Optum, Inc.

Unitedhealth Group Incorporated

Christopher A Hane

Vijay S Nori

Louis J Rumanes

Paul Bleicher

David R Anderson

Alexander Kravetz

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 categorical data objects. Certain embodiments of the present invention utilize systems, methods, and computer program products that perform predictive data analysis with respect to categorical data objects by utilizing at least one of predictive feature hierarchies, feature refinement routines, decision subsets of predictive features that are generated based at least in part on predictiveness measures for the predictive features, and/or the like.

Machine learning techniques using iterative feature refinement routines

Various embodiments of the disclosure provide apparatuses, systems, and computer program products for predictive data labelling using a dual-model system. Embodiments provide various advantages in accuracy of predicted labels, for example in various contexts such as medical data analysis for difficult to diagnose diseases. An example provided apparatus is configured to generate a positive, neutral, and negative candidate identifier sets and corresponding positive, neutral, and negative candidate index sets based in part on applying a candidate selection rule set to a candidate data set; train a candidate label probabilistic model based at least in part on a candidate label training subset associated with the candidate data set associated with the positive and negative candidate identifiers; generate a candidate positive-label probability set using at least the candidate label probabilistic model; train a historical record prediction model to predict the candidate positive-label probability set; and utilize the historical record prediction model.

Apparatus, computer program product, and method for predictive data labelling using a dual-prediction model system

There is a need for solutions that perform predictive natural language processing with improved efficiency and/or accuracy. This need can be addressed by, for example, by identifying an indexed representation of a natural language object; obtaining a vocabulary domain associated with one or more first phrases; determining an individual frequency for each first phrase based on a count of occurrences of the first phrase in the indexed representation; identifying one or more dominant phrases of the first phrases; for each dominant phrase, identifying any dependent phrases for the first dominant phrase; determining a semantically-adjusted frequency for each dominant phrase based on the individual frequency for the dominant phrase and each individual frequency for any dependent phrase for the dominant phrase; generating a structured representation of the natural language object based on each semantically-adjusted frequency associated with a dominant phrase; and providing the structured representation for the predictive analysis.

Predictive natural language processing using semantic feature extraction

Aggregate vectors corresponding to non-textual information/data are provided in a multi-dimensional space. A computing entity access a plurality of instances of medical information comprising medical codes. The computing entity generates one or more medical sentences from the plurality of instances of medical information. Each medical sentence comprises one or more medical codes. The computing entity generates an embedding vector dictionary comprising a plurality of multi-dimensional vectors based on a medical embedding model trained using machine learning and the one or more medical sentences. Each multi-dimensional vector corresponds to a medical code. The computing entity generates a plurality of aggregate vectors based on the embedding vector dictionary and analyzes at least a portion of the plurality of aggregate vectors to identify two or more aggregate vectors that are similar or different based on a distance between the two or more aggregate vectors in the multi-dimensional space.

Code vector embeddings for similarity metrics

Efficient SQL based multi-attribute clustering of data attributes may be used to identify the most relevant combination of data attributes to an outcome. A global outcome value may be calculated to represent an average of the outcome. A subset outcome value for each subset of data attributes of a plurality of attributes may be calculated to represent average of the outcome for the subset. For each subset of data attributes, a number of members associated with the subset may be compared to a threshold, and the subsets with less members than the threshold may be removed. The subset outcome value for each subset of data attributes may be compared to the global outcome value, and a report may be generated that identifies each subset for which the corresponding subset outcome value is greater than or less than the global outcome value.

Growing community of inventors

Irvine, CA, United States of America

Christopher A Hane