Generative ai clinical validation assistant

Abstract

Methods and systems for using generative artificial intelligence to generate medical data summaries for a patient are disclosed herein. A method includes: obtaining medical information for a patient; obtaining input data associated with the patient from a user, the input data including indications of one or more medical categories; identifying, using the input data, one or more trained generative Al models for analyzing the medical information for the patient; generating, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; and presenting, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

Description

Patent Application 31134 / 70821 / PC GENERATIVE Al CLINICAL VALIDATION ASSISTANTCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and the benefit of the filing date of provisional U.S. Patent Application No. 63 / 780,633, entitled “GENERATIVE Al CLINICAL VALIDATION ASSISTANT,” filed on March 31, 2025, and Greek Patent Application No. 20250100252, entitled “GENERATIVE Al CLINICAL VALIDATION ASSISTANT,” filed on March 28, 2025. The entire contents of the application(s) are hereby expressly incorporated herein by reference.FIELD OF THE DISCLOSURE

[0002] The present disclosure relates generally to techniques for using generative artificial intelligence (Al) to generate medical data summaries and, more particularly, to techniques for generating a summary based on medical data for a patient using one or more generative Al models trained on historical medical data related to particular medical categories selected by a clinician associated with the patient.BACKGROUND

[0003] The background description provided herein is for the purpose of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent it is described in this background section, as well as aspects of the description that may not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present disclosure.

[0004] In the current healthcare environment, nurses and other clinicians face significant challenges accessing the documents and resources necessary to provide high-quality care to patients. The challenges associated with conventional evidence-based care techniques are exacerbated by the dispersion of such resources across multiple locations and applications, making it challenging for clinicians to find relevant information efficiently. Furthermore, national staffing shortages and high patient acuity place additional pressure on clinicians to deliver high-quality care efficiently. Existing clinical systems often do not align with nursing / clinician workflows, leading to an increased administrative burden and less timePatent Application 31134 / 70821 / PC spent on direct patient care, as well as decreased computational efficiency of the clinical system.

[0005] There is a need for techniques for improving clinician access to necessary resources, thereby enabling clinicians to focus more on patient care and less on administrative tasks.SUMMARY OF THE INVENTION

[0006] The present techniques include methods and systems for generating medical data summaries for a patient using generative artificial intelligence (Al) models. The computer-implemented method may comprise: (1) obtaining, via one or more processors, medical information for a patient; (2) obtaining, via the one or more processors, input data associated with the patient from a user, the input data including indications of one or more medical categories; (3) identifying, via the one or more processors and using the input data, one or more trained generative Al models for analyzing the medical information for the patient, wherein; (i) each trained generative Al model of the one or more trained generative Al models is associated with a respective medical category of the one or more medical categories, and (ii) the each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources; (4) generating, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; and (5) presenting, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

[0007] In an example, a computing system for generating medical data summaries for a patient using generative artificial intelligence (Al) models comprises: one or more processors; and a non-transitory computer readable medium including computer executable instructions that, when executed by the one or more processors, cause the computing system to: (1) obtain medical information for a patient; (2) obtain input data associated with the patient from a user, the input data including indications of one or more medical categories; (3) identify, using the input data, one or more trained generative Al models for analyzing thePatent Application 31134 / 70821 / PC medical information for the patient, wherein: (i) each trained generative Al model of the one or more trained generative Al models is associated with a respective medical category of the one or more medical categories, and (ii) the each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources; (4) generate, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; and (5) present, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

[0008] In another example, a non-transitory computer-readable medium having stored thereon instructions that, when executed by one or more processors of a computing system, cause the computing system to: (1) obtain medical information for a patient; (2) obtain input data associated with the patient from a user, the input data including indications of one or more medical categories; (3) identify, using the input data, one or more trained generative Al models for analyzing the medical information for the patient, wherein: (i) each trained generative Al model of the one or more trained generative Al models is associated with a respective medical category of the one or more medical categories, and (ii) the each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources; (4) generate, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; and (5) present, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

[0009] In yet another example, a computer-implemented method for generating medical data summaries for a patient using generative artificial intelligence (Al) models, may comprise: (1) obtaining, via one or more processors, medical information for a patient; (2) obtaining, via the one or more processors, input data associated with the patient from a nurse practitioner, the input data including indications of one or more nursing categories; (3) identifying, via the one or more processors and using the input data, one or more trainedPatent Application 31134 / 70821 / PC generative Al models for analyzing the medical information for the patient, wherein: (i) each trained generative Al model of the one or more trained generative Al models is associated with a respective nursing category of the one or more nursing categories, and (ii) the each trained generative Al model is trained using a corresponding subset of training medical data from one or more nursing data sources of a plurality of nursing data sources; (4) generating, via the one or more trained generative Al models, one or more nursing summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of nursing data sources; and (5) presenting, via a graphical user interface on a computing device, at least one nursing summary of the one or more nursing summaries associated with the medical information for the patient for review by the nurse practitioner.BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The patent or application file contains at least one drawing executed in color.Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.

[0011] The figures described below depict various aspects of the system and methods disclosed herein. It should be understood that each figure depicts an example of aspects of the present systems and methods.

[0012] FIG. 1 illustrates an example block diagram of a computing system configured to implement methods for generating medical data summaries for a patient using generative artificial intelligence (Al) models, as described herein.

[0013] FIG. 2 illustrates an example block flow diagram depicting an exemplary method for generating medical data summaries for a patient using generative artificial intelligence (Al) models, implemented in the system of FIG. 1.

[0014] FIG. 3A illustrates an example graphical user interface (GUI) for presenting generated summaries, as described herein.

[0015] FIG. 3B illustrates an example graphical user interface (GUI) for presenting generated summaries, as described herein.Patent Application 31134 / 70821 / PC

[0016] FIG. 4 illustrates an example block flow diagram for developing a finetuned generative artificial intelligence (Al) model, according to some aspects.

[0017] FIG. 5 illustrates a combined block and logic diagram for training an example generative Al model, according to some aspects.

[0018] FIG. 6 illustrates a computer-implemented method for generating medical data summaries for a patient using generative artificial intelligence (Al) models, according to some aspects.DETAILED DESCRIPTION

[0019] Although the following text sets forth a detailed description of numerous different embodiments, it should be understood that the legal scope of the invention is defined by the words of the claims set forth at the end of this patent. The detailed description is to be construed as exemplary only and does not describe every possible embodiment, as describing every possible embodiment would be impractical, if not impossible. One could implement numerous alternate embodiments, using either current technology or technology developed after the filing date of this patent, which would still fall within the scope of the claims.

[0020] In the healthcare sector, particularly in nursing, there exists significant challenges in providing evidence based-care to patients. The disclosed invention addresses these challenges by leveraging generative artificial intelligence (GAI or generative Al) technology to develop a GAI clinical validation assistant. At a high level, the disclosed techniques centralize access to various distributed resources, ensuring that necessary documents and information are readily available to nurses at the point of care. More specifically, the disclosed techniques use generative Al models to pull relevant data (e.g., clinical guidelines, research articles, educational materials, enterprise content management documents, materials characterization data, historical electronic health records, healthcare analytics data, etc.) from these distributed resources, based on patient-specific information through the integration of electronic health records (EHRs). The disclosed invention significantly improves the efficiency of care delivery by reducing the time and resources nurses spend searching for evidence-based resources.Patent Application 31134 / 70821 / PC

[0021] Furthermore, the disclosed invention enhances healthcare delivery by significantly improving the computational efficiency and data management within clinical systems, particularly benefiting nursing workflows. Specifically, the disclosed invention reduces processing time by efficiently managing data queries and retrieval processes through the use of trained generative Al models. Additionally, the disclosed generative Al models, finetuned to a specific medical category, provide for improved analyzing of medical information (e.g., electronic health record data of a patient) and generation of summaries of relevant documents / information, thereby decreasing the computational resources required for the complexity of data processing necessitated by evidence-based care. The disclosed invention also enhances data storage and exchange mechanisms within clinical systems through the use of generative Al models and / or integrated application programming interfaces (APIs).

[0022] Furthermore, by centralizing access to distributed healthcare resources, the disclosed invention may reduce the redundancy of data storage across multiple applications and platforms. Moreover, the disclosed invention not only optimizes storage requirements, but also ensures that updates to any piece of information (e.g.. a clinician has access to out-of-date information) are universally reflected, maintaining data integrity and consistency.The disclosed invention improves computational efficiency by automating the identification and retrieval of relevant information for administering evidence-based care. The disclosed invention eliminates or minimizes the need for manual data entry and / or search efforts, thereby reducing the likelihood of human error and increasing the speed of information retrieval. Similarly, the instant techniques may block, delete, or otherwise prevent various types or sources of data from being used in analysis, data gathering, and / or training of a generative model, providing improvements to the overall performance and accuracy of the model and / or outcomes.

[0023] Additionally, through integration with existing EHR systems (e.g., a generative Al clinical validation assistant within nurses' existing digital environment), the disclosed invention ensures seamless data exchange and workflow alignment, thereby minimizing disruptions to clinical workflows, expediting user adoption, and facilitating real-time updates to patient records. For example, the disclosed invention may include a generative Al clinical validation assistant integrated with an electronic health records (EHR) systems, suchPatent Application 31134 / 70821 / PC as Epic, and configured to intelligently gather and present necessary resources based on individual patient data.

