| CPC A61B 5/4866 (2013.01) [A61B 5/1118 (2013.01); A61B 5/14532 (2013.01); A61B 5/4833 (2013.01); A61B 5/486 (2013.01); A61B 5/6801 (2013.01); A61B 5/6802 (2013.01); A61B 5/7267 (2013.01); A61B 5/7275 (2013.01); A61B 5/742 (2013.01); A61B 5/7475 (2013.01); A61B 5/749 (2013.01); G06N 20/00 (2019.01); G16H 10/40 (2018.01); G16H 10/60 (2018.01); G16H 15/00 (2018.01); G16H 20/10 (2018.01); G16H 20/60 (2018.01); G16H 40/67 (2018.01); G16H 50/20 (2018.01); G16H 50/30 (2018.01); G16H 50/50 (2018.01); A61B 5/0205 (2013.01)] | 17 Claims |
|
1. A method for tracking changes in a metabolic state of a patient, the method comprising:
generating, at a computing device, a patient-specific treatment recommendation for a time period, the patient-specific treatment recommendation including one or more actions for the patient to perform to improve the metabolic state;
receiving, at the computing device, at periodic intervals throughout the time period, recordings of patient data associated with the one or more actions indicating one or more of: 1) food items consumed by the patient during the time period and 2) medication taken by the patient during the time period, the recordings of patient data recorded by the patient;
iteratively, at the computing device, training a metabolic model for predicting a metabolic state of the patient, comprising:
accessing a patient-specific training dataset comprising a plurality of training examples, wherein each training example comprises previous patient data, biosignals of the patient, and a labeled metabolic state of the patient, wherein the patient-specific training dataset is periodically updated with patient health information and updated metabolic states corresponding to the updated patient health information; and
inputting, to the metabolic model, the patient-specific training dataset to draw correlations between each labeled metabolic state and the previous patient data and correlations between each labeled metabolic state and previous biosignals;
inputting, to the trained metabolic model, the recordings of patient data received during the time period to determine a predicted representation of a current metabolic state of the patient during the time period, wherein the predicted representation is a patient-specific prediction indicating a current metabolic performance of the patient;
inputting, to the trained metabolic model, biosignals recorded by at least one wearable sensor worn by the patient or lab test data collected for the patient during the time period to determine a true representation of the current metabolic state of the patient during the time period;
identifying, at the computing device, a discrepancy in the recordings of patient data based on a comparison of the predicted representation of the current metabolic state and the true representation of the current metabolic state;
computing, at the computing device, a score that rates an adherence of the patient to the patient-specific treatment recommendation based on the identified discrepancy;
presenting, via a graphical user interface on the computing device, when the discrepancy is higher than a predefined threshold, a notification informing the patient of the discrepancy and the computed score, comprising:
modifying the graphical user interface to display a user interface comprising a textual prompt requesting the patient to provide updated patient data for one or more potential instances of incorrectly recorded patient data causing the discrepancy;
receiving, via the graphical user interface, modified recordings of patient data related to instances of incorrectly recorded patient data that caused the discrepancy; and
inputting, at the computing device, the modified recordings of patient data to the trained metabolic model to determine an updated representation of the current metabolic state.
|