:


Diabetes Mellitus is a disease that affects millions of people worldwide. A high percentage of patients cannot normalize their glucose levels with the current treatments, so lately much research has been devoted to create an artificial pancreas. The main objective of the project is to improve the state of the art implementations of a continuous glucose sensor modifying their digital filters with novel techniques based on evolvable hardware. This objective be achieved using a Kalman filter capable of adapting in real time to the variable patient’s and sensor features by applying novel techniques based on evolvable hardware. The donor partner, University of Oslo, has an extensive background in the design and implementation of eHW (evolvable hardware) and adaptive systems. The project promoter, Complutense University of Madrid, will acquire the knowledge, techniques and methodologies required to develop eHW (evolvable hardware) filter while the team from the University of Oslo will acquire knowledge of a biomedical field in which eHW has a potential niche of application. The results will benefit the scientist community and mainly the patients with diabetes

Summary of project results

Diabetes mellitus is a disease that affects millions of people worldwide. It causes the sugar cannot enter the cells, raising the blood glucose levels (BG). A high percentage of patients cannot normalize their BG levels with the current treatments. Some of the issues affecting the reliability of the continuous glucose sensor (CG) are the noise, degeneration, calibration and adaptability of the CGS algorithms to the patient’s features. Currently, there are several approaches to address the problems of calibration, time-lag between BG and interstitial glucose (IG) measurements, and filtering of signals coming from CGS. All of them have in common the lack of adaptability to changes in the sensor’s parameters, in the patient’s parameters and modifications in the environment where the system operates. All these problems can be addressed comprehensively by applying evolvable hardware (eHW) techniques to the design of the filters that process the signal generated by the sensor. However, up-to-date no research group has tackled this problem with this approach. The goal of the project is to evaluate the viability of the eHW to improve CGS measurements. To this aim, partners designed and compare different implementations of evolvable digital filters and draw some conclusions about the most appropriate application in the CGS for each kind of implementation. The specific scientific activities carried out are: 1) Review of recent publications in the field of Evolutionary Hardware; 2) Design, development and implementation of the ZedBoard TM Zynq Evaluation Kit (XC7-Z020 ELQ484-1) of a parametrizable hardware module that implements an Evolutionary Strategy. 3) Integration of the module with a classifier of sonar signals and evaluation of the performance and physical characteristics of the full system; 4) Design and development of an ES in C programming language; 5) Implementation of a Zynq XC7-Z020 hard PS7 5.4 built-in dual-core ARM processor Cortex TM –A9 processor on the Zed Board TM Zynq Evaluation Kit (XC7-XZ020 ELQ484-1) using IP component libraries; 6) Integration of the processor with the classifier module using a communication system based on the AXI bus; 7) Evaluation of the performance and physical characteristics; 8) Implementation of a MicroBlaze processor; 11) Evaluation of the performance; 12) Evaluation and comparison of the different ES deployments for their usage in Evolutionary Hardware systems.

Summary of bilateral results

As a result of the close collaboration, partners are preparing a paper with the results of the project, to be submitted to a journal in the field of evolutionary and evolvable systems. In addition to the collaboration with Prof. Torrensen, Spanish researcher Prof. Garnica has also worked very closely with Prof. Kyrre Glette, who leads the Evolutionary and Reconfigurable sub-group in ROBIN team. Recently, ROBIN team has been granted with a project funded by the Norwegian Research Council to develop a robotic system to assist the elderly people in their homes. Partners are exploring the feasibility of incorporating lucose monitoring in the system, and hereby, provided it be feasible, open a line of future cooperation between the Spanish and Norwegian group.