Electronic health records agent application for highlighting contextually relevant information
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12488863-B1
Publication Date
2025-12-02
Expiration Date
Abstract
An improved electronic health records agent application (EHR agent) is disclosed herein. A server EHR agent receives an identifier for a healthcare worker and an identifier for a patient from a client computing device. The server EHR agent retrieves an audit table for the healthcare worker using the identifier for the healthcare worker and a virtual patient object (VPO) for the patient using the identifier for the patient. The audit table comprises a history of interactions of the healthcare worker with the EHR agent. The VPO comprises domains of clinical data for the patient aggregated from a plurality of sources. The server EHR agent provides the audit table and the VPO to a computer-implemented machine learning model, wherein the machine learning model identifies a subset of domains in the domains of clinical data that is likely to be used by the healthcare worker in treating the patient.
Core Innovation
The disclosed invention provides a server electronic health records agent application that accesses health records for patients that are inaccessible to an electronic health records application executing on a second server computing device. The server EHR agent receives an identifier for a healthcare worker and an identifier for a patient from a client computing device operated by the healthcare worker, with the patient currently being treated at a point of care, and retrieves an audit table that comprises a history of interactions of the healthcare worker with a client electronic health records agent application.
The server EHR agent retrieves a virtual patient object for the patient. The virtual patient object comprises identifiers for domains and clinical data for the patient aggregated from a plurality of sources, where the clinical data is stored and organized within the virtual patient object according to domains and each domain includes a different type of clinical data. The server EHR agent provides the audit table and the virtual patient object as input to a computer-implemented machine learning model with learned weights assigned thereto, and the model identifies a first domain and a second domain within the domains.
The first domain is determined to include a first subset of clinical data likely to be used by the healthcare worker at the point of care, while the second domain is determined to include a second subset of clinical data not likely to be used by the healthcare worker at the point of care. Based on the model-identified first and second domains, the server EHR agent generates an optimized virtual patient object that includes markings for the domain identifiers, transmits the optimized virtual patient object to the client EHR agent, and presents visually highlighted domain identifiers in a graphical user interface with the likely-use clinical data presented when the likely domain identifier is selected.
Claims Coverage
The independent claims are clm-00001, clm-00011, and clm-00017. Across these independent claims, the coverage centers on domain-based, machine-learning identification of likely vs not likely clinical-data domains for a point of care patient, using a healthcare-worker audit table and a domain-organized virtual patient object, followed by generating an optimized virtual patient object with markings and presenting visually highlighted domain identifiers in a client GUI.
Server EHR agent retrieves audit table and domain-organized VPO for point-of-care patient
A server electronic health records agent application receives a healthcare worker identifier and a patient identifier, retrieves an audit table for the healthcare worker, and retrieves a virtual patient object for the patient with clinical data aggregated from a plurality of sources and organized within the virtual patient object according to domains.
Machine learning model identifies likely and not likely clinical-data domains
The server EHR agent provides the audit table and the virtual patient object as input to a computer-implemented machine learning model having learned weights assigned thereto, wherein the model identifies a first domain and a second domain, the first domain comprising a first subset of clinical data determined to be likely to be used at the point of care, and the second domain comprising a second subset of clinical data determined to be not likely to be used at the point of care.
Optimized VPO marks domains and omits not-likely clinical data
The server EHR agent generates an optimized virtual patient object based on the virtual patient object and the first and second domains identified by the machine learning model, wherein the optimized virtual patient object includes domain identifiers, includes the first subset of clinical data via a marking assigned to the identifier for the first domain, and does not include the second subset of clinical data via a marking assigned to the identifier for the second domain.
Client GUI highlights likely and not included domains and presents selected clinical data
The server EHR agent transmits the optimized virtual patient object to a client EHR agent, where identifiers for the domains are presented within a graphical user interface, the identifier for the first domain is visually highlighted to indicate likely use, the identifier for the second domain is visually highlighted to indicate not included, and the subset of clinical data of the first domain is presented on the display when the identifier for the first domain is selected.
Healthcare-worker-specific ML model or weights customized by identifier
A healthcare worker identifier is used to select one machine learning model among a plurality of machine learning models assigned to different healthcare workers, or to retrieve a healthcare worker-specific set of learned weights and replace the model’s learned weights with that retrieved set.
Computer-readable medium storing server EHR agent performs domain-likelihood VPO optimization and GUI highlighting
A computer-readable storage medium stores a server electronic health records agent application configured to access health records from sources inaccessible to a server electronic health records application, and when executed performs acts of receiving healthcare worker and patient identifiers, retrieving an audit table and a domain-organized virtual patient object, using a machine learning model to identify a first likely domain and second not likely domain, generating an optimized virtual patient object with markings for the first and second domains, transmitting the optimized virtual patient object to a client EHR agent, and presenting visually highlighted domain identifiers and selected clinical data within a graphical user interface.
The independent claims collectively cover a server-side EHR agent that uses an audit table of healthcare-worker interactions together with a domain-organized virtual patient object and a machine learning model with learned weights to identify likely vs not likely patient clinical-data domains at a point of care. The server generates an optimized virtual patient object that includes and marks the likely domain data while omitting not-likely domain data, transmits it to a client EHR agent, and enables GUI visual highlighting and selection-based presentation of the likely-use clinical data.
Stated Advantages
Reducing irrelevant data display by focusing the optimized virtual patient object on domains likely to be used at point of care.
Saving network resources by transmitting an optimized virtual patient object rather than including not-likely domain clinical data.
Documented Applications
A distributed EHR agent system where a server EHR agent optimizes a virtual patient object for a point-of-care patient using a healthcare-worker audit table, and a client EHR agent presents likely vs not-likely clinical-data domains in a graphical user interface.
Interested in licensing this patent?