Insulin therapy determination
Techniques for determining insulin therapy are described. The techniques include obtaining patient characteristic information for a current patient; determining, with a machine-learning model, an insulin delivery therapy from a plurality of insulin delivery therapies for the current patient based on the patient characteristic information, wherein the machine-learning model is generated based on digital representations of a plurality of patients; and outputting information indicative of the determined insulin delivery therapy.
1 . A system for determining insulin therapy, the system comprising:
one or more processors; and
one or more processor-readable storage media storing instructions which, when executed by the one or more processors, cause performance of:
obtaining patient characteristic information and one or more target therapy efficacy metrics representative of a desired clinical outcome for a current patient, wherein the one or more target therapy efficacy metrics include a target time spent within a target range having an upper limit and a lower limit;
determining, based on an output of a machine-learning model that takes the patient characteristic information and the one or more target therapy efficacy metrics as an input, an insulin delivery therapy from a plurality of insulin delivery therapies for the current patient by selecting between an insulin delivery therapy that delivers the same amount of insulin for each meal, and an insulin delivery therapy that customizes insulin amounts, wherein
the machine-learning model has been trained to determine whether an insulin therapy that customizes insulin amounts for each meal or delivers the same amount of insulin for each meal will cause the current patient to achieve the desired clinical outcome represented in the one or more target therapy efficacy metrics based on the current patient having the patient characteristic information;
outputting information indicative of the determined insulin delivery therapy; and
delivering insulin, using an insulin pump, to the current patient in accordance with the determined insulin delivery therapy, wherein delivering the insulin causes a glucose level of the patient to decrease to within the target range.
2 . The system of claim 1 , wherein the plurality of insulin delivery therapies comprises at least a first insulin delivery therapy indicating the same amount of insulin for each meal, a second insulin delivery therapy indicating customized insulin amounts for each meal, and a third insulin delivery therapy indicating a first insulin amount for meals during a first set of days and a second insulin amount for meals during a second set of days.
3 . The system of claim 1 , wherein the patient characteristic information for the current patient comprises a first instance of patient characteristic information for the current patient, wherein the determined insulin delivery therapy comprises a first instance of insulin delivery therapy, and wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
obtaining a second instance of patient characteristic information for the current patient after obtaining the first instance of patient characteristic information;
determining, with the machine-learning model, a second instance of insulin delivery therapy from the plurality of insulin delivery therapies for the current patient based on the second instance of patient characteristic information;
outputting information indicative of the determined second instance of insulin delivery therapy.
4 . The system of claim 1 , wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
obtaining, for each patient of a plurality of patients, demographic information and information indicative of outcomes from therapy;
simulating, for each patient of the plurality of patients, using a respective digital representation, therapy outcomes for each insulin delivery therapy of the plurality of insulin delivery therapies to generate a plurality of simulated therapy outcomes;
selecting, for each patient of the plurality of patients, a respective insulin delivery therapy from the plurality of insulin delivery therapies based on the plurality of simulated therapy outcomes, wherein the selected respective insulin delivery therapy corresponds to an insulin delivery therapy associated with a best simulated therapy outcome of the plurality of simulated therapy outcomes; and
generating the machine-learning model based on the demographic information, information indicative of outcomes from therapy, and selected insulin delivery therapy of each patient of the plurality of patients.
5 . The system of claim 1 , wherein the machine-learning model is generated based on supervised training.
6 . The system of claim 1 , wherein generating the machine-learning model comprises generating a plurality of clusters, each cluster corresponding to an insulin delivery therapy of the plurality of insulin delivery therapies, and wherein determining, based on the output of the machine-learning model, the insulin delivery therapy comprises assigning the current patient to a cluster of the plurality of clusters based on the patient characteristic information.
7 . The system of claim 1 , wherein determining the insulin delivery therapy from the plurality of insulin delivery therapies for the current patient comprises determining the insulin delivery therapy without utilizing a digital representation for the current patient.
8 . A method for determining insulin therapy, the method comprising:
obtaining patient characteristic information and one or more target therapy efficacy metrics representative of a desired clinical outcome for a current patient, wherein the one or more target therapy efficacy metrics include a target time spent within a target range having an upper limit and a lower limit;
determining, based on an output of a machine-learning model that takes the patient characteristic information and the one or more target therapy efficacy metrics as an input, an insulin delivery therapy from a plurality of insulin delivery therapies for the current patient by selecting between an insulin delivery therapy that delivers the same amount of insulin for each meal, and an insulin delivery therapy that customizes insulin amounts, wherein:
the machine-learning model has been trained to determine whether an insulin therapy that customizes insulin amounts for each meal or delivers the same amount of insulin for each meal will cause the current patient to achieve the desired clinical outcome represented in the one or more target therapy efficacy metrics based on the current patient having the patient characteristic information;
outputting information indicative of the determined insulin delivery therapy; and
delivering insulin, using an insulin pump, to the current patient in accordance with the determined insulin delivery therapy, wherein delivering the insulin causes a glucose level of the patient to decrease to within the target range.
