US 12,288,215 B2
Multidimensional transaction pattern tracking and prediction
Gilbert Gatchalian, Union, NJ (US); Kevin A. Delson, Woodland Hills, CA (US); and Pratap Dande, St. Johns, FL (US)
Assigned to Bank of America Corporation, Charlotte, NC (US)
Filed by Bank of America Corporation, Charlotte, NC (US)
Filed on Dec. 1, 2022, as Appl. No. 18/072,918.
Prior Publication US 2024/0185248 A1, Jun. 6, 2024
Int. Cl. G06Q 20/40 (2012.01); G06Q 20/10 (2012.01)
CPC G06Q 20/4016 (2013.01) [G06Q 20/102 (2013.01); G06Q 20/4014 (2013.01)] 16 Claims
OG exemplary drawing
 
1. A multidimensional transaction pattern tracking and prediction engine, said engine comprising:
a central server in communication with a plurality of remote user devices, said central server comprising a set of machine-learning (ML) models, a processor, and a non-transitory memory storing computer executable instructions that, when run on the processor, are configured to:
detect, for a remote user device from the plurality of remote user devices, a series of digital transactions originating from an online account associated with the remote user device;
generate, via the set of ML models and based on the detected series of transactions, a multidimensional transactional profile for the remote user device, wherein generating the multidimensional transactional profile comprises:
passing the detected series of transactions through a first ML model to generate a first score representing a first dimension, wherein the first ML model is trained to detect transactional patterns along the first dimension;
passing the detected series of transactions through a second ML model to generate a second score representing a second dimension, wherein the second ML model is trained to detect transactional patterns along the second dimension;
selecting a first weighting for the first score and a second weighting for the second score; and
combining the weighted first score and the weighted second score to create the multidimensional transactional profile;
generate, based on the multidimensional transactional profile, a list of recommended transactional activities and a list of flagged transactional activities;
transmit the list of recommended transactional activities to the remote user device;
detect additional transactions originating from the account associated with the remote user device; and
in response to one of the additional transactions being from the list of flagged transactional activities, trigger an alert at the central server and freeze the account associated with the remote user device;
wherein:
the first and the second weightings are accessible on the remote user device; and
the remote user device is provided with a mechanism for adjusting the first and the second weightings.