Granted Patent B2

US 12,471,827 · App. 17/026,092 · Granted Nov 18, 2025

Inventors: Brandon K. Fornwalt (Danville, PA); Christopher Haggerty (Danville, PA); Sushravya Raghunath (Danville, PA); Christopher Good (Danville, PA); John Pfeifer (Lewisburg, PA); Alvaro Ulloa-Cerna (Danville, PA); Arun Nemani (Chicago, IL); Tanner Carbonati (Chicago, IL); Ashraf Hafez (Wheaton, IL)

Assignees: Tempus AI, Inc.; Geisinger Clinic

A61B5/361A61B5/0006A61B5/282G16H10/60G16H50/20G16H50/30

Abstract

Claims (70)

1 . A method comprising:

receiving electrocardiogram data associated with a subject and an electrocardiogram configuration including a plurality of leads and a time interval, the electrocardiogram data comprising, for each lead included in the plurality of leads, voltage data associated with at least a portion of the time interval;

receiving an age value associated with the subject;

receiving a sex value associated with the subject;

providing the age value, the sex value, and at least a portion of the electrocardiogram data to a trained model, the at least a portion of the electrocardiogram data including first voltage data and second voltage data, the trained model including a first branch having a first convolutional block, a second branch having a second convolutional block, and a concatenation layer configured to generate, based on an output of the first and second branch, a concatenated output, the trained model being trained to generate a risk score based at least in part on the concatenated output and at least one of the age value and the sex value associated with the subject, wherein the first voltage data of the at least a portion of the electrocardiogram data is received at the first branch of the trained model and the second voltage data is received at the second branch of the trained model, the first voltage data consisting of voltage data associated with a first temporal range of the time interval, the second voltage data consisting of voltage data associated with a second, different temporal range of the time interval, and the risk score being generated at least partially based on the first voltage data and the second voltage data;

receiving, from the trained model, the risk score indicative of a likelihood the subject will suffer from a cardiovascular condition within a predetermined period of time from when the electrocardiogram data was generated; and

outputting the risk score to at least one of a memory or a display for viewing by a medical practitioner or healthcare administrator,

wherein the trained model is trained using training electrocardiogram data associated with a plurality of training subjects, the plurality of training subjects not having a diagnosis of the cardiovascular condition at the time when the training electrocardiogram data of the training subject was generated.

2 . The method of claim 1 further comprising:

receiving electronic health record data associated with the subject; and

providing at least a portion of the electronic health record data to the trained model.

3 . The method of claim 2 , wherein the electronic health record data comprises at least one of a blood cholesterol measurement, a blood cell count, a blood chemistries lab, a troponin level, a natriuretic peptide level, a blood pressure, a heart rate, a respiratory rate, an oxygen saturation, a cardiac ejection fraction, a cardiac chamber volume, a heart muscle thickness, a heart valve function, a diabetes diagnosis, a chronic kidney disease diagnosis, a congenital heart defect diagnosis, a cancer diagnosis, a procedure, a medication, a referral for cardiac rehabilitation, or a referral for dietary counseling.

4 . The method of claim 1 further comprising:

determining that the risk score is above a predetermined threshold associated with the cardiovascular condition;

in response to determining that the risk score is above the predetermined threshold, generating a report including information and/or links to sources associated with at least one of treatments for the cardiovascular condition or causes of the cardiovascular condition; and

outputting the report to at least one of a memory or a display for viewing by a medical practitioner or healthcare administrator.

5 . The method of claim 1 , wherein the period of time is one year.

6 . The method of claim 1 , wherein the period of time is selected from a range of one day to thirty years.

7 . The method of claim 1 , wherein the trained model further comprises a deep neural network comprising a plurality of branches.

8 . The method of claim 7 , wherein each branch of the plurality of branches is associated with a corresponding temporal range of the time interval, and wherein each branch receives voltage data of the portion of the electrocardiogram data consisting of voltage data associated with the temporal range corresponding to the branch of the plurality of branches.

