US 12,737,821 B2Granted

Provider activity anomaly detection

AssigneeOptum, Inc.

InventorsRakesh P A, Vijaychandar Natesan, Kevin Obrien, Scott R. Vandeleest, Pawan Kumar Pandey, Ramesh Rangasamy Ganesan, Arun Kumar Kumar, Guntupalli Sumanth Naga Deepak, Aman Kharb, Diane T. Smith, Rahimunnisa Pathan

Life of this patent

Estimating the term…

patentgrep

Loading patent…

Abstract

Techniques for automatically detecting inactive addresses of providers using unsupervised learning approaches are provided. The techniques include determining an activity trend of an address. Responsive to determining that the address is associated with increasing activity, determining an active metric of the address, which indicates a likelihood of the address being active. Responsive to determining that the address is associated with decreasing activity, determining an inactive metric of the address, which indicates a likelihood of the address being inactive. The techniques further include determining whether the address is active or inactive based on the active metric and/or inactive metric. In some embodiments, the active metric or inactive metric is a weighted sum of z scores determined based on Gaussian distributions generated with respect to various benchmarks or provider features. In some embodiments, the weights used to determine the active metric or inactive metric are determined using RLHF techniques.

Claims20

Dependent claims

A system comprising:

one or more processors;
and
one or more memories storing processor-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
receiving a claim data set indicative of a claim count associated with an address of a provider over a particular time period;
determining, using a machine learning model trained using unsupervised learning techniques, that the address is associated with decreasing activity based at least in part on the claim data set, wherein the machine learning model comprises a plurality of parameters, and wherein the machine learning model is trained at least in part by:
(i) inputting, into the machine learning model, one or more claim counts associated with one or more addresses for a first time period, to obtain one or more outputs indicative of one or more predicted claim counts associated with the one or more addresses, respectively;
(ii) updating the plurality of parameters based on one or more differences between the one or more outputs and one or more claim counts of a second period subsequent to the first time period, respectively;
and
(iii) repeating operations (i) and (ii) until a predetermined condition is satisfied;
responsive to determining that the address is associated with decreasing activity, determining an inactive metric for the address based at least in part on the claim data set;
determining that the inactive metric fails to satisfy a first condition;
responsive to determining that the inactive metric fails to satisfy the first condition, determining an active metric for the address based at least in part on the claim data set;
determining that the active metric satisfies a second condition;
and
responsive to determining that the active metric satisfies the second condition, causing one or more data objects to indicate that the address is active.

12 more elements

A method comprising:

receiving, by one or more processors, a claim data set indicative of a claim count associated with an address of a provider over a particular time period;
determining, by the one or more processors and using a machine learning model trained using unsupervised learning techniques, that the address is associated with decreasing activity based at least in part on the claim data set, wherein the machine learning model comprises a plurality of parameters, and wherein the machine learning model is trained at least in part by:
(i) inputting, into the machine learning model, one or more claim counts associated with one or more addresses for a first time period, to obtain one or more outputs indicative of one or more predicted claim counts associated with the one or more addresses, respectively;
(ii) updating the plurality of parameters based on one or more differences between the one or more outputs and one or more claim counts of a second period subsequent to the first time period, respectively;
and
(iii) repeating operations (i) and (ii) until a predetermined condition is satisfied;
responsive to determining that the address is associated with decreasing activity, determining, by the one or more processors, an inactive metric for the address based at least in part on the claim data set;
determining, by the one or more processors, that the inactive metric fails to satisfy a first condition;
responsive to determining that the inactive metric fails to satisfy the first condition, determining, by the one or more processors, an active metric for the address based at least in part on the claim data set;
determining, by the one or more processors, that the active metric satisfies a second condition;
and
responsive to determining that the active metric satisfies the second condition, causing, by the one or more processors, one or more data objects to indicate that the address is active.

9 more elements

Dependencies are inferred from the claim text.

Facts

Application: 18/971,854Filed 6 Dec 2024 · us provisional application
Priority: 21 Aug 2024
Published: US 2026/0057453 A126 Feb 2026
Granted: 15 Sep 2026
Term ends: Not estimatedThe source records don’t support an estimate.

Classification8 codes

G06Q 40/084
Insurance: by insurance claims processing
G06N 3/0442
Recurrent networks, e.g. Hopfield networks: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
G06N 3/088
Learning methods: non-supervised learning, e.g. competitive learning
G06N 7/01
Computing arrangements based on specific mathematical models: probabilistic graphical models, e.g. probabilistic networks