A novel model-based estimator for real-time prediction of insulin-on-board

The availability of insulin measurements can improve automated insulin delivery technology for people with type 1 diabetes, who require exogenous insulin delivery. To reduce the risk of hypo-or hyper-glycemia, there is a strong need of calculating the amount of insulin that is yet to become active from the previous doses, known as the insulin-on-board. In this work, we propose an approach for the real-time estimation of insulin-on-board by means of an extended Kalman filter based on actual insulin levels measured using a microchip-based immunoassay. Moreover, the availability of further insulin measure-ments, collected with high accuracy by the laboratory-based ELISA, allows the development of a proba-bilistic description of the insulin measurement error, which is exploited in the tuning of the extended Kalman filter. The proposed approach for real-time quantification of the insulin-on-board will allow an informed refinement of insulin dosing, especially under varied conditions including stress and exercise.