[0024] In summary, the disclosed invention presents a novel solution to at least the longstanding issues of information dispersion and accessibility in healthcare. Further, the disclosed techniques significantly improve the computational efficiency, data management, and overall workflow within clinical settings, thereby enhancing and improving overall quality of patient care and the quality of the systems performing the patient care.

[0025] Turning first to FIG. 1, a system 100 configured to implement methods for generating medical data summaries includes a server computing device 102, a clinician computing device 104, a network 106, one or more medical data sources 110, and one or more application programming interfaces (APIs) 112. The server computing device 102, or server, includes one or more processors 120, one or more communication interfaces 130, and one or more memories 140. The clinician computing device 104 includes one or more processors 160, one or more memories 180, one or more communication interfaces 190, and one or more user interfaces 192. The one or more networks 106 may comprise any suitable network or networks, including a local area network (LAN), wide area network (WAN), the internet, or a combination thereof. For example, network(s) 106 may include a wireless cellular service (e.g., 4G, 5G, 6G, etc.). Generally, the network 106 enables bidirectional communication between the server computing device 102, the clinician computing device 104, and the one or more medical data sources 110. Additionally or alternatively, the network 106 may comprise one or more routers, wireless switches, or other such wireless connection points communicating to the components of the computing environment 100 via wireless communications based upon any one or more of various wireless standards, including by non-limiting example, IEEE 802.11a / b / c / g (Wi-Fi), Bluetooth, and / or the like. For ease of reading herein (and not for limitation purposes), the one or more networks 106 may be referred to using the singular tense.

[0026] The server computing device 102 may be an individual server, a group (e.g.. cluster) of multiple servers, or another suitable type of computing device or system (e.g., a collection of computing resources). For example, the server 102 may be a personal computer, a server, a mobile computing device, a smart phone, a tablet, a laptop, etc.Patent Application 31134 / 70821 / PC Additionally, the server 102 may be the property of a customer, a company, an organization, etc.

[0027] As mentioned above, the server 102 may include one or more processors 120 and one or more memories 140. The processors 120 may include any suitable number of processors and / or processor types, such as CPUs and one or more graphics processing units (GPUs). Generally, the processors 120 are configured to execute software instructions stored in a memory (e.g., the memory 140). For example, one or more CPUs of the one or more processors 120 of the server 102 may be configured to execute software instructions in the memory 140 for implementing at least the method 600 of FIG. 6 as well as other methods and / or techniques as described herein. As another example, one or more GPUs of the one or more processors 120 of the server 102 may be configured to train one or more machine learning models (e.g., the one or more generative Al models 146). The memory 140 may include one or more persistent memories (e.g., a hard drive / solid state memory) and may store one or more sets of computer executable instructions / modules, including a data processing module 141, an MU operation module 142, a prompt generation module 144, one or more generative Al models 146, and a validation module 148, as described in more detail below.

[0028] The one or more communication interfaces 130 includes at least one wireless communication interface (such as a network interface controller), which includes hardware, firmware, and / or software that is generally configured to communicate with other devices and / or over network 106 using one or more wireless communication protocols. For example, the communication interface 130 may be configured to transmit and receive data using a Bluetooth protocol, a Wi-Fi® (IEEE 802.11 standard) protocol, a near-field communication (NFC) protocol, a cellular protocol (such as global system for mobile communications (GSM), code-division multiple access (CDMA), long-term evolution (LTE), worldwide interoperability for microwave access (WiMAX), etc.), a peer-to-peer wireless protocol, a short-range wireless protocol, and / or other suitable wireless communication protocols.

[0029] Additionally, although not shown in FIG. 1 , it will be understood that, in some implementations, communication interface 130 may include one or more wired communication interfaces which may be utilized by the server 102 to communicativelyPatent Application 31134 / 70821 / PC connect to the medical data source(s) 110, the clinician computing device 104, a network of computing devices, and / or other devices of the environment 100 via one or more wired communication protocols and / or wired data protocols. In some embodiments, the communication interface 130 may include one or more suitable network interface controllers (NICs), such as wired / wireless controllers (e.g., Ethernet controllers), and facilitate bidirectional / multiplexed networking over the network 106 between the server 102 and other components of the environment 100 (e.g., the clinician computing device 104, the one or more medical data sources 110. another computing device, a remote computing device, etc.). In some embodiments, the communication interface 130 may include one or more transceivers to support various different wireless communication protocols; however, for ease of reading (and not limitation purposes) herein, the communication interface 130 may be referred to herein using the singular tense.

[0030] Returning to the memory 140, the data processing module 141, the ML operation module 142, the prompt generation module 144, and the validation module 148 may, generally, include machine-readable instructions corresponding to, for example, the operations represented by the flowcharts of this disclosure (e.g., the flowchart 600 of FIG.6).

[0031] For example, the memory 140 may include instructions for analyzing / evaluating medical information / data for a patient using the one or more generative Al models 146 (e.g., one or more base or foundation language models or one or more finetuned models, as described below with respect to FIG. 4 and FIG. 5) to generate one or more summaries associated with the medical information for the patient. In some embodiments, the memory 140 may obtain medical information for a patient from an EHR database (not depicted) (e.g., the medical information for a patient may be included in a corresponding electronic health record (EHR) for the patient) and / or other third party or off-premises database. For example, the medical information for a patient may include information related to medical history, medications, known allergies, vital signs, radiology or other medical images, etc. In some embodiments, each of the one or more generative Al models 146 is trained, and / or finetuned, on medical data associated with a particular medical category (e.g., a subset of historical medical data relevant to the patient of interest and the respective medicalPatent Application 31134 / 70821 / PC categories). For example, the medical data may include clinical guidelines, research articles, educational materials, etc.

[0032] Further, the memory 140 may include instructions for evaluating medical information for a patient with respect to historical medical data, or training medical data, using the one or more generative Al models 146, as described further below with respect to the ML operation module 142. Continuing with the above example, the memory 140 may input a prompt to generative Al model 146 that causes generative Al model 146 to (i) retrieve historical medical data related to the corresponding medical category and (ii) evaluate the medical information for the patient with respect to the retrieved medical data (sometimes referred to as “retrieval augmented generation”). In some embodiments, the one or more generative Al models 146 are selected based on clinician or other healthcare professional input. For example, the memory 140 may evaluate input data from a nurse (e.g., input data provided via clinician computing device 104) to identify the one or more generative Al models 146. As mentioned above, each of the one or more generative Al models 146 may be associated with a respective medical category. Furthermore, each generative Al model may have a particular architecture, or support certain functionalities (e.g., image processing, text extraction, etc.), that are suited for a particular medical category, in addition to the models being trained on a subset of medical data associated with a particular category. In this way, a clinician may configure and / or personalize the deployment of the generative Al models 146 via clinician computing device 104.

[0033] The data processing module 141 may include instructions for retrieving, and / or receiving, and processing medical data / information for a patient (e.g., medical history, medications, known allergies, vital signs, radiology or other medical images, etc.). For example, the data processing module 141 may receive an indication of a patient, such as a patient identifier or patient name, and may retrieve medical information for the patient from a patient information data source, such as an EHR database (not depicted), or another type of patient information database, communicatively connected to the server 102 (e.g., via network 106). As another example, the data processing module 141 may obtain or receive medical information for a patient from the clinician computing device 104 (e.g., the medical information for the patient is stored locally on the clinician computing device 104).Patent Application 31134 / 70821 / PC

[0034] Tn some embodiments, the data processing module 141 obtains, or receives, input data from a clinician via the clinician computing device 104. For example, the data processing module 141 may cause the clinician computing device 104 to display a graphical user interface presenting a plurality of selectable medical categories (e.g., situation, background, surgical history, vital signs, nursing physical assessment, etc.), as described further below with respect to FIG. 2-6. Moreover, the data processing module 141 may obtain input data including indications of one or more medical categories (e.g., medical categories selected by a nurse) via the clinician computing device 104. Additionally or alternatively, the data processing module 141 may evaluate the medical information for a patient to determine one or more medical categories associated with the patient and / or the medical information. For example, the patient may be associated with particular groups of patients (e.g., patients of a particular demographic, patients with a particular condition, patients with common medical history attributes, etc.) and the data processing module 141 may identify particular medical categories associated with each group of patients based on the medical information. Additionally, the data processing module 141 may provide the determined / identified medical categories associated with a patient to the ML operation module 142.

[0035] Generally, the ML operation module 142 may include instructions for implementing one or more ML models, such as the one or more generative Al models 146. In some embodiments, the ML operation module 142 identifies, based on the one or more medical categories from the data processing module 141, one or more trained generative Al models (e.g., from among the generative Al models 146) for analyzing medical information from the data processing module 141. Further, the ML operation module 142 may also include instructions for training or finetuning the one or more generative Al models 146. For example, the ML operation module 142 may finetune or train a generative Al model of the one or more models 146 on training medical data corresponding to each respective medical category from the data processing module 141. In some embodiments, the ML operation module 142 may obtain the training medical data from the one or more medical data sources 110. For example, the medical data sources 110 may include healthcare platforms / systems (e.g., Lippincott, OnBase, Micrometrics, Cemer, etc.) that store clinical guidelines, researchPatent Application 31134 / 70821 / PC articles, educational materials, enterprise content management documents, materials characterization data, historical electronic health records, healthcare analytics data, etc.

