Apparatus, method, and program for predicting hypoglycemia, and apparatus, method, and program for generating hypoglycemia prediction model

1. An apparatus for predicting hypoglycemia, comprising:

a storage device that stores a machine learning model; and

a controller that:

inputs blood glucose-related information into the machine learning model; and

predicts an occurrence of postprandial hypoglycemia using the machine learning model to which the blood glucose-related information is inputted,

wherein the blood glucose-related information comprises blood glucose at a reference time, a first blood glucose change rate from a mealtime to a postprandial peak blood glucose level time, a second blood glucose change rate from a postprandial peak blood glucose level time to the reference time, and a third blood glucose change rate from a previous time to the reference time, the previous time being a time before a predetermined time from the reference time,

wherein the machine learning model is a decision tree model comprising a root node, a plurality of first intermediate nodes branching from the root node, and a plurality of second intermediate nodes branching from the first intermediate nodes,

wherein a root node of the decision tree model performs classification based on a first predetermined value of the blood glucose at the reference time,

wherein first one of the first intermediate nodes performs classification based on a second predetermined value of one of the first, second and third blood glucose change rates, and second one of the first intermediate nodes performs classification based on a third predetermined value of the blood glucose at the reference time, and

wherein a second intermediate node, which branches from the second one of the first intermediate nodes among the second intermediate nodes, performs classification based on a fourth predetermined value of other one of the first, second and third blood glucose change rates.

2. The apparatus for predicting hypoglycemia according to claim 1 , wherein the machine learning model is learned using training data consisting of previous blood glucose-related information and information about the occurrence of postprandial hypoglycemia.

3. The apparatus for predicting hypoglycemia according to claim 1 , wherein the controller predicts the occurrence of postprandial hypoglycemia within a predetermined time interval after the reference time.

4. The apparatus for predicting hypoglycemia according to claim 1 , wherein the controller

generates the blood glucose-related information by measuring and processing continuous blood glucose information,

generates the blood glucose-related information by processing the continuous blood glucose information received from another apparatus, or

receives the blood glucose-related information from another apparatus.

5. The apparatus for predicting hypoglycemia according to claim 1 , wherein the blood glucose-related information is updated in real time.

6. The apparatus for predicting hypoglycemia according to claim 1 , which is any one of an insulin pump, a continuous glucose monitor, a composite apparatus of an insulin pump and a continuous glucose monitor, an artificial pancreas apparatus, a wearable apparatus, and a hand-held apparatus.

7. The apparatus for predicting hypoglycemia according to claim 1 , wherein the controller generates the blood glucose-related information by processing continuous blood glucose information received from another apparatus or receives the blood glucose-related information from another apparatus, and

wherein the apparatus for predicting hypoglycemia is any one of an insulin pump, a wearable apparatus and a hand-held apparatus.

8. The apparatus for predicting hypoglycemia according to claim 1 , wherein the controller generates the blood glucose-related information by measuring and processing continuous blood glucose information, and

wherein the apparatus for predicting hypoglycemia is any one of a continuous glucose monitor, a composite apparatus of an insulin pump and a continuous glucose monitor and an artificial pancreas apparatus.

9. A hypoglycemia prediction method performed by a computing device storing a machine learning model, comprising:

inputting blood glucose-related information into the machine learning model; and

predicting an occurrence of postprandial hypoglycemia using the machine learning model to which the blood glucose-related information is inputted,

wherein the blood glucose-related information comprises blood glucose at a reference time, a first blood glucose change rate from a mealtime to a postprandial peak blood glucose level time, a second blood glucose change rate from the postprandial peak blood glucose level time to the reference time, and a third blood glucose change rate from a previous time to the reference time, the previous time being a time before a predetermined time from the reference time,

wherein the machine learning model is a decision tree model comprising a root node, a plurality of first intermediate nodes branching from the root node, and a plurality of second intermediate nodes branching from the first intermediate nodes,

wherein a root node of the decision tree model performs classification based on a first predetermined value of the blood glucose at the reference time,

wherein first one of the first intermediate nodes performs classification based on a second predetermined value of one of the first, second and third blood glucose change rates, and second one of the first intermediate nodes performs classification based on a third predetermined value of the blood glucose at the reference time, and

wherein a second intermediate node, which branches from the second one of the first intermediate nodes among the second intermediate nodes, performs classification based on a fourth predetermined value of other one of the first, second and third blood glucose change rate.

10. The hypoglycemia prediction method according to claim 9 , herein the machine learning model is learned using training data consisting of previous blood glucose-related information and information about the occurrence of postprandial hypoglycemia.

11. A non-transitory storage medium that stores a hypoglycemia prediction program for predicting the occurrence of postprandial hypoglycemia according to the hypoglycemia prediction method according to claim 9 .