US 12,353,964 B2
Cross-entity similarity determinations using machine learning frameworks
Daniel G. McCreary, St. Louis Park, MN (US); Parker J. Erickson, Plymouth, MN (US); Mark G. Megerian, Rochester, MN (US); and Alex Li, Maple Grove, MN (US)
Assigned to OPTUM, INC., Minnetonka, MN (US)
Filed by Optum, Inc., Minnetonka, MN (US)
Filed on May 4, 2021, as Appl. No. 17/302,481.
Prior Publication US 2022/0358395 A1, Nov. 10, 2022
Int. Cl. G06N 20/00 (2019.01); G06N 5/02 (2023.01); G06N 5/04 (2023.01)
CPC G06N 20/00 (2019.01) [G06N 5/027 (2013.01); G06N 5/04 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
generating, by one or more processors and based on a group of graph node sequences from a predefined number of random walks across an entity relationship graph corresponding to a first predictive entity and a second predictive entity, a graph-based feature set;
generating, by the one or more processors and by processing the graph-based feature set using a graph node embedding machine learning model, a first graph-based entity embedding for the first predictive entity and a second graph-based entity embedding for the second predictive entity;
generating, by the one or more processors and by processing the first graph-based entity embedding and the second graph-based entity embedding using an entity encoding machine learning model, a first predictive entity embedding and a second predictive entity embedding, wherein the first predictive entity embedding and the second predictive entity embedding each comprise one or more values corresponding to each dimension of a multidimensional embedding space;
determining, by the one or more processors and using a similarity determination machine learning model to process the first predictive entity embedding and the second predictive entity embedding, a predicted cross-entity similarity measure;
updating, by the one or more processors, a queue of a plurality of data objects to include a predictive data object describing the second predictive entity, wherein the plurality of data objects are included in the queue according to a cross-entity similarity measure condition;
generating, by the one or more processors, at least one API-based data object corresponding to at least a portion of the queue; and
transmitting, by the one or more processors, the at least one API-based data object to an end user interface.