[0036] In some embodiments, the ML operation module 142 and / or the data processing module 141 may process the data from the one or more medical data sources 110 to generate structured training data or input data. For example, the ML operation module 142 and / or the module 141 may sort and / or organize the data by risk level and volume (e.g., clinical guidelines or procedures may be sorted based on complexity). Continuing with this example, the module 142 and / or the module 141 may organize the data (e.g., procedures and guidelines, research articles, educational materials, etc.) from the medical data sources 110 by high risk, low volume, from simple to complex, etc.

[0037] At a high level, the ML operation module 142 may generate a respective summary for each determined and / or indicated medical category by inputting a corresponding subset of training medical data, from the medical data sources 110, and the medical information for a patient to each respective generative Al model identified by the ML operation module 142. In some embodiments, an example trained generative Al model (e.g., as described below with respect to FIGs. 4-6) may be a foundational model trained to have a strong understanding of natural language and excel at natural language processing tasks (e.g., the model may be pre-trained to understand natural language inputs). Advantageously, the ML operation module 142 may summarize a medical information regarding a patient using multiple generative Al models 146, whereby each generative Al model is instructed (e.g., via an input prompt from prompt generation module 144) to evaluate the medical information with respect to medical data (e.g., from the medical data sources 110) associated with a corresponding medical category. Moreover, the described techniques eliminate the need for additional training or finetuning of the models to improve the processing capabilities of a generative Al model with respect to a particular type of medical data or medical category. The architecture, training, and / or implementation of exemplary machine learning models and / or exemplary generative Al models (e.g., the generative Al models 146) are described further below with respect to FIG. 4 and FIG. 5.

[0038] The prompt generation module 144 may include instructions for generating prompts and / or instruction sets for the one or more generative Al models 146. In somePatent Application 31134 / 70821 / PC embodiments, the prompt generation module 144 generates a prompt by interpolating at least a portion of the medical information for a patient and / or training medical data (e.g., data snippets from the medical data sources 110) into a template prompt. For example, the prompt generation module 144 may include, or store, a plurality of template or base prompts respectively associated with one or more medical categories and / or one or more generative Al models. Further, the prompt generation module 144 may generate respective prompts for each generative Al model 146 identified by the ML operation module 142, based on a template prompt associated with the corresponding medical category. In some embodiments, the prompt generation module 144 may store template prompts that include additional context for guiding the generative Al models in the generation of content. For example, the additional context may cause a generative Al model to generate intermediary conclusion(s) while generating an output, thereby helping the model generate relevant information for the instructed task by implementing a logical flow. As another example, the additional context may cause a generative Al model to evaluate accuracy of the intermediary conclusions while generating the output, thereby forcing the model to evaluate the accuracy of conclusions, assumptions, inferences, etc. By performing such intermediary conclusion generation, the generative Al model may reduce errors, hallucinations, and / or other such disruptive influences on the model output.

[0039] Generally, the prompt generation module 144 may generate a prompt for a generative Al model of the one or more generative Al models 146 that includes a set of instructions and a set of input data, whereby the set of instructions guides the model in generating medical data summaries for a patient based on the set of input data. In some embodiments, the prompt generation module 144 may train generative Al models 146 (e.g.. a base or foundational language model, as described with respect to FIG. 4 and FIG. 5) using prompts that include sets of instructions (e.g., commands, questions, etc.) and examples outputs for the generative Al models 146 (e.g., an example output associated with a medical category from the data processing module 141), thereby eliminating the need for task specific training or fine-tuning (sometimes referred to as “one-shot training” or “one-shot learning”). Additionally, the prompt generation module 144 may, ostensibly, train the generative Al models 146 using retrieval augmented generation (e.g., the generated promptsPatent Application 31134 / 70821 / PC may cause / instruct a model to retrieve data from an external source and evaluate the input data with respect to the retrieved data).

[0040] The validation module 148 may include instructions for evaluating the accuracy of medical summaries generated by the generative Al models 146, via the ML operation module 142. For example, the validation module 148 may validate the accuracy of a medical summary against data snippets, from medical data associated with a corresponding medical category, that are input into the respective generative Al model via a corresponding prompt. In some embodiments, the validation module may validate the accuracy of a medical summary against the respective medical data using one or more additional generative Al models included in generative Al models 146.

[0041] As mentioned above, the clinician computing device 104 may include one or more processors 160, one or more memories 180, one or more communication interfaces 190, and one or more user interfaces 192. The clinician computing device 104 may be an individual computing device, a group (e.g., cluster) of multiple computing devices, or another suitable type of computing device or system (e.g., a collection of computing resources). For example, the clinician computing device 104 may be a personal computer, a laptop, a mobile computing device, a smart phone, a tablet, a server, etc. Additionally, the clinician computing device 104 may be the property of a customer, a company, an organization, etc. The processors 160 may include any suitable number of processors and / or processor types, such as CPUs and one or more graphics processing units (GPUs). Generally, the processors 160 are configured to execute software instructions stored in a memory (e.g.. the memory 180). For example, the processors 160 may be configured to execute software instructions in the memory 180 for implementing the method 600 of FIG. 6 and / or other methods or techniques described herein. The memory 180 may include one or more persistent memories (e.g., a hard drive / solid state memory) and may store one or more sets of computer executable instructions / modules. In some embodiments, an example user of the clinician computing device 104 may be a clinician or nurse. For example, a “clinician” may refer to any individual interacting with a patient in a clinical context (e.g., physicians, nurses, physicians assistants, certified nursing assistants, etc.).Patent Application 31134 / 70821 / PC

[0042] The one or more communication interfaces 190 (similar to communication interface 130) includes at least one wireless communication interface (such as a network interface controller), which includes hardware, firmware, and / or software that is generally configured to communicate with other devices and / or over network 106 using one or more wireless communication protocols. Additionally, although not shown in FIG. 1, it will be understood that, in some implementations, communication interface 190 may include one or more wired communication interfaces which may be utilized by the clinician computing device 104 to communicatively connect to the medical data source(s) 110, the server computing device 102, a network of computing devices, and / or other devices via one or more wired communication protocols and / or wired data protocols. In some embodiments, the communication interface 190 may include one or more suitable NICs. such as wired / wireless controllers (e.g., Ethernet controllers), and facilitate bidirectional / multiplexed networking over the network 106 between the clinician computing device 104 and other components of the environment 100 (e.g., the server computing device 102, the one or more medical data sources 110, another computing device, a remote computing device, etc.). In some embodiments, the communication interface 190 may include one or more transceivers to support various different wireless communication protocols; however, for ease of reading (and not limitation purposes) herein, the communication interface 190 may be referred to herein using the singular tense.

[0043] The one or more user interfaces 192 of the clinician computing device 104 may receive clinician input and communicate output data to the clinician. Further, the user interface 192 may be configured to present interactive graphical user interfaces (GUIs) for review by a clinician and receive feedback data from the clinician (e.g.. via the input / output devices 194). For example, the user interface 192 may present interactive graphical user interfaces and / or graphical elements for review by a clinician, such as the graphical user interfaces depicted in FIG. 3A and FIG. 3B, as discussed in greater detail below. As another example, a clinician may provide input data, such as indications of one or more medical categories, using the user interface 192. Continuing with the above example, the input data from the clinician may be sent to the server computing device 102 and the data processing module 141 may process the input data.Patent Application 31134 / 70821 / PC

[0044] The user interfaces 192 may include one or more input / output (I / O) device(s) 194. For instance, the I / O devices 194 may include one or more suitable types of user input devices, such as keyboards, touch screen displays, microphones, mice, touchpads, and / or any suitable types of remote and / or local user input devices. Similarly, the I / O devices 194 may include one or suitable types of output devices, such as touch screen displays, speakers, and the like. In some embodiments, the I / O devices 194 may include one or more display(s) / screen(s) 196 for presenting or displaying information to a user. The one or more display s / screens 196 may use any suitable display technology (e.g., LED, OLED, LCD, etc.). For example, the I / O devices 194 and / or the displays / screens 196 may be configured to present graphical representations of information and / or data associated with the embodiments described herein, such as information related to a medical data or data snippets (e.g., citations) from the medical data sources 110, information related to training and / or finetuning the generative Al models 146, etc. Generally, the user interface(s) 192 may present curated nurse patient summary and direct links (e.g., hyperlinks and / or web links) to contextual, native, and evidence-based practice resources from multiple disperse locations to the fingertips of nurses at the point of care. In some embodiments, the I / O device 194 and the display / screen 196 are integrated as a touchscreen display. In some embodiments, the user interfaces 192, the I / O devices 194, and / or the displays / screens 196 are not integral to the clinician computing device 104 and receive instructions from the clinician computing device 104 via wired and / or wireless transmissions over communication interface 190, for example. In some embodiments, the user interfaces 192, the I / O devices 194, and / or the displays / screens 196 may include one or more local interfaces, and / or may include one or more remote interfaces that are communicatively connected to the clinician computing device 104 via the network 106 (e.g.. that are provided by an application, web browser, or other software executing on a device of a user). For ease of reading (and not limitation) purposes, the user interfaces 192, the I / O devices 194, and / or the displays / screens 196 may be referred to herein using the singular tense.