9 . The method of claim 1 , wherein the trained model comprises a deep neural network comprising a convolutional component and a dense layer component.

10 . The method of claim 9 , wherein the convolutional component comprises an inception block comprising a plurality of convolutional layers.

11 . The method of claim 1 , wherein the plurality of leads comprises a lead I, a lead V2, a lead V4, a lead V3, a lead V6, a lead II, a lead VI, and a lead V5.

12 . The method of claim 11 , wherein the electrocardiogram data comprises first voltage data associated with the lead I and a first portion of the time interval, second voltage data associated with the lead V2 and a second portion of the time interval, third voltage data associated with the lead V4 and a third portion of the time interval, fourth voltage data associated with the lead V3 and the second portion of the time interval, fifth voltage data associated with the lead V6 and the third portion of the time interval, sixth voltage data associated with the lead II and the first portion of the time interval, seventh voltage data associated with the lead II and the second portion of the time interval, eighth voltage data associated with the lead II and the third portion of the time interval, ninth voltage data associated with the lead VI and the first portion of the time interval, tenth voltage data associated with the lead VI and the second portion of the time interval, eleventh voltage data associated with the lead VI and the third portion of the time interval, twelfth voltage data associated with the lead V5 and the first portion of the time interval, thirteenth voltage data associated with the lead V5 and the second portion of the time interval, and fourteenth voltage data associated with the lead V5 and the third portion of the time interval.

13 . The method of claim 12 , wherein the time interval comprises a ten second time period, the first portion of the time interval comprises a first half of the time interval, the second portion of the time interval comprises a third quarter of the time interval, and the third portion of the time interval comprises a fourth quarter of the time interval.

14 . The method of claim 12 , wherein the trained model comprises a first channel, a second channel, and a third channel, and the providing step comprises:

providing the first voltage data, the sixth voltage data, the ninth voltage data, and the twelfth voltage data to the first channel;

providing the second voltage data, the fourth voltage data, the seventh voltage data, the tenth voltage data, and the thirteenth voltage data to the second channel; and

providing the third voltage data, the fifth voltage data, the eighth voltage data, the eleventh voltage data, and the fourteenth voltage data to the third channel.

15 . The method of claim 11 , wherein each of the plurality of leads is associated with the time interval.

16 . The method of claim 1 , wherein the electrocardiogram data is indicative of a heart condition based on cardiological standards.

17 . The method of claim 1 , wherein the electrocardiogram data is not indicative of a heart condition based on cardiological standards.

18 . The method of claim 1 , wherein the cardiovascular condition is mortality.

19 . The method of claim 1 , wherein the cardiovascular condition is atrial fibrillation.

20 . A method comprising:

receiving subject electrocardiogram data associated with a subject and an electrocardiogram configuration including a plurality of leads and a time interval from an electrocardiogram device, the subject electrocardiogram data comprising, for each lead included in the plurality of leads, voltage data associated with at least a portion of the time interval;

providing at least a portion of the subject electrocardiogram data to a trained model, the at least a portion of the subject electrocardiogram data including first voltage data consisting of voltage data associated with a first temporal range of the time interval and second voltage data consisting of voltage data associated with a second, different temporal range of the time interval, the trained model including a first branch having a first convolutional block, a second branch having a second convolutional block, and a concatenation layer configured to generate, based on an output of the first and second branch, a concatenated output, the trained model being trained to output a risk score based at least in part on the concatenated output, wherein the first voltage data is received at the first branch of the trained model and the second voltage data is received at the second branch of the trained model, the first branch being configured to receive electrocardiogram data consisting of voltage data associated with the first temporal range, and the second branch being configured to receive electrocardiogram data consisting of voltage data associated with the second temporal range, wherein the first temporal range and the second temporal range do not overlap;

receiving, from the trained model, a risk score indicative of a likelihood the subject will suffer from a cardiovascular condition within a predetermined period of time from when the subject electrocardiogram data was generated;

generating a report based on the risk score; and

outputting the report to at least one of a memory or a display for viewing by a medical practitioner or healthcare administrator,

wherein the trained model is trained using training electrocardiogram data associated with a plurality of training subjects, the plurality of training subjects not having a diagnosis of the cardiovascular condition at the time when the training electrocardiogram data of the training subject was generated.

