Last April 14, 2021 Ignacio Hidalgo presented the paper Title: HERACLES-II: “Bioinspired Adaptive System for glycemic control based on smart sensors and accessories.”
Abstract : Diabetes Mellitus type 1 is a chronic disease characterized by elevated blood glucose due to a defect in insulin production, which affects more than 600 thousand people in Spain alone. Patients with this disease need, for life, both to measure their blood glucose and to inject insulin subcutaneously. In routine clinical practice, blood glucose can be measured by continuous glucose monitors and insulin is injected either manually or by a continuous subcutaneous infuser or insulin pump. On the other hand, fully autonomous glycemic control would require a predictive model to estimate the future evolution of blood glucose. With this information, a control algorithm would determine the insulin dose to be administered by the insulin pump. To date, despite the progress made, we can affirm that the estimation and prediction of blood glucose is still an open problem. For example, the data-driven models currently used to predict future glucose values make use of a subset of the quantities that affect glucose. Nowadays there are intelligent devices equipped with biosensors capable of providing physiological information in real time and therefore, a new avenue is opening up to incorporate other variables into the models. Likewise, the filters used to estimate and predict blood glucose use these models, without taking advantage of the abundance of physiological data provided by smart devices. Finally there is a need for electronic devices capable of incorporating all this information, providing in real time a more accurate blood glucose estimate. In this work we continue the research carried out in the TIN-2014-54806-R project, which has produced important results confirming its starting hypotheses. We propose, on the one hand, to extend the methodology to incorporate to the models information provided by smart devices to obtain a more accurate estimation and prediction of the evolution of blood glucose concentration. On the other hand, building on the results of filters obtained in the mentioned project, a hardware prototype of an intelligent device. This device must integrate, in real time, both the information provided by the biosensors and the information from the models in order to generate the estimation and prediction of glucose. At the end of the project, on the one hand, we will have completed the design of an integrated circuit of a Bayesian filter using models based on the data provided by the biosensors of smart devices. This must have physical characteristics (size, performance and power consumption) that allow its uninterrupted and autonomous operation, at least for a period of time comparable to that of smart devices. On the other hand, we will have implemented a set of software tools that based on evolutionary algorithms and the results of the TIN-2014-54806-R project will generate predictive models of glucose behavior. The input values will be provided by biometric sensors that in real time will be sent to the models of the insulin guideline recommender system, or even to the control algorithm in charge of generating the insulin pump performance in an eventual integration.
More information (glucmodel.ucm.es, glucnet.ucm.es).
You can find the video of the presentation at ITC Youtube channel