[0045] The application programming interfaces (APIs) 112 may facilitate interaction between components and / or devices of the computing system 100. Generally, the APIs 112 may be configured to receive data, and / or information, from a component of the computing system 100 and to provide such data to a different component of the computing system 100.Patent Application 31134 / 70821 / PC For example, the APIs 112 may be configured to exchange information between the generative Al models 146 and the medical data sources 110. As another example, the APIs 112 may be configured to provide vectorized input data to the generative Al models 146, the ML operation module 142, the data processing module 141, etc. In some embodiments, the one or more APIs 112 may include a computer vision API that includes visual processing model / application, for instance, a convolutional neural network (CNN), an image-to-graph transformer, a graph neural network (GNN), a multilayer perceptron, etc. Generally, an exemplary computer vision API 112 may generate graph representations (e.g., a text file) of visual data, and may provide the graph representations to the one or more generative Al models 146, thereby enabling the one or more generative Al models 146 to interpret visual data. In some embodiments, the APIs 112 may be configured such that the generative Al models 146 may access external data (e.g., from the medical data sources 110) via the APIs 112, while generating an output.

[0046] The one or more medical data sources 110 may store medical data associated with various medical categories (e.g., situation, background, surgical history, vital signs, nursing physical assessment, etc.). For example, the medical data sources 110 may store clinical guidelines, research articles, educational materials, enterprise content management documents, materials characterization data, historical electronic health records, healthcare analytics data, etc. Continuing with this example, the medical data sources 110 may include one or more healthcare platforms / systems (e.g., Lippincott, OnBase, Micrometrics, Cemer, etc.) that are, for instance, associated with a particular clinician and / or type of practice.

[0047] FIG. 2 illustrates an example block-flow diagram for a computer-implemented method 200 for generating medical data summaries for a patient using generative artificial intelligence (Al) models, implemented by a computing system (e.g., system 100). It will be understood that the method 200 may be performed by other systems and / or components thereof (e.g., other systems configured to perform similar operations, and / or components thereof).

[0048] At block 202, input data from a clinician may be obtained and provided to the ML operation module 142 and the prompt generation module 144 (e.g., via the data processing module 141). In some embodiments, the server computing device 102 may obtain the inputPatent Application 31134 / 70821 / PC data from the clinician computing device 104. Generally, the data processing module 141 may process the input data, and / or the medical information for a patient, to identify one or more medical categories, as described above, with respect to FIG. 1. Depending on the implementation, the medical categories may be determined based on elements in the medical information and / or input data (e.g., as determined and / or used while training a model at the data processing module 141), indicated by a user, generated based on a template, etc.

[0049] At block 204, patient data, such as medical information associated with a patient, is provided to the ML operation module 142 and the prompt generation module 144 (e.g., via the data processing module 141). Generally, the data processing module 141 may obtain medical information for a patient (e.g., patient data 204) from an EHR database, a similar patient information data store, the clinician computing device 104, or another component of the computing environment 100, as described above with respect to FIG. 1.

[0050] Based on the one or more identified medical categories (block 202) and the patient data 204, the ML operation module 142 may identify one or more generative Al models, of the generative Al models 146, for processing the patient data. Depending on the implementation, the ML operation module 142 may identify the one or more generative Al models based on one or more characteristics of the generative Al models, one or more categories of analysis to be performed on the corresponding identified medical categories, one or more user preferences, etc.

[0051] Based on the one or more identified medical categories (block 202) and the patient data 204, the prompt generation module 144 may generate one or more prompts (block 206) for the generative Al model(s) 146. Generally, the one or more prompts 206 may include at least (i) a portion of the patient data 204, or a portion of the medical information, and (ii) additional context for guiding a generative Al model in generating an output or evaluating input data.

[0052] In some embodiments, the generative Al models 146 are communicatively connected to the one or more medical data sources 110 via a bus interface 208 (e.g., and called or accessed via APIs 112). Further, the additional context included in prompts 206 may cause the generative Al models 146 to retrieve one or more data snippets 210 from the medical data sources 110. In some embodiments, the generative Al models 146 mayPatent Application 31134 / 70821 / PC generate one or more queries for the medical data sources 110 to retrieve medical data related to a particular medical category. Additionally, the generative Al models 146 may issue the one or more queries to and / or against the medical data sources 110. In some embodiments, the data snippets 210 may be, or include, citations to medical data associated with an identified medical category (e.g., citations to a clinical guideline related to the medical category). In some implementations, the generative Al models 146 may therefore improve the overall accuracy and functionality of the models by providing and / or crosschecking data against particular data and citing accordingly. As such, the overall hallucination rate of the model may be decreased and / or mitigated. In further implementations, a user and / or another module may confirm that the citations are accurate and / or to existing sources, and may modify or reinitiate the corresponding generative Al model to generate a new summary if the citation(s) are inaccurate.

[0053] At a high level, the generative Al models 146 may evaluate the patient data 204 with respect to the obtained data snippets 210. Further, the generative Al models 146 may augment the generation of content (e.g., a summary related to a particular medical category) by retrieving data related to the medical category (sometimes referred to as “retrieval augmented generation”, as mentioned above), thereby training or finetuning the generative Al models 146 for the medical category.

[0054] In response to the one or more prompts 206, the generative Al models 146 may generate respective medical summaries 212 (e.g., a summary for each input prompt and / or medical category). Generally, the generative Al models 146 may be instructed (e.g., via the prompts 206) to summarize portions of the patient data 204 with respect to corresponding data snippets 210 (e.g., the corresponding data snippets 210 are provided as context to a model when tasked with generating a summary). Additionally, the validation module 148 may evaluate the accuracy of the medical summaries 212 against the corresponding data snippets 210, to identify hallucinations within the medical summaries 212.

[0055] In response to the validating, the one or more medical summaries 212 may be transmitted to the clinician computing device 104 for review by a clinician. As described below with respect to FIG. 3 A and FIG. 3B, the clinician computing device 104 may present, via one or more graphical user interfaces (e.g., via user interfaces 192), the one orPatent Application 31134 / 70821 / PC more medical summaries 212, the one or more data snippets 210, the patient data 204, and / or the input data 202.

[0056] FIG. 3A illustrates an example graphical user interface (GUI) 300a for presenting medical summaries, generated by a computing system (e.g., system 100). for review by a clinician. It will be understood that the GUI 300a may be generated, presented, and / or displayed by other systems and / or components thereof (e.g., other systems configured to perform similar operations, and / or components thereof). Generally, the example GUI 300a may be presented via the user interface 192 of the clinician computing device 104.

[0057] As depicted in FIG. 3A, the GUI 300a may include a nurse patient summary section 302 and a nursing references section 304. The nurse patient summary section 302 may include an overview section 306 and a sources section 308. The nursing references section 304 may include a search interface 312, a search results section 314, and a contextual references section 316. Generally, the medical summaries generated via the generative Al models 146 (e.g., the medical summaries 212), as described above with respect to FIG. 1 and FIG. 2, may be presented within the overview 306. Furthermore, in the illustrated example, the overview section 306 includes medical summaries for a plurality of medical categories (e.g., situation, background, surgical history, vital signs, labs, diagnostic studies, nursing physical assessment, last bowel movement, active intravenous lines, active gastrointestinal tubes and drains, mental health and / or psychosocial status, respiratory ventilation data, last patient safety check, current diet status, etc.). In some embodiments, the medical summaries presented within the overview section 306 may correspond to medical categories selected by a clinician of the patient (e.g., a nurse practitioner). Additionally, the data snippets 210 and / or citations to data from the medical data sources 110 may be presented within the sources section 308. In some embodiments, the nursing references section 304 may include hyperlinks and / or web portals associated with the one or more medical data sources 110. For example, the nursing references section 304 may include hyperlinks to resources from various healthcare platforms / systems (e.g., Lippincott, OnBase, Micrometrics, Cemer, etc.).

[0058] FIG. 3B illustrates an example graphical user interface (GUI) 300b for presenting medical summaries, generated by a computing system (e.g., system 100), review by aPatent Application 31134 / 70821 / PC clinician. Tt will be understood that the GUT 300b may be generated, presented, and / or displayed by other systems and / or components thereof (e.g., other systems configured to perform similar operations, and / or components thereof). Generally, the example GUI 300b may be presented via the user interface 192 of the clinician computing device 104. As depicted in FIG. 3B, the GUI 300b may include a medical category 320 (e.g., treatment in the illustrated example) and a medical summary 322 (e.g., generated by the generative Al models 146, as described above with respect to FIG. 1 and FIG. 2). The GUI 300b may also include one or more sources (e.g., source 330, source 332, and source 334) for each respective medical summary (e.g., medical summary 322). In some implementations, the GUI 300b may be or include additional or alternate categories for other healthcare providers. For example, the exemplary embodiment of Fig. 3A may include medical categories of interest to a nurse or nurse practitioner (e.g., within overview section 306), while the exemplary embodiment of Fig. 3B may indicate medical categories (e.g., medical category 320) of interest to a physician.

