MACHINE LEARNING HEALTH ANALYSIS WITH A MOBILE DEVICE
Disclosed herein are devices, systems, methods and platforms for continuously monitoring the health status of a user, for example the cardiac health status. The present disclosure describes systems, methods, devices, software, and platforms for continuously monitoring a user's low-fidelity health-indicator data (for example and without limitation PPG signals, heart rate or blood pressure) from a user-device in combination with corresponding (in time) data related to factors that may impact the health-indicator (“other-factors”) to determine whether a user has normal health as judged by or compared to, for example and not by way of limitation, either (i) a group of individuals impacted by similar other-factors, or (ii) the user him/herself impacted by similar other-factors.
1 . A method comprising:
receiving low-fidelity health-indicator data of a user and other-factor data of the user at a current time;
inputting the low-fidelity health-indicator data and the other-factor data into a machine learning (ML) model;
predicting by the ML model, health-indicator data of the user at a future time based on the low-fidelity health-indicator data and the other-factor data;
at the future time, determining if measured health indicator data of the user is outside a normal range based on the predicted health-indicator data and the measured health-indicator data, wherein the measured health-indicator data is measured at the future time; and
in response to determining that the measured health-indicator data is outside the normal range, providing the user with a notification that the measured health-indicator data is outside the normal range.
2 . The method of claim 1 , wherein determining if the measured health-indicator data is outside the normal range comprises:
determining a loss based on the predicted health-indicator data and the measured health-indicator data; and
determining that the measured health-indicator data is outside the normal range if the loss exceeds a predefined loss threshold.
3 . The method of claim 2 , wherein the loss is determined based on an absolute value of a difference between the predicted health-indicator data and the measured health-indicator data.
4 . The method of claim 2 , wherein the predicted health-indicator data comprises a probability distribution of health-indicator values of the user at the future time.
5 . The method of claim 4 , further comprising:
sampling the probability distribution to select a particular health-indicator value, wherein the loss is determined based on the selected particular health-indicator value and the measured health-indicator data.
6 . The method of claim 5 , wherein sampling the probability distribution is performed using a mean value of the probability distribution, a maximum value of the probability distribution or a random sampling of the probability distribution.
7 . The method of claim 1 , further comprising:
training the ML model using a set of low-fidelity health-indicator training data labeled with high-fidelity measurement data, wherein the low-fidelity health-indicator training data and the high-fidelity measurement data is from a population of subjects.
8 . The method of claim 1 , further comprising:
in response to determining that the measured health-indicator data is outside the normal range, calculating by the ML model, an amount of time the measured health-indicator data will be outside the normal range.
9 . The method of claim 1 , wherein the notification comprises one or more of: an instruction to obtain a high-fidelity measurement and an instruction to contact a physician.
10 . The method of claim 2 , further comprising:
determining second predicted health-indicator data for a subsequent time using a weighted combination of the predicted health-indicator data and the measured health indicator data, wherein the combination is based at least in part on a size of the loss.
11 . An apparatus comprising:
a memory; and
a processing device operatively coupled to the memory, the processing device to:
receive low-fidelity health-indicator data of a user and other-factor data of the user at a current time;
input the low-fidelity health-indicator data and the other-factor data into a machine learning (ML) model;
predict by the ML model, health-indicator data of the user at a future time based on the low-fidelity health-indicator data and the other-factor data;
at the future time, determine if measured health indicator data of the user is outside a normal range based on the predicted health-indicator data and the measured health-indicator data, wherein the measured health-indicator data is measured at the future time; and
in response to determining that the measured health-indicator data is outside the normal range, provide the user with a notification that the measured health-indicator data is outside the normal range.
12 . The apparatus of claim 11 , wherein to determine if the measured health-indicator data is outside the normal range, the processing device is to:
determine a loss based on the predicted health-indicator data and the measured health-indicator data; and
determine that the measured health-indicator data is outside the normal range if the loss exceeds a predefined loss threshold.
13 . The apparatus of claim 12 , wherein the processing device determines the loss based on an absolute value of a difference between the predicted health-indicator data and the measured health-indicator data.
14 . The apparatus of claim 12 , wherein the predicted health-indicator data comprises a probability distribution of health-indicator values of the user at the future time.
15 . The apparatus of claim 14 , wherein the processing device is further to:
sample the probability distribution to select a particular health-indicator value, wherein the loss is determined based on the selected particular health-indicator value and the measured health-indicator data.
16 . The apparatus of claim 15 , wherein the processing device samples the probability distribution using a mean value of the probability distribution, a maximum value of the probability distribution or a random sampling of the probability distribution.
17 . The apparatus of claim 11 , wherein the processing device is further to:
train the ML model using a set of low-fidelity health-indicator training data labeled with high-fidelity measurement data, wherein the low-fidelity health-indicator training data and the high-fidelity measurement data is from a population of subjects.
18 . The apparatus of claim 11 , wherein the processing device is further to:
in response to determining that the measured health-indicator data is outside the normal range, calculate by the ML model, an amount of time the measured health-indicator data will be outside the normal range.
19 . The apparatus of claim 11 , wherein the notification comprises one or more of: an instruction to obtain a high-fidelity measurement and an instruction to contact a physician.
20 . The apparatus of claim 12 , wherein the processing device is further to:
determine second predicted health-indicator data for a subsequent time using a weighted combination of the predicted health-indicator data and the measured health indicator data, wherein the combination is based at least in part on a size of the loss.