In August, Francis Doyle visited the Artificial Pancreas Trondheim (APT) project and gave his Volterra lecture entitled “A Systems Approach to Treating Diabetes”. Prof Doyle is currently the Dean of John A. Paulson School of Engineering and Applied Sciences (SEAS), Harvard University, and scholar in chemical engineering. Also, he has an active research group, and pursues the challenge to develop optimal automated blood sugar control, effectively curing type 1 diabetes mellitus (T1DM). His team has chosen a systems engineering approach, combining the fields of systems biology and functional biomedical control.
In his lecture he enlightened the audience on the difficulties in automating the process. As insulin, the main regulator reducing blood sugar, is lacking in type 1 diabetes, administration of the drug solves the immediate problem when blood sugar is too high. However, as glucose is critical to the body’s function, the ratio between an adequate and deadly dose of insulin is in the range of 1 to 2. Doubling the amount of paracetamol, by taking two instead of one pill, is harmless. With insulin, doubling the dose could be critical. Therefore, insulin administrated to diabetes patients must be done in a way that ensures that the blood sugar level does not dip below the minima of 80 mg/dl. Consequently, most patients have on average higher blood sugar than normal individuals, just to be on the safe side. The regulation is especially difficult for kids, having less ability to control and monitor blood sugar themselves, combined with an active and varied rhythm of life.
To achieve an automated system, a reliable and accurate infusion pump for insulin, a sensitive and rapid glucose detector, integration of the components, and software controlling the system are needed. However, there are both physiological and computational challenges. Measurements of glucose by non-invasive sensors (i.e. on the skin), will always be delayed compared with the actual glucose level in the blood. Also, the insulin release is less direct when infused subcutaneous compared to release into the blood stream. The lack of direct measurements could be overcome by computational approaches that adjust for delays and identifies major impact on the blood glucose level at an early stage, like digestion of a meal or physical exhaustion. Again, the system must ensure that dips below the critical 80 g/dl level do not occur.
By moving the sensors into the peritoneal cavity, more direct monitoring of glucose levels can be conducted. Likewise, insulin could also be administered here. Still, the feedback control algorithms need to be optimized, clinically verified and approved. Currently, Francis Doyle and colleagues are running clinical trials on the effect of new algorithms in patients testing out the approach in their daily life, at home away from the clinic. The efforts will hopefully result in a closed-loop system performing automated and optimized blood glucose regulation, normalizing the life of the many patients with T1DM, and reliving them from lifelong effects of inadequate blood sugar level control.