[0059] FIG. 4 depicts a block-flow diagram outlining a method 400 for building and applying a generative artificial intelligence (Al) model, specifically a language model (LM) (e.g., such as a large language model (LLM)), according to some aspects. The method 400 may include the creation and / or utilization of an LM to develop a nurse practitioner assistant. The method 400 begins with data preparation and sampling (block 402), which includes processing data pertinent to a given application of the LM. Subsequently, the method 400 involves the construction of an LM (block 404) which encompasses the formulation of the model's architecture (block 406), as well as pretraining (block 408) with an attention mechanism (block 410). The pretraining phase is undertaken to establish a foundational model (block 412) that learns from the prepared data, suited to the LM's intended use. Within the foundational model (block 412), additional steps of training (block 414) and model evaluation (block 416) are performed. During this phase, the model may also benefit from pretrained weights (block 418), which can accelerate the training by drawing on previously acquired patterns and knowledge.

[0060] After the foundational model is established, the process includes finetuning (block 420), which may include the calibration and optimization stage, utilizing more focused data pertaining to the specific application of the LM. This phase may be utilized to hone andPatent Application 31134 / 70821 / PC improve the LM to produce accurate outcomes within the intended operational domain.Finally, an instructions dataset (block 424) is provided to the nurse practitioner assistant (block 401). This dataset may contain structured instructions and prompts that facilitate the LM in generating relevant analyses and responses. These prompts promote the practical implementation of the trained model, enabling the model to function effectively as part of the nurse practitioner assistant. FIG. 4 illustrates a structured approach for developing a specialized LM aimed at a certain application, starting with data preparation and sampling, progressing through model construction with architecture development and pretraining, and culminating in the finetuning and deployment of a nurse practitioner assistant that operates in conjunction with a dedicated instructions dataset. This visual aid encapsulates the sequence of steps necessary to forge a tool that supports the specific application for which the LM is designed, according to some aspects.

[0061] The language model (LM) outlined in method 400 directly supports the functions of the server computing device 102, specifically the ML operation module 142, within the computing environment 100. The initial steps of data preparation and sampling for the LM (block 402) align with the data collection and preparation activities of the ML operation module 142 and the data processing module 141 to develop ML models that can accurately summarize medical information for a patient. Further, the ML operation module 142 utilizes a variety of data sources, including healthcare platforms / systems (e.g., Lippincott, OnBase, Micrometrics, Center, etc.) and other medical data resources to train neural network models and generative Al models. In some embodiments, the data processing module 141 and / or the ML operation module 142 organizes and preprocesses data obtained from medical data sources 110, ensuring suitability of the data for training the generative Al models 146.Furthermore, the communication interfaces 130 of the server computing device 102 enable interconnected exchanges between the foundational models 412 and external resources, such as the medical data sources 110 (e.g., via the APIs 112), to develop the generative Al models 146 by enriching the foundational model 412 with diverse data inputs.

[0062] The construction phase of the LM, including the formulation of the model's architecture (block 406) and pretraining with an attention mechanism (blocks 408 and 410). parallels the development process within the ML operation module 142. This process involves using machine learning and digital medical libraries to build models capable ofPatent Application 31134 / 70821 / PC understanding and accurately summarizing medical information and / or patient data. The foundational model (block 412) of the LM, which undergoes additional training (block 414) and evaluation (block 416), reflects the iterative training and refinement process of ML models within the ML operation module 142. This module may leverage historical medical information and example summaries to enhance the accuracy and efficiency of the Al-based summarization process. The finetuning stage (block 420) of the LM may operate and / or be utilized for optimizing the model's performance for its specific application, mirroring the continuous improvement efforts of the ML operation module 142 to ensure the models remain relevant and accurate as medical documentation (e.g., clinical guidelines, procedures, educational materials, etc.) evolves and / or changes. Furthermore, the prompt generation module 144 may generate prompts that cause the generative Al modes 146 to perform context retrieval (sometimes referred to as “retrieval augmented generation”) from medical data sources 110, thereby enabling the foundational model to emphasize critical medical categories (e.g., surgical history, vital signs, etc.) accurately.

[0063] Additionally, the instructions dataset (block 424) provided to the nurse practitioner assistant (block 401) guides the model in summarizing a particular type of medical information for the patient. In some embodiments, the prompt generation module 144 generates input instructions for refining the generative Al models 146 during finetuning. For example, the described techniques may involve generating prompts that cause the generative Al models 146 to perform context retrieval from medical data sources 110. Furthermore, the data snippets from medical data sources 110 serve as references to improve the accuracy and reliability of the generated summaries, thereby reducing the likelihood of hallucinations or false information. Furthermore, the prompt generation module 144 improves the likelihood of a foundational model being able to achieve clinically relevant results by interpolating patient-specific information and combining it with structured templates, as described above with respect to FIG. 1. At a high level, the validation module 148 may ensure that the nurse practitioner assistant (block 401) accurately functions as intended, avoiding potential errors or hallucinations in its responses.

[0064] Additionally, the different medical categories may require different models to accurately analyze and summarize the corresponding medical data. For example, some categories may include image or video data, such as x-rays or other medical images.Patent Application 31134 / 70821 / PC Continuing with the above example, the ML operation module 142 and / or the data processing module 141 may employ optical character recognition (OCR) or object recognition techniques to extract relevant data for the generative Al models 146. For example, the architecture, training, and implementation of the foundational models 412 and / or the generative Al models 146 may be tailored to the unique requirements and demands of each medical category. Furthermore, an example generative Al model, such as the model 146, may include image recognition, or OCR, capabilities for analyzing image or text data within a specific medical category.

[0065] As mentioned above, the described techniques may include obtaining input data from a clinician (e.g., via clinician computing device 104), which includes indications of one or more medical categories. Based on the medical categories, the server computing device 102 identifies one or more trained generative Al models associated with each category. For example, each trained generative Al model 146 is specialized and efficient in processing specific types of medical data. Furthermore, each prompt provided to the respective generative Al models also corresponds to a specific medical category, guiding the models in generating accurate and relevant medical summaries. This approach ensures that the generated summaries are tailored for the specific medical category, enhancing the accuracy and specificity of the information provided. For example, if a prompt is associated with the medical category of vital signs, generative Al model 146 may be trained specifically on analyzing vital sign information and can provide a summary that is specific to vital signs and includes all the relevant details. Similarly, for other medical categories such as surgical history or nursing physical assessments, specific generative Al models are chosen to provide targeted analysis and summaries.

[0066] FIG. 5 illustrates a neural network-based model architecture for processing and analyzing medical information for a patient (block 502). The process begins with data collection (block 530) which is then passed through preprocessing layers, specifically a data normalization layer (block 512a) and a feature extraction layer (block 512b). The aforementioned layers are followed by a dropout layer (block 514) to prevent overfitting.The core of the architecture is the neural network loop (block 516), which is iterated N times, where N is a positive integer. Each iteration consists of a normalization layer (block 520a), followed by an attention layer (block 522) with its own dropout layer (block 524a),Patent Application 31134 / 70821 / PC another normalization layer (block 520b), a dense layer (block 526), and another dropout layer (block 524b). The process concludes with a final normalization layer (block 517) and a linear output layer (block 518), producing the final output from the neural network-based model. This architecture is designed to handle and analyze data for summarizing medication information for a patient, or patient data.

[0067] The model architecture depicted in FIG. 5 may be used to analyze and evaluate medical information and data, particularly in the context of summarizing a patients medical information / data. The neural network-based architecture facilitates the processing of diverse data through a series of layers and loops designed to understand and identify patterns related to medical information and particular medical categories (e.g., medical history, medications, known allergies, vital signs, radiology or other medical images, etc.). Initially, the collected data is processed through preprocessing layers, including normalization and feature extraction layers, which help the model understand the significance of each data point within the context of a particular medical category. As such, the model is enabled for effectively summarizing medical information as the system is enabled to replicate and / or identify the nuances of the particular medical categories, including common data types, common formats, and common content. The dropout layers introduced after the preprocessing layers and within the neural network loop serve to prevent overfitting by randomly omitting some of the units from the layers during training. The dropout layers, then, ensure that the model does not become too reliant on the training data, allowing the model to generalize more effectively to new, unseen data. The neural network loop, iterated N times, is where the bulk of the analysis happens. Each iteration consists of a series of layers including normalization, attention, and dense layers, each followed by dropout layers. The normalization layers help stabilize the learning process, while the attention layers allow the model to focus on different parts of the input data to better understand the relationships between various factors relevant to a particular medical category. The dense layers, on the other hand, are fully connected layers that help in learning non-linear combinations of the features. The final normalization layer ensures that the data is normalized before passing it to the linear output layer, which produces the final output of the model. By leveraging the capabilities of Al-based nursing tools, clinicians can engage with an example machinePatent Application 31134 / 70821 / PC learning model to evaluate (e.g., summarize) medical information for a patient on a granular level (e.g., the medical summaries may be confined to particular categories).

[0068] FIG. 6 illustrates an example computer-implemented method 600 for generating medical data summaries for a patient using generative artificial intelligence (Al) models, implemented by a computing system (e.g., system 100). It will be understood that the method 600 may be performed by other systems and / or components thereof (e.g., other systems configured to perform similar operations, and / or components thereof).