9 . The method of claim 8 , wherein the plurality of insulin delivery therapies comprises at least a first insulin delivery therapy indicating the same amount of insulin for each meal, a second insulin delivery therapy indicating customized insulin amounts for each meal, and a third insulin delivery therapy indicating a first insulin amount for meals during a first set of days and a second insulin amount for meals during a second set of days.
10 . The method of claim 8 , wherein the patient characteristic information for the current patient comprises a first instance of patient characteristic information for the current patient, wherein the determined insulin delivery therapy comprises a first instance of insulin delivery therapy, and wherein the method further comprises:
obtaining a second instance of patient characteristic information for the current patient after obtaining the first instance of patient characteristic information;
determining, with the machine-learning model, a second instance of insulin delivery therapy from the plurality of insulin delivery therapies for the current patient based on the second instance of patient characteristic information;
outputting information indicative of the determined second instance of insulin delivery therapy.
11 . The method of claim 8 , further comprising:
obtaining, for each patient of a plurality of patients, demographic information and information indicative of outcomes from therapy;
simulating, for each patient of the plurality of patients, using a respective digital representation, therapy outcomes for each insulin delivery therapy of the plurality of insulin delivery therapies to generate a plurality of simulated therapy outcomes;
selecting, for each patient of the plurality of patients, a respective insulin delivery therapy from the plurality of insulin delivery therapies based on the plurality of simulated therapy outcomes; and
generating the machine-learning model based on the demographic information, information indicative of outcomes from therapy, and selected insulin delivery therapy of each patient of the plurality of patients.
12 . The method of claim 8 , wherein the machine-learning model is generated based on supervised training.
13 . The method of claim 8 , wherein the machine-learning model is generated based on unsupervised training.
14 . The method of claim 8 , wherein determining the insulin delivery therapy from the plurality of insulin delivery therapies for the current patient comprises determining the insulin delivery therapy without utilizing a digital representation for the current patient.
15 . One or more non-transitory processor-readable storage media storing instructions which, when executed by the one or more processors, cause performance of:
obtaining patient characteristic information and one or more target therapy efficacy metrics representative of a desired clinical outcome for a current patient, wherein the one or more target therapy efficacy metrics include a target time spent within a target range having an upper limit and a lower limit and a maximum allowable number of hypoglycemic events;
determining, based on an output of a machine-learning model that takes the patient characteristic information and the one or more target therapy efficacy metrics as an input, an insulin delivery therapy from a plurality of insulin delivery therapies for the current patient by selecting between an insulin delivery therapy that delivers the same amount of insulin for each meal, and an insulin delivery therapy that customizes insulin amounts, wherein:
the machine-learning model has been trained to determine whether an insulin delivery therapy that customizes insulin amounts for each meal or delivers the same amount of insulin for each meal will cause the current patient to achieve the desired clinical outcome represented in the one or more target therapy efficacy metrics based on the current patient having the patient characteristic information, and
wherein the machine-learning model has been trained to account for an accuracy of the current patient in counting carbohydrates based on the patient characteristic information in the determination of the insulin delivery therapy;
outputting information indicative of the determined insulin delivery therapy; and
delivering insulin, using an insulin pump, to the current patient in accordance with the determined insulin delivery therapy, wherein delivering the insulin causes a glucose level of the patient to decrease to within the target range.
16 . The one or more non-transitory processor-readable storage media of claim 15 , wherein the plurality of insulin delivery therapies comprises at least a first insulin delivery therapy indicating the same amount of insulin for each meal, a second insulin delivery therapy indicating customized insulin amounts for each meal, and a third insulin delivery therapy indicating a first insulin amount for meals during a first set of days and a second insulin amount for meals during a second set of days.
17 . The one or more non-transitory processor-readable storage media of claim 15 , wherein the patient characteristic information for the current patient comprises a first instance of patient characteristic information for the current patient, wherein the determined insulin delivery therapy comprises a first instance of insulin delivery therapy, and wherein the one or more non-transitory processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
obtaining a second instance of patient characteristic information for the current patient after obtaining the first instance of patient characteristic information;
determining, with the machine-learning model, a second instance of insulin delivery therapy from the plurality of insulin delivery therapies for the current patient based on the second instance of patient characteristic information;
outputting information indicative of the determined second instance of insulin delivery therapy.
18 . The one or more non-transitory processor-readable storage media of claim 15 , further storing instructions which, when executed by the one or more processors, cause performance of:
obtaining, for each patient of a plurality of patients, demographic information and information indicative of outcomes from therapy;
simulating, for each patient of the plurality of patients, using a respective digital representation, therapy outcomes for each insulin delivery therapy of the plurality of insulin delivery therapies to generate a plurality of simulated therapy outcomes;
selecting, for each patient of the plurality of patients, a respective insulin delivery therapy from the plurality of insulin delivery therapies based on the plurality of simulated therapy outcomes; and
generating the machine-learning model based on the demographic information, information indicative of outcomes from therapy, and selected insulin delivery therapy of each patient of the plurality of patients.
19 . The one or more non-transitory processor-readable storage media of claim 15 , wherein the machine-learning model is generated based on supervised training.
20 . The one or more non-transitory processor-readable storage media of claim 15 , wherein determining the insulin delivery therapy from the plurality of insulin delivery therapies for the current patient comprises determining the insulin delivery therapy without utilizing a digital representation for the current patient.