Granted Patent B1

US 12,530,404 · App. 17/684,166 · Granted Jan 20, 2026

Inventor: Sears Merritt (Groton, MA)

Assignee: Massachusetts Mutual Life Insurance Company

G06F16/9027G06F16/951G06N20/00H04L63/0428H04L67/2895H04L67/306H04L67/56H04L67/01

Abstract

Claims (39)

1 . A method comprising:

generating, by a processor, a feature vector corresponding to one or more attributes of a user received via at least one of one or more input elements displayed on a computing device operated by the user or retrieved from one or more databases;

determining, by the processor, a data structure type, from a plurality of data structure types indicating different methods of data organization, of the feature vector;

searching, by the processor, a plurality of machine-learning models stored in a database, wherein each machine-learning model corresponds to a data structure type from the plurality of data structure types to identify a mortality machine-learning model that is suitable for processing the data structure type of the feature vector;

executing, by the processor, the mortality machine-learning model to determine a likelihood of mortality for the user by applying the feature vector to the mortality machine-learning model, wherein the mortality machine-learning model is trained based on a machine-learning technique and a respective survival time in each of a set of historic user data records and their corresponding attributes;

calculating, by the processor, contribution of at least one first factor analyzed by the mortality machine-learning model that contributed to increasing the likelihood of mortality;

calculating, by the processor, contribution of at least one second factor analyzed by the mortality machine-learning model that contributed to decreasing the likelihood of mortality; and

updating, by the processor, an electronic page based on the likelihood satisfying a threshold, wherein the updated page includes:

an indication of the at least one first factor and its corresponding contribution to the likelihood of mortality, and at least one second factor and its corresponding contribution to the likelihood of mortality.

2 . The method of claim 1 , wherein the one or more attributes correspond to at least one of an age of the user, a blood disorder, a status of the user being a smoker, medical diagnostic tests, gamma glutamyltransferase, aspartate transaminase (AST) and alanine transaminase (ALT) ratio, urine based nicotine metabolites, globulin, prescription drugs, blood urea nitrogen, creatinine, urine total protein, alkaline phosphatase, hemoglobin, systolic blood pressure, disability claims, diastolic blood pressure, motor vehicle violations, estimated glomerular filtration rate, glucose, anti hcv (hepatitus c), vascular disorders, cholesterol, triglycerides, urine specific gravity, urine, respiratory disorders, cocaine metabolites, albumin globulin ratio, urine creatinine, high density lipoprotein, mental disorders, family history of diabetes, family history of vascular disorders, pulse at rest, urine glucose, diabetes, height, prostate specific antigen, weight, cancer disorder, serum glutamic oxaloacetic transaminase AST, body mass index, total bilirubin, family history of cancer, serum glutamic-pyruvic transaminase ALT, urine pc ratio, fructosamine, HIV positive, prescription drug history, credit risk history, physical activity history, or medical examination history.

3 . The method of claim 1 , wherein the mortality machine-learning model is a risk model pre-generated based on a set of dynamically updateable database records for a set of users excluding the user operating the computing device.

4 . The method of claim 1 , wherein the database is an in-memory database.

5 . The method of claim 1 , wherein the feature vector is generated via an application programming interface.

6 . A computer system comprising:

a server in communication with a computing device, the server configured to:

generate a feature vector corresponding to one or more attributes of a user received via at least one of one or more input elements displayed on the computing device operated by the user or retrieved from one or more databases;

determine a data structure type, from a plurality of data structure types indicating different methods of data organization, of the feature vector;

search, based on the data structure type of the feature vector, a plurality of machine-learning models stored in a database, wherein each machine-learning model corresponds to a data structure type from the plurality of data structure types to identify a mortality machine-learning model that is suitable for processing the data structure type of the feature vector;

execute the mortality machine-learning model to determine a likelihood of mortality for the user by applying the feature vector to the mortality machine-learning model, wherein the mortality machine-learning model is trained based on a machine-learning technique and a respective survival time in each of a set of historic user data records and their corresponding attributes;

calculate contribution of at least one first factor analyzed by the mortality machine-learning model that contributed to increasing the likelihood of mortality;

calculate contribution of at least one second factor analyzed by the mortality machine-learning model that contributed to decreasing the likelihood of mortality; and

update an electronic page based on the likelihood satisfying a threshold, wherein the updated page includes:

an indication of the at least one first factor and its corresponding contribution to the likelihood of mortality, and at least one second factor and its corresponding contribution to the likelihood of mortality.