[0069] At block 602, a computing device (e.g., server computing device 102 and / or clinician computing device 104 of FIG. 1) obtains medical information for a patient. For example, the medical information for the patient may include medical history, medications, known allergies, vital signs, radiology or other medical images, etc. In some embodiments, the medical information for the patient is obtained from an electronic health record (EHR) database. Additionally or alternatively, the computing system may obtain the medical information for a patient, or an EHR for the patient, from a user computing device (e.g., clinician computing device 104).

[0070] At block 604, the computing device obtains input data associated with the patient from a user (e.g., a nurse practitioner or another type of clinician). In some embodiments, the input data includes indications of one or more medical categories. Additionally, the method 600 may include presenting, via a graphical user interface (GUI) of a user computing device (e.g., the clinician computing device 104), a plurality of selectable medical categories for review by the user. Further, the method 600 may include obtaining, via the GUI, selections from among the plurality of medical categories from the user. As depicted in FIG. 3A, the medical categories may include various configurable categories, such as (i) situation, (ii) background, (iii) surgical history, (iv) vital signs, (v) nursing physical assessment, etc. In some embodiments, the input data includes indications of one or more respective sub-categories for each medical category of the one or more medical categories. For example, the nursing physical assessment category may include one or more configurable sub-categories, such as (i) alertness and orientation, (ii) ability to communicate, (iii) intervention history (e.g., number of intravenous attempts over a period of time), etc.Patent Application 31134 / 70821 / PC

[0071] Tn some embodiments, the method 600 includes, based on the input data, generating one or more prompts for one or more trained generative Al models using at least a portion of the medical information for the patient. For example, each prompt of the one or more prompts may correspond to a respective medical category of the one or more medical categories. In some embodiments, the method 600 includes generating the one or more prompts by causing the computing system to identify, based on the one or more medical categories, one or more base prompts for the one or more trained generative Al models. For example, each respective base prompt may include additional context that causes the each trained generative Al model to generate intermediary conclusions while generating an output, and evaluate accuracy of the intermediary conclusions while generating the output. Continuing with the above example, each respective base prompt may be formatted for a specific medical category. For example, a first base prompt may be associated with a medical category that involves data extraction and may include a corresponding set of instructions that guide a model in extracting data from input data. As a further example, a second base prompt may be associated with a medical category that involves reasoning and may include a corresponding set of instructions that guides a model in generating inferences based on input data (e.g., the complexity of the second base prompt may exceed that of the first base prompt).

[0072] Moreover, the computing device may generate, refine, and / or fine-tune each respective base prompt for a particular medical category, and each respective base prompt may correspond to a respective type of generative Al model. In some embodiments, the method 600 includes generating the one or more prompts by causing the computing system to identify one or more respective prompt sections for each medical category based on the one or more respective sub-categories for each medical category. Additionally or alternatively, the computing device may generate the one or more prompts by extracting information associated with each respective medical category of the one or more medical categories from the medical information for the patient. Further, the computing device may generate the one or more prompts by identifying one or more respective prompt sections for the each medical category based on the one or more respective sub-categories for the each medical category.Patent Application 31134 / 70821 / PC

[0073] The one or more trained generative Al models (e.g., the generative Al models 146) may be a machine learning transformer model such as a transformer-based language model or large language model (e.g., a generative pre-trained transformer, a bidirectional encoder representations from transformers model, etc.), as described in further detail above with respect to FIG. 4 and FIG. 5. Further, each prompt of the one or more prompts, similar to each respective medical category, may correspond to a respective generative Al model of the one or more generative Al models. For instance, a prompt that is associated with data extraction may correspond to a model that excels in data extraction tasks, whereas a prompt that is associated with reasoning may correspond to a model that excels in inference / reasoning tasks.

[0074] At block 606, the computing device identifies one or more trained generative Al models for analyzing the medical information for the patient using the input data. In some embodiments, each trained generative Al model of the one or more trained generative Al models (e.g., generative Al models 146) is associated with a respective medical category of the one or more medical categories. For example, the one or more trained generative Al models may be identified based on the one or more medical categories indicated in the input data. In some embodiments, the each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources (e.g., medical data sources 110). Additionally and / or alternatively, the method 600 may include obtaining historical medical data from at least one data source of the medical data sources 110. In some embodiments, one or more medical data sources of the plurality of medical data sources may correspond to a respective medical category.Furthermore, the method 600 may include obtaining training data for a particular medical category, and a corresponding generative Al model, from medical data sources associated with the particular category. For example, the method 600 may include generating, via the one or more trained generative Al models, one or more queries based on the medical information for the patient, wherein each respective query of the one or more queries is associated with a respective medical category of the one or more medical categories.Continuing with this example, the method 600 may include issuing each respective query to and / or against the at least one data source (e.g., the one or more medical data sources associated with the corresponding medical category).Patent Application 31134 / 70821 / PC

[0075] At block 608, the computing device generates one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources via the one or more trained generative Al models. In some embodiments, the method 600 includes generating the one or more summaries associated with the medical information for the patient by causing the computing system to obtain, via the one or more trained generative Al models and based on the medical information for the patient, the historical medical data from the at least one data source. For example, each summary of the one or more summaries may be generated using one or more respective data snippets from the historical medical data. Continuing with the above example, each summary may be associated with a respective medical category of the one or more medical categories. In some embodiments, the method 600 includes generating the one or more summaries associated with the medical information for the patient by causing the computing system to validate each summary against the one or more respective data snippets using one or more additional generative Al models.

[0076] As mentioned above, the instant techniques may block, delete, or otherwise prevent various types or sources of data from being used in analysis, data gathering, and / or training of a generative model, providing improvements to the overall performance and accuracy of the model and / or outcomes. For example, an example prompt may cause a generative Al model to ignore or redact data related to historical patients (e.g., medical information for an individual other than the patient) from data obtained from the one or more medical data sources. As another example, an example prompt may cause a generative Al model to only access particular resources from among the one or more medical data sources 110.

[0077] At block 610, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user is presented via a graphical user interface on a computing device. In some embodiments, the method 600 includes presenting, via a graphical user interface and with each summary, indications of the one or more respective data snippets to the user. In some embodiments, the method 600 further includes presenting, via the graphical user interface, the one or more nursing summaries in a curated order and one or more web links to contextual practice resources, native practice resources, and evidence-based practice resources included in the one or more nursing dataPatent Application 31134 / 70821 / PC sources. Moreover, the method 600 may include presenting curated nurse patient summary and direct links (e.g., hyperlinks and / or web links) to contextual, native, and evidence-based practice resources from multiple disperse locations to the fingertips of nurses at the point of care.

[0078] Generally, the method 600 may include generating a respective summary for each determined / indicated medical category by inputting a corresponding subset of training medical data, from the medical data sources 110, and the medical information for a patient to each respective generative Al model. As mentioned above, the described techniques may advantageously include summarizing medical information for a patient using an ensemble of generative Al models 146, whereby each generative Al model is instructed (e.g., via an input prompt) to evaluate the medical information with respect to medical data (e.g., from the medical data sources 110) associated with a corresponding medical category.Furthermore, the method 600 eliminates the need for additional training or finetuning of a generative Al model with respect to a particular language processing task by identifying and associating the proper model(s) for generating and / or analyzing data with the corresponding medical information category. Moreover, the method 600 may reduce the presence of hallucinations or other errors, at least through confirmation of accuracy, introduction of citations, and other such techniques as described herein.

[0079] The following list of examples reflects a variety of the embodiments explicitly contemplated by the present disclosure. Those of ordinary skill in the art will readily appreciate that the examples below are neither limiting of the embodiments disclosed herein, nor exhaustive of all of the embodiments conceivable from the disclosure above, but are instead meant to be exemplary in nature.

[0080] Example 1. A computer- implemented method for generating medical data summaries for a patient using generative artificial intelligence (Al) models, the method comprising: obtaining, via one or more processors, medical information for a patient; obtaining, via the one or more processors, input data associated with the patient from a user, the input data including indications of one or more medical categories; identifying, via the one or more processors and using the input data, one or more trained generative Al models for analyzing the medical information for the patient, wherein: each trained generative AlPatent Application 31134 / 70821 / PC model of the one or more trained generative AT models is associated with a respective medical category of the one or more medical categories, and the each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources; generating, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; and presenting, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

[0081] Example 2. The computer-implemented method of example 1 , wherein generating the one or more summaries associated with the medical information for the patient includes: obtaining, via the one or more trained generative Al models and based on the medical information for the patient, the historical medical data from the at least one data source, wherein: each summary of the one or more summaries is generated using one or more respective data snippets from the historical medical data, and the each summary is associated with a respective medical category of the one or more medical categories.

[0082] Example 3. The computer-implemented method of example 2, wherein obtaining the historical medical data from the at least one data source includes: generating, via the one or more trained generative Al models, one or more queries based on the medical information for the patient, wherein each respective query of the one or more queries is associated with a respective medical category of the one or more medical categories; and issuing, via the one or more processors, the each respective query against the at least one data source.

[0083] Example 4. The computer-implemented method of example 2, wherein generating the one or more summaries associated with the medical information for the patient includes: validating, via the one or more processors, the each summary against the one or more respective data snippets using one or more additional generative Al models.

