US 2022/0019913 A1Published application

Weighted adaptive filtering based loss function to predict the first occurrence of multiple events in a single shot

AssigneeOptum Technology, Inc.

InventorsV Kishore Ayyadevara, Sree Harsha Ankem, Raghav Bali, Rohan Khilnani, Vineet Shukla, Saikumar Chintareddy, Ranraj Rana Singh

Life of this patent

Estimating the term…

patentgrep

Loading patent…

Abstract

There is a need for more effective and efficient prediction data analysis. This need can be addressed by, for example, solutions for performing first-occurrence multi-disease prediction. In one example, a method includes determining a per-event-type loss value for each event type of a group of event types; determining a cross-event-type loss value based at least in part on each per-event-type loss value; training a multi-event-type prediction model based at least in part on the cross-event type loss value; generating a first-occurrence prediction based at least in part on the multi-event-type prediction model, wherein the first occurrence-prediction comprises a first-occurrence prediction item for each event type of the group of event types; and performing one or more prediction-based actions based at least in part on the first-occurrence prediction.

Claims20

Dependent claims

A computer-implemented method for generating a first-occurrence prediction for a prediction input set in relation to a plurality of event types, the computer-implemented method comprising:

identifying a plurality of training data fields, wherein each training data field of the plurality of training data fields is associated with a training input set and an event label set;
processing each training input set for a training data field of the plurality of training data fields using a multi-event prediction model to determine an inferred event probability set for the training data field;
for each event type of the plurality of event types:
determining a related subset of the plurality of training data fields for the event type, wherein a training data field is in the related subset if the event label set for the training data field describes that the training data field describes no-occurrence of the event type in relation to the training data field or first-occurrence of the event type in relation to the training data field;
and
determining a per-event-type loss value for the event type based at least in part on each event label set for a training data field in the related subset and each inferred event probability set for a training data field in the related subset;
determining, based at least in part on each per-event-type loss value for an event type of the plurality of event types, a cross-event-type loss value;
training the multi-event-type prediction model based at least in part on the cross-event type loss value;
generating the first-occurrence prediction based at least in part on the multi-event-type prediction model, wherein the first occurrence-prediction comprises a first-occurrence prediction item for each event type of the plurality of event types;
and
performing one or more prediction-based actions based at least in part on the first-occurrence prediction.

8 more elements

An apparatus for generating a first-occurrence prediction for a prediction input set in relation to a plurality of event types, the apparatus comprising at least one processor and at least one memory including program code, the at least one memory and the program code configured to, with the processor, cause the apparatus to at least:

identify a plurality of training data fields, wherein each training data field of the plurality of training data fields is associated with a training input set and an event label set;
process each training input set for a training data field of the plurality of training data fields using a multi-event prediction model to determine an inferred event probability set for the training data field;
for each event type of the plurality of event types:
determine a related subset of the plurality of training data fields for the event type, wherein a training data field is in the related subset if the event label set for the training data field describes that the training data field describes no-occurrence of the event type in relation to the training data field or first-occurrence of the event type in relation to the training data field;
and
determine a per-event-type loss value for the event type based at least in part on each event label set for a training data field in the related subset and each inferred event probability set for a training data field in the related subset;
determine, based at least in part on each per-event-type loss value for an event type of the plurality of event types, a cross-event-type loss value;
train the multi-event-type prediction model based at least in part on the cross-event type loss value;
generate the first-occurrence prediction based at least in part on the multi-event-type prediction model, wherein the first occurrence-prediction comprises a first-occurrence prediction item for each event type of the plurality of event types;
and
perform one or more prediction-based actions based at least in part on the first-occurrence prediction.

8 more elements

A computer program product for generating a first-occurrence prediction for a prediction input set in relation to a plurality of event types, the computer program product comprising at least one non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions configured to:

identify a plurality of training data fields, wherein each training data field of the plurality of training data fields is associated with a training input set and an event label set;
process each training input set for a training data field of the plurality of training data fields using a multi-event prediction model to determine an inferred event probability set for the training data field;
for each event type of the plurality of event types:
determine a related subset of the plurality of training data fields for the event type, wherein a training data field is in the related subset if the event label set for the training data field describes that the training data field describes no-occurrence of the event type in relation to the training data field or first-occurrence of the event type in relation to the training data field;
and
determine a per-event-type loss value for the event type based at least in part on each event label set for a training data field in the related subset and each inferred event probability set for a training data field in the related subset;
determine, based at least in part on each per-event-type loss value for an event type of the plurality of event types, a cross-event-type loss value;
train the multi-event-type prediction model based at least in part on the cross-event type loss value;
generate the first-occurrence prediction based at least in part on the multi-event-type prediction model, wherein the first occurrence-prediction comprises a first-occurrence prediction item for each event type of the plurality of event types;
and
perform one or more prediction-based actions based at least in part on the first-occurrence prediction.

8 more elements

Dependencies are inferred from the claim text.