7 . The computer system of claim 6 , wherein the one or more attributes correspond to at least one of an age of the user, a blood disorder, a status of the user being a smoker, medical diagnostic tests, gamma glutamyltransferase, aspartate transaminase (AST) and alanine transaminase (ALT) ratio, urine based nicotine metabolites, globulin, prescription drugs, blood urea nitrogen, creatinine, urine total protein, alkaline phosphatase, hemoglobin, systolic blood pressure, disability claims, diastolic blood pressure, motor vehicle violations, estimated glomerular filtration rate, glucose, anti hcv (hepatitus c), vascular disorders, cholesterol, triglycerides, urine specific gravity, urine, respiratory disorders, cocaine metabolites, albumin globulin ratio, urine creatinine, high density lipoprotein, mental disorders, family history of diabetes, family history of vascular disorders, pulse at rest, urine glucose, diabetes, height, prostate specific antigen, weight, cancer disorder, serum glutamic oxaloacetic transaminase AST, body mass index, total bilirubin, family history of cancer, serum glutamic-pyruvic transaminase ALT, urine pc ratio, fructosamine, HIV positive, prescription drug history, credit risk history, physical activity history, or medical examination history.

8 . The computer system of claim 6 , wherein the mortality machine-learning model is a risk model pre-generated based on a set of dynamically updateable database records for a set of users excluding the user operating the computing device.

9 . The computer system of claim 6 , wherein the database is an in-memory database.

10 . The computer system of claim 6 , wherein the feature vector is generated via an application programming interface.

11 . A system comprising:

a server comprising a processor and a non-transitory computer-readable medium containing instructions that when executed by the processor causes the processor to perform operations comprising:

generating, by a processor, a feature vector corresponding to one or more attributes of a user received via at least one of one or more input elements displayed on a computing device operated by the user or retrieved from one or more databases;

determining, by the processor, a data structure type, from a plurality of data structure types indicating different methods of data organization, of the feature vector;

searching, by the processor, a plurality of machine-learning models stored in a database, wherein each machine-learning model corresponds to a data structure type from the plurality of data structure types to identify a mortality machine-learning model that is suitable for processing the data structure type of the feature vector;

executing, by the processor, the mortality machine-learning model to determine a likelihood of mortality for the user by applying the feature vector to the mortality machine-learning model, wherein the mortality machine-learning model is trained based on a machine-learning technique and a respective survival time in each of a set of historic user data records and their corresponding attributes; and

calculating, by the processor, contribution of at least one first factor analyzed by the mortality machine-learning model that contributed to increasing the likelihood of mortality;

calculating, by the processor, contribution of at least one second factor analyzed by the mortality machine-learning model that contributed to decreasing the likelihood of mortality; and

updating, by the processor, an electronic page based on the likelihood satisfying a threshold, wherein the updated page includes:

an indication of the at least one first factor and its corresponding contribution to the likelihood of mortality, and at least one second factor and its corresponding contribution to the likelihood of mortality.

12 . The system of claim 11 , wherein the one or more attributes correspond to at least one of an age of the user, a blood disorder, a status of the user being a smoker, medical diagnostic tests, gamma glutamyltransferase, aspartate transaminase (AST) and alanine transaminase (ALT) ratio, urine based nicotine metabolites, globulin, prescription drugs, blood urea nitrogen, creatinine, urine total protein, alkaline phosphatase, hemoglobin, systolic blood pressure, disability claims, diastolic blood pressure, motor vehicle violations, estimated glomerular filtration rate, glucose, anti hcv (hepatitus c), vascular disorders, cholesterol, triglycerides, urine specific gravity, urine, respiratory disorders, cocaine metabolites, albumin globulin ratio, urine creatinine, high density lipoprotein, mental disorders, family history of diabetes, family history of vascular disorders, pulse at rest, urine glucose, diabetes, height, prostate specific antigen, weight, cancer disorder, serum glutamic oxaloacetic transaminase AST, body mass index, total bilirubin, family history of cancer, serum glutamic-pyruvic transaminase ALT, urine pc ratio, fructosamine, HIV positive, prescription drug history, credit risk history, physical activity history, or medical examination history.

13 . The system of claim 11 , wherein the mortality machine-learning model is selected based on the processor analyzing the feature vector.

Assignments (1)

Continuity (2)

References Cited (48)