[0084] Example 5. The computer-implemented method of example 4, further including: presenting, via a graphical user interface and with the each summary, indications of the one or more respective data snippets to the user.Patent Application 31134 / 70821 / PC

[0085] Example 6. The computer-implemented method of example 1 , further including: based on the input data, generating, via the one or more processors, one or more prompts for the one or more trained generative Al models using at least a portion of the medical information for the patient, wherein each prompt of the one or more prompts corresponds to a respective medical category of the one or more medical categories.

[0086] Example 7. The computer-implemented method of example 6, wherein generating the one or more prompts includes: identifying, via the one or more processors and based on the one or more medical categories, one or more base prompts for the one or more trained generative Al models, wherein each respective base prompt includes additional context that causes the each trained generative Al model to: generate intermediary conclusions while generating an output, and evaluate accuracy of the intermediary conclusions while generating the output.

[0087] Example 8. The computer-implemented method of example 6, wherein generating the one or more prompts includes: extracting, via the one or more processors, information associated with each respective medical category of the one or more medical categories from the medical information for the patient.

[0088] Example 9. The computer-implemented method of example 6, wherein the input data includes indications of one or more respective sub-categories for each medical category of the one or more medical categories, and wherein generating the one or more prompts includes: identifying, via the one or more processors, one or more respective prompt sections for the each medical category based on the one or more respective sub-categories for the each medical category.

[0089] Example 10. The computer-implemented method of example 1. wherein: the one or more medical categories include one or more nursing categories; and the one or more summaries associated with the medical information for the patient include one or more nursing summaries.

[0090] Example 11. The computer-implemented method of example 1 , wherein the plurality of medical data sources include a plurality of nursing data sources.Patent Application 31134 / 70821 / PC

[0091] Example 12. The computer-implemented method of example 1, wherein the each trained generative Al model is trained using a corresponding subset of training medical data from one or more nursing data sources of a plurality of nursing data sources.

[0092] Example 13. A computing system for generating medical data summaries for a patient using generative artificial intelligence (Al) models, the computing system comprising: one or more processors; one or more non-transitory memories, the one or more non-transitory memories having stored thereon computer-executable instructions that, when executed by the one or more processors, cause the computing system to: obtain medical information for a patient; obtain input data associated with the patient from a user, the input data including indications of one or more medical categories; identify, using the input data, one or more trained generative Al models for analyzing the medical information for the patient, wherein: each trained generative Al model of the one or more trained generative Al models is associated with a respective medical category of the one or more medical categories, and the each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources; generate, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; and present, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

[0093] Example 14. The computing system of example 13, the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, generate the one or more summaries associated with the medical information for the patient by causing the computing system to: obtain, via the one or more trained generative Al models and based on the medical information for the patient, the historical medical data from the at least one data source, wherein: each summary of the one or more summaries is generated using one or more respective data snippets from the historical medical data, and the each summary is associated with a respective medical category of the one or more medical categories.Patent Application 31134 / 70821 / PC

[0094] Example 15. The computing system of example 14, the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, obtain the historical medical data from the at least one data source by causing the computing system to: generate, via the one or more trained generative Al models, one or more queries based on the medical information for the patient, wherein each respective query of the one or more queries is associated with a respective medical category of the one or more medical categories; and issue the each respective query against the at least one data source.

[0095] Example 16. The computing system of example 14, the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, generate the one or more summaries associated with the medical information for the patient by causing the computing system to: validate the each summary against the one or more respective data snippets using one or more additional generative Al models.

[0096] Example 17. The computing system of example 16, the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, further cause the computing system to: present, via a graphical user interface and with the each summary, indications of the one or more respective data snippets to the user.

[0097] Example 18. The computing system of example 13, the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, further cause the computing system to: based on the input data, generate one or more prompts for the one or more trained generative Al models using at least a portion of the medical information for the patient, wherein each prompt of the one or more prompts corresponds to a respective medical category of the one or more medical categories.

[0098] Example 19. The computing system of example 18, the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, generate the one or more prompts by causing the computing system to: identify, based on the one or more medical categories, one or more base promptsPatent Application 31134 / 70821 / PC for the one or more trained generative Al models, wherein each respective base prompt includes additional context that causes the each trained generative Al model to: generate intermediary conclusions while generating an output, and evaluate accuracy of the intermediary conclusions while generating the output.

[0099] Example 20. The computing system of example 18, the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, generate the one or more prompts by causing the computing system to: extract information associated with each respective medical category of the one or more medical categories from the medical information for the patient.

[0100] Example 21. The computing system of example 18. wherein the input data includes indications of one or more respective sub-categories for each medical category of the one or more medical categories, and the one or more non-transitory memories having stored thereon computer executable instructions that, when executed by the one or more processors, generate the one or more prompts by causing the computing system to: identify one or more respective prompt sections for the each medical category based on the one or more respective sub-categories for the each medical category.

[0101] Example 22. The computing system of example 13, wherein: the one or more medical categories include one or more nursing categories; and the one or more summaries associated with the medical information for the patient include one or more nursing summaries.

[0102] Example 23. The computing system of example 13, wherein the plurality of medical data sources include a plurality of nursing data sources.

[0103] Example 24. The computing system of example 13. wherein the each trained generative Al model is trained using a corresponding subset of training medical data from one or more nursing data sources of a plurality of nursing data sources.

[0104] Example 25. A tangible, non-transitory computer readable medium storing computer-readable instructions that, when executed by one or more processors of a computing system, cause the computing system to: obtain medical information for a patient; obtain input data associated with the patient from a user, the input data including indications of one or more medical categories; identify, using the input data, one or more trainedPatent Application 31134 / 70821 / PC generative Al models for analyzing the medical information for the patient, wherein: each trained generative Al model of the one or more trained generative Al models is associated with a respective medical category of the one or more medical categories, and the each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources; generate, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; and present, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

[0105] Example 26. The tangible, non-transitory computer readable medium of example 25, wherein the computer-readable instructions, when executed by the one or more processors, further cause the computing system to: obtain, via the one or more trained generative Al models and based on the medical information for the patient, the historical medical data from the at least one data source, wherein: each summary of the one or more summaries is generated using one or more respective data snippets from the historical medical data, and the each summary is associated with a respective medical category of the one or more medical categories.

[0106] Example 27. The tangible, non-transitory computer readable medium of example 25, wherein: the one or more medical categories include one or more nursing categories; and the one or more summaries associated with the medical information for the patient include one or more nursing summaries.

[0107] Example 28. The tangible, non-transitory computer readable medium of example 25, wherein the plurality of medical data sources include a plurality of nursing data sources.

[0108] Example 29. The tangible, non-transitory computer readable medium of example 25, wherein the each trained generative Al model is trained using a corresponding subset of training medical data from one or more nursing data sources of a plurality of nursing data sources.

[0109] Example 30. A computer-implemented method for generating nursing summaries for a patient using generative artificial intelligence (Al) models, the method comprising:Patent Application 31134 / 70821 / PC obtaining, via one or more processors, medical information for a patient; obtaining, via the one or more processors, input data associated with the patient from a nurse practitioner, the input data including indications of one or more nursing categories; identifying, via the one or more processors and using the input data, one or more trained generative Al models for analyzing the medical information for the patient, wherein: each trained generative Al model of the one or more trained generative Al models is associated with a respective nursing category of the one or more nursing categories, and the each trained generative Al model is trained using a corresponding subset of training medical data from one or more nursing data sources of a plurality of nursing data sources; generating, via the one or more trained generative Al models, one or more nursing summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of nursing data sources; and presenting, via a graphical user interface on a computing device, at least one nursing summary of the one or more nursing summaries associated with the medical information for the patient for review by the nurse practitioner.

[0110] Example 31. The computer-implemented method of example 30, further including: presenting, via the graphical user interface, the one or more nursing summaries in a curated order and one or more web links to contextual practice resources, native practice resources, and evidence-based practice resources included in the one or more nursing data sources.

[0111] Throughout this specification, plural instances may implement components, operations, or structures described as a single instance. Although individual operations of one or more methods are illustrated and described as separate operations, one or more of the individual operations may be performed concurrently, and nothing requires that the operations be performed in the order illustrated. Structures and functionality presented as separate components in example configurations may be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of the subject matter herein.Patent Application 31134 / 70821 / PC

[0112] Additionally, certain embodiments are described herein as including logic or a number of routines, subroutines, applications, or instructions. These may constitute either software (e.g., code embodied on a machine-readable medium or in a transmission signal) or hardware. In hardware, the routines, etc., are tangible units capable of performing certain operations and may be configured or arranged in a certain manner. In example embodiments, one or more computer systems (e.g., a standalone, client or server computer system) such as system 100 of FIG. 1 or one or more hardware modules of a computer system (e.g., a processor or a group of processors) such as system 100 may be configured by software (e.g., an application or application portion) as a hardware module that operates to perform certain operations as described herein. Similarly, the computer systems or hardware modules of the computer systems may be configured to execute stored instructions on a memory as described herein to perform any such operations as described herein.

