International Business Machines Corporation

Azka Javaid

Paul R Bastide

Fang Lu

Jayanth Krishnan

A plurality of instances of a patient's health condition are received as patient health information from a plurality of sources. Ground-truth health conditions of the patient are generated based on the patient health information and expert review of the patient's level of engagement for each instance of provider assessment. A regression model predicting a probability of patient engagement level with a provider is generated, based on the instances of patient condition assessment and the engagement level expert review associated with each instance. A current set of patient information and natural language processing results of a dialog between the patient and a provider performing a condition assessment of the patient, are received. The current patient information and the dialog are applied to the regression model, determining a probability of patient engagement level. An indicator of the patient engagement level is sent to the provider's computing device.

Determining engagement level of an individual during communication

A computer system adapts a model analyzing data. Information sources are analyzed to determine one or more changes for a computerized model employed for analyzing data. One or more current projects each using an implementation of the computerized model with at least one of the determined changes are identified. The implementations are compared to the employed computerized model to determine differences. One or more adaptations for the employed computerized model are determined in response to the determined differences satisfying a threshold, wherein the one or more adaptations for the employed computerized model are based on the determined changes in the corresponding implementation of the computerized model. At least one adaption is installed into a platform hosting the employed model for modification of the employed model. Embodiments of the present invention further include a method and program product for adapting a model analyzing data in substantially the same manner described above.

Intelligent framework updater to incorporate framework changes into data analysis models

Dynamic prioritization of an action item can include retrieving an electronically rendered text from one or more sampled text sources in response to recognizing a user-related task conveyed within the electronically rendered text. Dynamic prioritization of an action item can also include generating an action item corresponding to the user-related task and prioritizing the action item by assigning to the action item a priority computed based on one or more priority factors corresponding to the user-related task. Each of the one or more priority factors can be automatically selected from a collection of predetermined priority factors. Additionally, dynamic prioritization of an action item can include recomputing the priority assigned to the action item in response to a change in task-related circumstances, as determined based on other electronically rendered text.

Growing community of inventors

Brooklyn, NY, United States of America

Azka Javaid