Constructing an intelligent insulin decision-making system for dynamic glucose control in type 2 diabetes mellitus via a multicentre federated learning algorithm, comparing the performance of the federated learning model, the local model and the initial model, and evaluating their feasibility and safety.

Constructing an intelligent insulin decision-making system for dynamic glucose control in type 2 diabetes mellitus via a multicentre federated learning algorithm, comparing the performance of the federated learning model, the local model and the initial model, and evaluating their feasibility and safety.

Inclusion Criteria:

• type 2 diabetes inpatients receiving insulin therapy

Exclusion Criteria:

• use of insulin pumps or glucocorticoids during hospitalisation
• less than two days of insulin therapy

type 2 diabetes who