Joint state estimation prediction that evaluates differences in predicted vs. corresponding received data

A system for reconciling diabetes-management data pertaining to a subject, the system comprising:

• a continuous glucose monitor (CGM) configured to generate glucose measurements of the subject;
• a memory configured to store the glucose measurements as trusted data and to store untrusted data pertaining to the subject, wherein the untrusted data comprises user-entered data including at least one of insulin data, meal data, or activity data, and wherein each of the untrusted insulin data, untrusted meal data, and untrusted activity data is associated with timing data;
• and
• at least one processor operatively coupled to the CGM and the memory and configured to:
• receive, at an input estimator executing on the at least one processor, the trusted data and the untrusted data;
• generate, using a mathematical model of the subject's glucose-insulin dynamics, an estimate of the untrusted data from the trusted data;
• reconcile, using an input reconciler executing on the at least one processor, the untrusted data with the estimate of the untrusted data to produce reconciled data;
• and
• output the reconciled data.