| CPC G16H 20/17 (2018.01) [A61M 5/142 (2013.01); A61M 2005/14208 (2013.01); A61M 2230/201 (2013.01); G06N 3/08 (2013.01); G06N 20/00 (2019.01)] | 4 Claims |
|
1. A method comprising:
determining an insulin injection amount by:
obtaining, with a computer apparatus, characteristic information and a blood glucose content at a current time of a target user through the computer apparatus, the characteristic information comprising clinical information and dietary information, wherein the clinical information comprises height, weight, fasting blood glucose content, and connecting peptide content;
determining, with the computer apparatus, an insulin injection amount at each time of a plurality of times after the current time of the target user through the computer apparatus and based on the characteristic information of the target user, the blood glucose content at the current time of the target user, a predetermined blood glucose prediction model, and a predetermined insulin injection amount prediction model; and
outputting, with the computer apparatus, the determined insulin injection amount at each time of the plurality of times after the current time of the target user through the computer apparatus; and
administering the insulin injection amount to the target user without the target user wearing a continuous blood glucose monitoring device;
wherein the determining the insulin injection amount at each time of the target user comprises:
inputting the characteristic information of the target user and the blood glucose content at the current time of the target user into the predetermined insulin injection amount prediction model to predict the insulin injection amount and determining the insulin injection amount at the current time;
inputting the insulin injection amount at the current time, the characteristic information of the target user, and the blood glucose content at the current time of the target user into the predetermined blood glucose prediction model to perform blood glucose prediction, and determining a blood glucose content of the target user at a next time corresponding to the current time; and
regarding the next time as a current time and returning to the inputting the characteristic information of the target user and the blood glucose content at the current time of the target user into the predetermined insulin injection amount prediction model to predict the insulin injection amount and determining the insulin injection amount at the current time; until the insulin injection amount at each time is determined, and obtaining the respective corresponding insulin injection amount at each time of the target user;
wherein determining an insulin injection amount further comprises:
obtaining, with the computer apparatus, a marked training set and a unmarked training set, the marked training set comprising the characteristic information, the blood glucose content of each first-sample user at each sampling time, and the insulin injection amount of each first-sample user at each sampling time; the unmarked training set comprising characteristic information and a blood glucose content of each second-sample user at each sampling time;
initializing and obtaining, with the computer apparatus, an initial insulin injection amount prediction model; and
training, with the computer apparatus, the initial insulin injection amount prediction model based on the marked training set and the unmarked training set, and determining the predetermined insulin injection amount prediction model;
wherein the initial insulin injection amount prediction model comprises first initial insulin injection amount prediction models;
wherein the determining the predetermined insulin injection amount prediction model comprises:
sampling, with computer apparatus, the marked training set and determining mark training subsets, wherein the number of the mark training subsets is the same as the number of the first initial insulin injection amount prediction models, the mark training subset comprising the characteristic information, the blood glucose content at each sampling time, and the insulin injection amount at each sampling time of some sample users of the first-sample users;
training, with computer apparatus, the corresponding first initial insulin injection amount prediction model based on the unmarked training set and the mark training subsets to obtain trained initial insulin injection amount prediction models;
performing weighted summation and averaging, with computer apparatus, on each of the trained initial insulin injection amount prediction models based on weights, and obtaining an integrated insulin injection amount prediction model; and
determining, with computer apparatus, the predetermined insulin injection amount prediction model according to the integrated insulin injection amount prediction model;
wherein the determining the predetermined insulin injection amount prediction model according to the integrated insulin injection amount prediction model comprises adjusting, with computer apparatus, each of the weights in the integrated insulin injection amount prediction model based on the marked training set, and determining the predetermined insulin injection amount prediction model; and
wherein the determining the predetermined blood glucose prediction model comprises:
obtaining, with computer apparatus, a sample marked training set; the sample marked training set comprising characteristic information, a blood glucose content at each sampling time, and an insulin injection amount at each sampling time of each third sample user;
initializing and obtaining, with computer apparatus, an initial blood glucose prediction model; and
training, with computer apparatus, the initial blood glucose prediction model based on the marked training set, and determining the predetermined blood glucose prediction model.
|