Granted Patent B2

US 12,614,630 · App. 18/310,801 · Granted Apr 28, 2026

Inventors: Christopher Kanan (Pittsford, NY); Rodrigo Ceballos Lentini (Flemington, NJ); Jillian Sue (New York, NY); Thomas Fuchs (New York, NY); Leo Grady (Darien, CT)

Assignee: Paige.AI, Inc.

G16H30/40G06T7/0012G16H50/20G06T2207/20081G06T2207/20084

Abstract

Claims (68)

1 . A computer-implemented method for monitoring health of a population, the method comprising:

receiving a query from a user requesting monitoring health of a population, wherein the query identifies at least one geographic region of a plurality of geographic regions;

in response to receiving the query, determining a current prevalence of at least one data variable and/or at least one health variable in the at least one geographic region, via a machine learning model trained to determine whether the at least one data variable and/or the at least one health variable is present in the at least one geographic region, by:

determining that the at least one data variable and/or the at least one health variable is present in a plurality of digital pathology images and/or reports input into the machine learning model over an interval of time, wherein the machine learning model is trained using whole slide images and diagnostic data corresponding to the plurality of geographic regions;

determining a first trend in a relationship between the at least one data variable and/or the at least one health variable and the at least one geographic region at a time of the query;

determining a second trend in the relationship between the at least one data variable and/or the at least one health variable and the at least one geographic region at at least one prior point in time;

determining, using the trained machine learning model, a comparison of the first trend and the second trend;

determining, based on the comparison, that the first trend in the relationship between the at least one data variable and/or the at least one health variable and the at least one geographic region is irregular within a predetermined threshold with respect to the second trend between the at least one data variable and/or the at least one health variable and the at least one geographic region from the at least one prior point in time and associated with a timing of the received query; and

upon determining that the first trend in the relationship between the at least one data variable and/or the at least one health variable the at least one geographic region, and the timing of the received query is irregular within the predetermined threshold, generating an alert.

2 . The computer-implemented method of claim 1 , further comprising:

generating an updated relationship between the at least one geographic region, the at least one data variable, and/or the at least one health variable based on the determined relationship, comprising:

providing the updated relationship to the machine learning model;

applying the machine learning model to predict at least one future relationship over a time period in the at least one geographic region; and

generating the updated relationship between the at least one geographic region, the at least one data variable, and/or the at least one health variable based on the at least one future relationship.

3 . The computer-implemented method of claim 2 , wherein the machine learning model comprises:

a convolutional neural network; a graph convolutional network; an autoregressive model; a recurrent neural network; and/or a capsule network.

4 . The computer-implemented method of claim 2 , wherein applying the machine learning model to predict at least one future relationship over time period in the at least one geographic region comprises:

receiving time-stamped human subject data, associated with a plurality of patients, from the at least one geographic region;

inferring health related variables from the time-stamped human subject data; and

determining disease states and applying at least one natural language processing model to at least one clinic note to extract at least one relevant variable.

5 . The computer-implemented method of claim 4 , wherein the time-stamped human subject data comprises at least one of a plurality of digital images of pathology specimens, genetic data, pathogenic data, and/or clinical notes.

6 . The computer-implemented method of claim 4 , further comprising:

training a machine learning model to predict future relationships over time in a geographic region; and

determining, based on the predicted future relationships, whether to continue monitoring health of a population.

7 . The computer-implemented method of claim 2 , wherein applying a machine learning model to predict at least one future relationship over time in the at least one geographic region comprises using at least one of: a convolutional neural network; a graph convolution network; an autoregressive model; a recurrent neural network; and a capsule network.

8 . The computer-implemented method of claim 1 , wherein determining the relationship between the at least one geographic region, at least one data variable, or the at least one health variable comprises using correlation machine learning and/or geographic visual overlay.

9 . The computer-implemented method of claim 1 , wherein a query is received from a user for one of a particular health variable, a particular data variable.

10 . The computer-implemented method of claim 1 , wherein a relationship is determined in a particular geographic region.

