US 11,386,126 B1Granted

System and method for linking data records from different databases

AssigneeVeeva Systems Inc.

InventorsLorenz Lehmann, Igor Praznik, Alexander Darius Emami-Nemini

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods for linking HCPs from an MDM system and publication authors from a second source data system. Distances between the HCPs and the publication authors may be determined, and an HCP and a publication author may be connected if their distance is below a threshold. The connected data points may be used to train a machine learning model to determine a probability that an HCP matches a publication author. An upper bound and a lower bound may be set. Data points with the probability over the upper bound may be sent out as matches, data points below the lower bound may be disregarded as not a match, and data points between the upper bound and lower bound may be labeled for manual verification.

Claims12

Dependent claims

A computer-implemented method for linking data records in a first category from a first source data system and data records in a second category from a second source data system, the method comprising:

at a data linking controller in a server, receiving a plurality of the data records in the first category from the first source data system, wherein a data record in the first category comprises a first type of element and a second type of element;
at the data linking controller, receiving a plurality of data records in the second category from the second source data system, wherein a data record in the second category comprises the first type of element and the second type of element, and wherein at least a data record in the first category from the first source data system and at least a data record in the second category from the second source data system are about a same individual;
determining by the data linking controller, distances between the first data record in the first category and each of the plurality of data records in the second category;
determining by the data linking controller distances between the first data record in the second category and each of the plurality of data records in the first category;
selecting by the data linking controller the distances below a first threshold;
creating by the data linking controller a connection for each selected distance;
creating by the data linking controller a data point set, wherein each data point represents a data record in the first category, and a data record in the second category which has a connection with the data record in the first category;
receiving classified data points;
and
training a machine learning model for determining a probability that the first data record in the first category matches the first data record in the second category.

7 more elements

Dependencies are inferred from the claim text.