Diabetes prediction using glucose measurements and machine learning

Abstract

Diabetes prediction using glucose measurements and machine learning is described. In one or more implementations, the observation analysis platform includes a machine learning model trained using historical glucose measurements and historical outcome data of a user population to predict a diabetes classification for an individual user. The historical glucose measurements of the user population may be provided by glucose monitoring devices worn by users of the user population, while the historical outcome data includes one or more diagnostic measurements obtained from sources independent of the glucose monitoring devices. Once trained, the machine learning model predicts a diabetes classification for a user based on glucose measurements collected by a wearable glucose monitoring device during an observation period spanning multiple days. The predicted diabetes classification may then be output, such as by generating one or more notifications or user interfaces based on the classification.

What is claimed is: 
     
         1 . A method comprising:
 obtaining glucose measurements of a user;   processing the glucose measurements of the user to extract one or more glucose features;   predicting a diabetes classification of the user by providing the one or more extracted glucose features to one or more machine learning models as input, the one or more machine learning models generated based on historical glucose measurements provided by glucose monitoring devices worn by users of a user population and historical outcome data of the user population; and   outputting the diabetes classification.   
     
     
         2 . The method as described in  claim 1 , wherein the diabetes classification is an indication describing a state of the user as having one of diabetes, prediabetes, or no diabetes. 
     
     
         3 . The method as described in  claim 1 , wherein the diabetes classification is an indication describing a state of the user as having gestational diabetes or no gestational diabetes. 
     
     
         4 . The method as described in  claim 1 , wherein:
 the diabetes classification is an indication of one or more adverse effects of diabetes the user is predicted to experience; and   the historical outcome data describes adverse effects of diabetes observed in users of the user population.   
     
     
         5 . The method of  claim 1 , wherein the glucose measurements of the user are collected by a wearable glucose monitoring device during an observation period. 
     
     
         6 . The method as described in  claim 5 , wherein the wearable glucose monitoring device includes a sensor that is inserted subcutaneously into skin of the user during the observation period to collect the glucose measurements. 
     
     
         7 . The method as described in  claim 6 , wherein the glucose measurements comprise time-sequenced glucose measurements collected by the wearable glucose monitoring device during the observation period. 
     
     
         8 . The method as described in  claim 7 , wherein the time-sequenced glucose measurements are collected by the wearable glucose monitoring device continuously at predetermined intervals during the observation period. 
     
     
         9 . The method as described in  claim 1 , further comprising:
 obtaining the historical glucose measurements and the historical outcome data of the user population, the historical glucose measurements provided by glucose monitoring devices worn by users of the user population; and   generating the one or more machine learning models by providing the historical glucose measurements to the one or more machine learning models, comparing training classifications of diabetes received from the one or more machine learning models to diabetes classifications indicated by the historical outcome data, and adjusting weights of the one or more machine learning models.   
     
     
         10 . The method as described in  claim 9 , wherein the historical outcome data is associated with one or more diagnostic measures independent of the historical glucose measurements provided by the glucose monitoring devices worn by users of the user population. 
     
     
         11 . The method as described in  claim 1 , wherein the historical outcome data includes labels of traces of the historical glucose measurements that indicate whether a respective user of the user population is clinically diagnosed with diabetes based on one or more diagnostic measures. 
     
     
         12 . The method as described in  claim 11 , wherein the one or more diagnostic measures include at least one of Hemoglobin A1c (HbA1c), an Oral Glucose Tolerance Test (OGTT), or Fasting Plasma Glucose (FPG). 
     
     
         13 . The method as described in  claim 1 , wherein the one or more extracted glucose features include a time over threshold measure corresponding to an amount of time during an observation period that the glucose measurements of the user are above a glucose threshold. 
     
     
         14 . The method as described in  claim 1 , wherein the one or more extracted glucose features include a time within range measure corresponding to an amount of time during an observation period that the glucose measurements of the user are between a first glucose level and a second glucose level that is less than the first glucose level. 
     
     
         15 . The method as described in  claim 1 , wherein the one or more extracted glucose features include a rate-of-change measure corresponding to a difference in glucose measurements over a unit of time. 
     
     
         16 . The method as described in  claim 1 , wherein the machine learning model predicts the diabetes classification of the user using at least two extracted glucose features. 
     
     
         17 . The method as described in  claim 1 , wherein the outputting comprises outputting a glucose observation report that includes the diabetes classification and at least one of:
 one or more treatment recommendations for the user based on the diabetes classification;   a visual representation of the glucose measurements; or   one or more glucose statistics of the user generated based on the glucose measurements.   
     
     
         18 . A method comprising:
 obtaining historical glucose measurements of users in a user population;   obtaining historical outcome data of the users in the user population;   generating instances of training data that include a training input portion and an expected output portion, the training input portion including one or more features of a respective user's glucose measurements, and the expected output portion including one or more values of outcome data corresponding to the respective user; and   training a machine learning model to predict a diabetes classification using the instances of training data.   
     
     
         19 . The method of  claim 18 , wherein the historical outcome data indicates whether or not a respective user of the user population is clinically diagnosed with diabetes. 
     
     
         20 . A system comprising:
 a wearable glucose monitoring device comprising a sensor to collect glucose measurements of a user;   a storage device to maintain the glucose measurements of the user collected during an observation period; and   a prediction system to obtain the glucose measurements of the user collected during the observation period, and predict a diabetes classification of the user by processing the glucose measurements of the user to extract one or more glucose features and providing the one or more extracted glucose features to one or more machine learning models as input.