11 . A system for monitoring health of a population, the system comprising:

at least one memory storing instructions; and

at least one processor configured to execute the instructions to perform operations comprising:

receiving a query from a user requesting monitoring health of a population, wherein the query identifies at least one geographic region of a plurality of geographic regions;

in response to receiving the query, determining a current prevalence of at least one data variable or at least one health variable in the at least one geographic region, via a machine learning model trained to determine whether the at least one data variable or the at least one health variable is present in the at least one geographic region, by:

determining that the at least one data variable or the at least one health variable is present in a plurality of digital pathology images and/or reports input into the machine learning model over an interval of time, wherein the machine learning model is trained using whole slide images and diagnostic data corresponding to the plurality of geographic regions;

determining a relationship between the at least one geographic region, the at least one data variable, or the at least one health variable;

determining a first trend in the determined relationship at a first period of time based on the current prevalence of the at least one data variable or the at least one health variable in the at least one geographic region;

determining a second trend in the determined relationship at a second period of time;

generating an alert based on an irregularity between the first trend, the second trend, and a timing of the received query in the determined relationship being detected;

training the machine learning model to predict future relationships over time in the at least one geographic region; and

determining, based on the predicted future relationships, whether to continue monitoring health of a population.

12 . The system of claim 11 , the operations further comprising:

generating an updated relationship between the at least one geographic region, the at least one data variable, and/or the at least one health variable based on the determined relationship, comprising:

providing the at least one updated relationship to the machine learning model;

applying the machine learning model to predict at least one future relationship over a time period in the at least one geographic region; and

generating the updated relationship between the at least one geographic region, the at least one data variable, and/or the at least one health variable based on the at least one future relationship.

13 . The system of claim 12 , wherein the machine learning model comprises:

a convolutional neural network; a graph convolutional network, an autoregressive model, a recurrent neural network, and/or a capsule network.

14 . The system of claim 12 , wherein applying the machine learning model to predict at least one future relationship over a time period in the at least one geographic region comprises:

receiving time-stamped human subject data, associated with a plurality of patients, from the at least one geographic region;

inferring health related variables from the time-stamped human subject data; and

determining disease states and applying at least one natural language processing model to at least one clinic note to extract at least one relevant variable.

15 . The system of claim 14 , wherein the time-stamped human subject data comprises at least one of digital images of pathology specimens, genetic data, pathogenic data, and/or clinical notes.

16 . The system of claim 14 , the operations further comprising:

training a machine learning model to predict future relationships over time in a geographic region; and

determining, based on the predicted future relationships, whether to continue monitoring health of a population.

17 . The system of claim 12 , wherein applying a machine learning model to predict at least one future relationship over time in the at least one geographic region comprises using at least one of: a convolutional neural network; a graph convolution network; an autoregressive model; a recurrent neural network; and a capsule network.

18 . The system of claim 11 , wherein determining the at least one relationship between the at least one geographic region, at least one data variable, or at least one health variable comprises using correlation machine learning and/or geographic visual overlay.

19 . The system of claim 11 , wherein a query is received from a user for one of a particular health variable, a particular data variable.

20 . A non-transitory computer readable medium storing instructions that, when executed by a processor, cause the processor to perform operations for monitoring health of a population, the operations comprising:

receiving a query from a user requesting monitoring health of a population, wherein the query identifies at least one geographic region of a plurality of geographic regions;

in response to receiving the query, determining a current prevalence of at least one data variable or at least one health variable in the at least one geographic region, via a machine learning model trained to determine whether the at least one data variable or the at least one health variable is present in the at least one geographic region, by:

determining that the at least one data variable or the at least one health variable is present in a plurality of digital pathology images and/or reports input into the machine learning model over an interval of time, wherein the machine learning model is trained using whole slide images and diagnostic data corresponding to the plurality of geographic regions;

determining a relationship between the at least one geographic region, the at least one data variable, or the at least one health variable;

determining a first trend in the determined relationship at a first period of time based on the current prevalence of the at least one data variable or the at least one health variable in the at least one geographic region;

determining a second trend in the determined relationship at a second period of time;

generating an alert based on an irregularity between the first trend, the second trend, and a timing of the received query in the determined relationship being detected;

training the machine learning model to predict future relationships over time in the at least one geographic region; and

determining, based on the predicted future relationships, whether to continue monitoring health of a population.

Assignments (3)

Continuity (3)

References Cited (10)