[0113] In various embodiments, a hardware module may be implemented mechanically or electronically. For example, a hardware module may comprise dedicated circuitry or logic that is permanently configured (e.g., as a special-purpose processor, such as a field programmable gate array (FPGA) or an application- specific integrated circuit (ASIC)) to perform certain operations. A hardware module may also comprise programmable logic or circuitry (e.g., as encompassed within a general-purpose processor or other programmable processor) that is temporarily configured by software to perform certain operations. It will be appreciated that the decision to implement a hardware module mechanically, in dedicated and permanently configured circuitry, or in temporarily configured circuitry (e.g., configured by software) may be driven by cost and time considerations.

[0114] Accordingly, the term "hardware module" should be understood to encompass a tangible entity, be that an entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a certain manner or to perform certain operations described herein. Considering embodiments in which hardware modules are temporarily configured (e.g., programmed), each of the hardware modules need not be configured or instantiated at any one instance in time. For example, where the hardware modules comprise a general-purpose processor configured using software, the general-purpose processor may be configured as respective different hardware modules at different times. Software may accordingly configure a processor, for example, to constitute aPatent Application 31134 / 70821 / PC particular hardware module at one instance of time and to constitute a different hardware module at a different instance of time.

[0115] Hardware modules can provide information to, and receive information from, other hardware modules. Accordingly, the described hardware modules may be regarded as being communicatively coupled. Where multiple of such hardware modules exist contemporaneously, communications may be achieved through signal transmission (e.g., over appropriate circuits and buses) that connects the hardware modules. In embodiments in which multiple hardware modules are configured or instantiated at different times, communications between such hardware modules may be achieved, for example, through the storage and retrieval of information in memory structures to which the multiple hardware modules have access. For example, one hardware module may perform an operation and store the output of that operation in a memory device to which it is communicatively coupled. A further hardware module may then, at a later time, access the memory device to retrieve and process the stored output. Hardware modules may also initiate communications with input or output devices, and can operate on a resource (e.g., a collection of information).

[0116] The various operations of the example methods described herein may be performed, at least partially, by one or more processors that are temporarily configured (e.g., by software) or that are permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors may constitute processor-implemented modules that operate to perform one or more operations or functions. The modules referred to herein may, in some example embodiments, comprise processor-implemented modules.

[0117] Similarly, the methods or routines described herein may be at least partially processor-implemented. For example, at least some of the operations of a method may be performed by one or more processors or by processor-implemented hardware modules. The performance of certain of the operations may be distributed among the one or more processors, not only residing within a single machine (having different processing abilities), but also deployed across a number of machines. In some example embodiments, the processors may be located in a single location (e.g.. deployed in the field, in an officePatent Application 31134 / 70821 / PC environment, or as part of a server farm), while in other embodiments the processors may be distributed across a number of locations.

[0118] Unless specifically stated otherwise, discussions herein using words such as "processing," "computing," "calculating," "determining," "presenting," "displaying," or the like may refer to actions or processes on a GPU thread that manipulates or transforms data represented as physical (e.g., electronic, magnetic, or optical) quantities within one or more memories (e.g., volatile memory, non-volatile memory, or a combination thereof), registers, or other machine components that receive, store, transmit, or display information.

[0119] As used herein any reference to "one embodiment" or "an embodiment" means that a particular element, feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment.

[0120] Some embodiments may be described using the expression "coupled" and "connected" along with their derivatives. For example, some embodiments may be described using the term "coupled" to indicate that two or more elements are in direct physical or electrical contact. The term "coupled," however, may also mean that two or more elements are not in direct contact with each other, but yet still co-operate or interact with each other. The embodiments are not limited in this context.

[0121] As used herein, the terms "comprises," "comprising," "includes." "including," "has," "having" or any other variation thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Further, unless expressly stated to the contrary, "or" refers to an inclusive or and not to an exclusive or. For example, a condition A or B is satisfied by any one of the following: A is true (or present) and B is false (or not present), A is false (or not present) and B is true (or present), and both A and B are true (or present).

[0122] In addition, use of the "a" or "an" are employed to describe elements and components of the embodiments herein. This is done merely for convenience and to give aPatent Application 31134 / 70821 / PC general sense of the description. This description, and the claims that follow, should be read to include one or at least one and the singular also includes the plural unless it is obvious that it is meant otherwise.

[0123] This detailed description is to be construed as an example only and does not describe every possible embodiment, as describing every possible embodiment would be impractical, if not impossible. One could implement numerous alternate embodiments, using either current technology or technology developed after the filing date of this application.

Claims

Patent Application 31134 / 70821 / PC WHAT IS CLAIMED:

1. A computer-implemented method for generating medical data summaries for a patient using generative artificial intelligence (Al) models, the method comprising:obtaining, via one or more processors, medical information for a patient;obtaining, via the one or more processors, input data associated with the patient from a user, the input data including indications of one or more medical categories;identifying, via the one or more processors and using the input data, one or more trained generative Al models for analyzing the medical information for the patient, wherein:each trained generative Al model of the one or more trained generative Al models is associated with a respective medical category of the one or more medical categories, andthe each trained generative Al model is trained using a corresponding subset of training medical data from one or more data sources of a plurality of medical data sources;generating, via the one or more trained generative Al models, one or more summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of medical data sources; andpresenting, via a graphical user interface on a computing device, at least one summary of the one or more summaries associated with the medical information for the patient for review by the user.

2. The computer-implemented method of claim 1, wherein generating the one or more summaries associated with the medical information for the patient includes;obtaining, via the one or more trained generative Al models and based on the medical information for the patient, the historical medical data from the at least one data source, wherein:each summary of the one or more summaries is generated using one or more respective data snippets from the historical medical data, andthe each summary is associated with a respective medical category of the one or more medical categories.Patent Application 31134 / 70821 / PC 3. The computer-implemented method of claim 2, wherein obtaining the historical medical data from the at least one data source includes:generating, via the one or more trained generative Al models, one or more queries based on the medical information for the patient, wherein each respective query of the one or more queries is associated with a respective medical category of the one or more medical categories; andissuing, via the one or more processors, the each respective query against the at least one data source.

4. The computer-implemented method of claim 2, wherein generating the one or more summaries associated with the medical information for the patient includes:validating, via the one or more processors, the each summary against the one or more respective data snippets using one or more additional generative Al models.

5. The computer-implemented method of claim 4, further including: presenting, via a graphical user interface and with the each summary, indications of the one or more respective data snippets to the user.

6. The computer-implemented method of claim 1, further including:based on the input data, generating, via the one or more processors, one or more prompts for the one or more trained generative Al models using at least a portion of the medical information for the patient,wherein each prompt of the one or more prompts corresponds to a respective medical category of the one or more medical categories.

7. The computer-implemented method of claim 6, wherein generating the one or more prompts includes:identifying, via the one or more processors and based on the one or more medical categories, one or more base prompts for the one or more trained generative Al models,wherein each respective base prompt includes additional context that causes the each trained generative Al model to:Patent Application 31134 / 70821 / PC generate intermediary conclusions while generating an output, and evaluate accuracy of the intermediary conclusions while generating the output.

8. The computer-implemented method of claim 6, wherein generating the one or more prompts includes:extracting, via the one or more processors, information associated with each respective medical category of the one or more medical categories from the medical information for the patient.

9. The computer-implemented method of claim 6, wherein the input data includes indications of one or more respective sub-categories for each medical category of the one or more medical categories, and wherein generating the one or more prompts includes:identifying, via the one or more processors, one or more respective prompt sections for the each medical category based on the one or more respective sub-categories for the each medical category.

10. The computer- implemented method of claim 1, wherein:the one or more medical categories include one or more nursing categories; and the one or more summaries associated with the medical information for the patient include one or more nursing summaries.

11. The computer-implemented method of claim 1, wherein the plurality of medical data sources include a plurality of nursing data sources.

12. The computer- implemented method of claim 1, wherein the each trained generative Al model is trained using a corresponding subset of training medical data from one or more nursing data sources of a plurality of nursing data sources.

13. A computer-implemented method for generating nursing summaries for a patient using generative artificial intelligence (Al) models, the method comprising:Patent Application 31134 / 70821 / PC obtaining, via one or more processors, medical information for a patient; obtaining, via the one or more processors, input data associated with the patient from a nurse practitioner, the input data including indications of one or more nursing categories;identifying, via the one or more processors and using the input data, one or more trained generative Al models for analyzing the medical information for the patient, wherein:each trained generative Al model of the one or more trained generative Al models is associated with a respective nursing category of the one or more nursing categories, andthe each trained generative Al model is trained using a corresponding subset of training medical data from one or more nursing data sources of a plurality of nursing data sources;generating, via the one or more trained generative Al models, one or more nursing summaries associated with the medical information for the patient based on historical medical data from at least one data source of the plurality of nursing data sources; andpresenting, via a graphical user interface on a computing device, at least one nursing summary of the one or more nursing summaries associated with the medical information for the patient for review by the nurse practitioner.

14. The computer-implemented method of claim 13, further including: presenting, via the graphical user interface, the one or more nursing summaries in a curated order and one or more web links to contextual practice resources, native practice resources, and evidence-based practice resources included in the one or more nursing data sources.

15. A computing system for generating medical data summaries for a patient using generative artificial intelligence (Al) models, the computing system comprising a transceiver and processing hardware configured to implement a method according to any one of the preceding claims.

Images