21 . A system, comprising:

at least one processor coupled to at least one memory comprising instructions, the at least one processor executing the instructions to:

receive electrocardiogram data associated with a subject and an electrocardiogram configuration including a plurality of leads and a time interval, the electrocardiogram data comprising, for each lead included in the plurality of leads, voltage data associated with at least a portion of the time interval;

provide at least a portion of the electrocardiogram data to a trained model including a first branch having a first convolutional block, a second branch having a second convolutional block, and a concatenation layer configured to generate, based on an output of the first and second branch, a concatenated output, the at least a portion of the electrocardiogram data including first voltage data and second voltage data, the trained model being trained to output a risk score based at least in part on the concatenated output, wherein the first voltage data is provided to the first branch, the first voltage data being temporally restricted within a first temporal range of the time interval, and the second voltage data is provided to the second branch, the second voltage data being temporally restricted within a second temporal range of the time interval, wherein the second temporal range does not overlap with the first temporal range;

receive, from the trained model, a risk score indicative of a likelihood the subject will suffer from a cardiovascular condition within a predetermined period of time from when the electrocardiogram data was generated from the trained model; and

output the risk score to at least one of a memory or a display for viewing by a medical practitioner or healthcare administrator,

wherein the trained model is trained using training electrocardiogram data associated with a plurality of training subjects, the plurality of training subjects not having a diagnosis of the cardiovascular condition at the time when the training electrocardiogram data of the training subject was generated.

22 . A method, comprising:

receiving electrocardiogram data associated with a subject and an electrocardiogram configuration including a plurality of leads and a time interval, the electrocardiogram data comprising, for each lead included in the plurality of leads, voltage data associated with at least a portion of the time interval;

receiving demographic data associated with the subject;

providing a first temporally coherent portion of the electrocardiogram data to a first branch of a trained model the first branch having a first convolutional block, the first temporally coherent portion consisting of first voltage data acquired during a first range of the time interval;

providing a second temporally coherent portion of the electrocardiogram data to a second branch of the trained model, the second branch having a second convolutional block, the second temporally coherent portion consisting of second voltage data acquired during a second range of the time interval, wherein the second range is subsequent to the first range;

providing the demographic data to the trained model;

generating, at the first branch a first output based on the first voltage data, and generating, at the second branch, a second output based on the second voltage data;

concatenating, at a concatenation layer, the first output and the second output with the demographic data;

generating, at the trained model, a risk score indicative of a likelihood the subject will suffer from a cardiovascular condition within a predetermined period of time from when the electrocardiogram data was generated based on the concatenated output and the demographic data;

receiving the risk score from the trained model; and

outputting the risk score to at least one of a memory or a display for viewing by a medical practitioner or healthcare administrator,

wherein the trained model is trained using training electrocardiogram data associated with a plurality of training subjects, the plurality of training subjects not having a diagnosis of the cardiovascular condition at the time when the training electrocardiogram data of the training subject was generated.

23 . The method of claim 22 , wherein the demographic data comprises a sex of the subject.

24 . The method of claim 22 , wherein the demographic data comprises an age of the subject.

25 . The method of claim 22 , wherein the cardiovascular condition is mortality.

26 . The method of claim 22 , wherein the cardiovascular condition is atrial fibrillation.

27 . The method of claim 22 , wherein the time period is at least six months.

28 . The method of claim 27 , wherein the time period is at least one year.

29 . The method of claim 22 , wherein the plurality of leads comprises a lead I, a lead V2, a lead V4, a lead V3, a lead V6, a lead II, a lead VI, and a lead V5.

30 . The method of claim 22 further comprising:

generating a report based on the risk score; and

outputting the report to the display for viewing by a medical practitioner or healthcare administrator.

Assignments (6)

Continuity (4)

References Cited (138)