Optum, Inc.

A method comprises receiving an incident report comprising a textual description of an incident; generating a regularized incident report in which out-of-vocabulary terms in the received incident report are replaced with in-vocabulary terms; determining importance measures for a plurality of incident report terms, wherein each of the incident report terms is in the regularized incident report; generating an incident matrix in which similarity values are defined for combinations of terms in the incident report and terms in a predetermined term set; generating an incident vector based on the incident matrix and the importance measures for the terms in the incident report; applying one or more machine learning (ML) models that identify, based on the incident vector, relevant software support records and/or software modules, wherein the relevant software support records and the software modules are potentially relevant to the incident; and outputting data identifying relevant software support records and/or software modules.

Techniques for data clustering and outcome generation may comprise determining a state sequence associated with a sample of one or more samples that satisfies an occurrence threshold and generating one or more timing features for the state sequence. The techniques may further comprise clustering, by executing a machine-learned model, the sample corresponding to at least the state sequence into a cluster of a set of clusters based on the state sequence and the one or more timing features. The techniques may further comprise generating a data object indicating one or more outcomes associated with the cluster that includes the sample. These data objects indicate outcomes that are specific enough to provide meaningful insights contained in associations between the input data and the associated clusters and eliminate/reduce the misleading information commonly output by typical techniques.

Systems and computer-implemented methods are disclosed for detecting a system anomaly. A computer-implemented method comprises: receiving, by a data storage module, time-series data from a plurality of sensors of an information technology infrastructure, each sensor corresponding to a respective metric; detecting a plurality of anomalies in the time-series data stored in the data storage module; generating a knowledge graph by: determining connections between the plurality of metrics based on the time-series data; and for each connection, determining a respective weight based on an impact score for metrics joined by the connection; and configuring a root cause determination engine to output one or more metrics as root cause candidates, the one or more metrics based on the knowledge graph, in response to input of a query associated with at least one metric.

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.

Various embodiments of the present disclosure provide image and prediction processing techniques for providing improved image-based predictions. The techniques may include generating a plurality of derivative images from a production line image by cutting the production line image into a plurality of portions. The techniques include generating a plurality of comparable derivative images by rotating each of the plurality of derivative images to a particular orientation relative to a production line item reflected by the production line image. The techniques include generating, using a machine learning model, an anomaly prediction for the production line image based on an image comparison between the plurality of comparable derivative images. The techniques include initiating the performance of the prediction-based action based on the anomaly prediction.

Various embodiments of the present invention provide methods, apparatus, systems, computing devices, computing entities, and/or the like for performing predictive data analysis with respect to structured data objects. Certain embodiments of the present invention utilize systems, methods, and computer program products that perform predictive data analysis with respect to structured data objects by utilizing at least one of cross-table data similarity score generation machine learning models and unsupervised anomalous table row detection machine learning models.

A computer-implemented method includes receiving first query data associated with a first query, identifying a system error that is associated with the first unique entity identifier based on an indication that the first query was improperly approved, extracting a cause of the system error from the first query data, determining a machine-readable invalidity code that corresponds to the cause of the system error, updating a system error database with the first unique entity identifier and the machine-readable invalidity code as a first pair, and outputting, the first pair including the first unique entity identifier and the machine-readable invalidity code from the system error database, the first pair being configured to be processed by a front-end system for identifying a potential system error associated with second query data for a second query.

Systems and methods are disclosed for generating training data for a machine learning model. The method includes receiving, from one or more data sources, a first machine learning training data set that includes a plurality of data points; determining an input outlier score of a first data point of the plurality of data points; determining an output outlier score of the first data point; generating a total output score of the first data point based on the input outlier score of the first data point and the output outlier score of the first data point, the total output score of the first data point representing a likelihood that the first data point is an inconsistently annotated data point; comparing the total output score of the first data point with a pre-determined threshold; based on the comparison, generating a second machine learning training data set that excludes the first data point.

Systems and methods are disclosed for training a machine-learning model to predict and manage a condition of an entity. The method includes receiving historical data associated with a target entity from a plurality of data sources; deriving feature(s) from the historical data; determining a condition of the target entity by applying the feature(s) to a machine-learning model trained by: receiving a plurality of datasets associated with each entity of a plurality of entities; determining a specific condition associated with each entity based on the plurality of datasets; generating an identifier for each entity based on the determined specific condition; deriving training feature(s) for each entity from historical training data associated with the entity; and inputting the identifier and the training feature(s) for each entity to the machine-learning model to learn associations between the identifiers and the training feature(s) associated with the plurality of entities.

Various embodiments of the present invention provide methods, apparatus, systems, computing devices, computing entities, and/or the like for performing predictive data analysis operations. For example, certain embodiments of the present invention utilize systems, methods, and computer program products that perform predictive data transformation operations by machine learning models, where the predictive data transformation is performed based at least in part on a cross comparison of a pair of columns, accounting for the similarity of both the column names and the column values, inferred by the outputs of a machine learning model. Additionally, certain embodiments of the present invention utilize systems, methods, and computer program products that perform anomaly detection by using machine learning models that operate based at least in part on a comparison of the fixed-size representation of the column values resulting from the machine learning model.