Revolutionizing diabetes management by combining in silico models and AI control for vagus neuroprostheses

Descripción del proyecto

Estimulación automatizada del nervio vago para controlar la glucemia

Las células del cuerpo utilizan la glucosa para producir energía. La diabetes se caracteriza por una regulación deficiente de la glucemia, lo cual resulta difícil de tratar con eficacia. Se necesita urgentemente un método automatizado que se adapte a las condiciones cambiantes de los pacientes. La estimulación del nervio vago (ENV) puede modular el metabolismo de la glucosa. Sin embargo, debido a que el nervio vago inerva muchos órganos internos, se requiere una investigación cuidadosa de la interacción entre el nervio y el implante y de los circuitos estimulados. El equipo del proyecto DiabetManager, financiado por el Consejo Europeo de Investigación, tiene como objetivo desarrollar la primera neuroprótesis de ENV para la regulación personalizada de la glucemia en personas diabéticas. El trabajo incluirá el desarrollo de un modelo «in silico» del nervio vago, un diseño de interfaz impulsado por modelos, políticas de estimulación basadas en inteligencia artificial, experimentos con animales y ensayos en humanos.

Objetivo

Diabetes is a major world health problem, approaching epidemic proportions and causing long term health damages due to inappropriate blood glucose regulation. Pharmacological therapies are limited by numerous side effects (as renal insufficiency or heart failure) and dose management issues. Their efficacy is also hindered by often problematic, patients’ compliance. This results in an unmet need for automatized, closed-loop glucose metabolism regulation, adaptable to variable patients’ conditions. Vagus nerve (VN) is a major component of the autonomous nervous system innervating internal organs, and its electrical stimulation showed preliminary results in modulating glucose metabolism. However, underlying mechanisms remain unknown, posing a high-risk to its feasibility, and could lead to indiscriminate stimulation causing multiple side effects (compromising breathing or heart rate). Fine glucose regulation cannot be achieved without better understanding of nerve-implant interaction and circuits stimulated by VNS. I aim to fill present scientific gaps by developing the first VNS neuroprosthesis for personalized closed-loop-regulation of glucose levels in diabetics. This results in a significant challenge: defining stimulation policies able to selectively block or activate specific neural pathways, while counteracting metabolic shifts and implant-nerve alterations. To that aim I will exploit models-driven interface design and AI-based stimulation policies. Animal experiments will investigate the physiological responses to optimal policies, maximising glucose regulation while minimizing adverse events, inspecting both cervical and abdominal VNS. Optimal VNS will be validated in human trials achieving device assessment in patients. DiabetManager will shed light on the underlying mechanisms of VNS-mediated glucose metabolism control, while providing versatile VN in-silico model, unprecedent AI-architecture for its adaptive stimulation, and disruptive health treatment.