Optum Services (Ireland) Limited

Various embodiments of the present disclosure provide feature engineering techniques for improving machine learning model development, performance, and maintenance. The feature engineering techniques include generating a model description vector from a textual model description for a target machine learning model and using the model description vector to generate one or more description-based similarity vectors that each include one or more of a plurality of similarity scores for a plurality of machine learning features. The feature engineering techniques include generating a label-based similarity vector based on a comparison between training data for the target machine learning model and a plurality of feature values of the plurality of machine learning features. The feature engineering techniques include providing a predictive feature set for the target machine learning model based on the one or more similarity vectors and the one or more label-based similarity vectors.

Various embodiments of the present disclosure provide computer forecasting techniques for initiating presentation of an interactive user interface. The techniques may include receiving one or more candidate prediction-based actions and generating a plurality of causal risk-based impact scores with respect to a candidate prediction-based action. The techniques include generating a plurality of causal quality-based impact scores and an action sequence for a plurality of evaluation entities and generating a causal net impact score based on (i) an aggregation of the plurality of causal risk-based impact scores and the plurality of causal quality-based impact scores and (ii) a sequence impact metric corresponding to the action sequence. The techniques include generating a sequence ranking for the action sequence and initiating a presentation of an interactive user interface reflective of the action sequence and the sequence ranking.

Various examples of the present disclosure utilize a sequential model pipeline to predict time-based effect output that improves the functionality of a computer in various aspects. The techniques comprise receiving a labeled output comprising a historical ground truth cause label and a historical ground truth effect label, generating a time-based cause output using a first machine learning model of the sequential model pipeline, generating a time-based effect output using a second machine learning model of the sequential model pipeline, and initiating a performance of a prediction-based action based on the time-based effect output.

Various embodiments of the present disclosure provide a technique for forecasting causal predictions that improves the functionality of a computer in various aspects. The techniques comprise receiving a historical sequence for a time-based prediction, generating, by a connected model framework, a first time-based output for a first time position in a prediction sequence for the time-based prediction, generating, using a directed acyclic graph, an output modification for a second time position in the prediction sequence; generating, using a machine learning model, a second time-based output for the second time position in the prediction sequence based on the output modification and the first time-based output, and initiating performance of a prediction-based action based on the prediction sequence.

Various embodiments of the present disclosure provide a data filtering and expansion technique that improves the functionality of a computer in various aspects. The techniques comprise generating a target object embedding for an unlabeled data object within a set of data objects that comprise a subset of unlabeled data objects and a subset of labeled data objects. The techniques comprise generating an embedding similarity score for the unlabeled data object and extracting the unlabeled data object from the subset of unlabeled data objects based on a comparison between the embedding similarity score and a quantile similarity threshold. The techniques comprise providing an expansion message that comprises the unlabeled data object to expand an initially filtered set of results.

Various embodiments of the present disclosure provide a similarity-based output transferring technique that improves the functionality of a computer in various aspects. The techniques comprise receiving a feature vector for a target data object of a set of data objects, generating a feature matrix for the set of data objects based on the feature vector; generating a similarity matrix for the target data object based on the feature matrix; generating a set of model coefficients by normalizing the similarity matrix based on a normalization threshold; generating a transferred prediction for the target data object by applying the set of model coefficients to a set of causal predictions that respectively correspond to the set of data objects; and initiating a performance of a prediction-based action based on the transferred prediction.

Various embodiments of the present disclosure provide techniques for improving causal inference modeling with respect to predictive labels in a complex predictive domain. The techniques of the present disclosure may include generating a positive cohort and a negative cohort of data objects from a dataset based on an associated with a predictive label, generating a positive cohort impact measure for the positive cohort and one or more negative cohort impact measures for the negative cohort, and generating an object impact measure for a particular data object of the negative cohort based on the positive cohort impact measure and at least one of the negative cohort impact measures.

Various embodiments of the present disclosure provide data storage, processing, and prediction techniques for providing predictive insights within large data prediction domains. The techniques may include generating, using a plurality of source tables for a prediction domain, a global graph for the prediction domain. The techniques may include generating, using a graph-based machine learning model, a plurality of node-level weights for the plurality of graph nodes based on a plurality of node attributes corresponding to the plurality of graph nodes. The techniques may include generating, using the graph-based machine learning model, a plurality of semantic-level weights for the plurality of weighted edges based on a designated predictive task for the global graph. The techniques may include generating plurality of graph node embeddings and initiating the performance of the designated predictive task based on the plurality of graph node embeddings.

Various embodiments of the present disclosure provide a data compression pipeline that improves the functionality of a computer in various aspects. The techniques comprise receiving an entity cohort file, extracting a cohort-level optimization dataset from the entity cohort file, generating, using a first and second predictive model, a scaled risk score and a scaled event score for an entity represented within the entity cohort file, generating a cohort score for the entity cohort file based on the scaled risk score, the scaled event score, and the subset of target entities of the cohort-level optimization dataset, and storing a compressed entity cohort file that identifies the cohort score, the subset of target entities, and an entity-level score for each entity within the subset of target entities.

Various embodiments of the present invention utilize systems, methods, and computer program products that perform measurement device calibration management by utilizing calibration offset generation machine learning models that are generated using a model training routine that comprises, for each measurement environment feature value: (i) determining a plurality of inferred measurements by a measurement device in relation to a ground-truth measurement operation via performing the ground-truth measurement operation under simulated measurement conditions characterized at least in part by varying a measurement environment feature that is associated with the measurement environment feature value across a per-feature spectrum for the measurement environment feature; and (ii) generating the calibration offset generation machine learning model based at least in part on comparing the plurality of inferred measurements and a ground-truth measurement output for the ground-truth measurement operation.