Systems, methods, and computer program products for a platform for the development and deployment of artificial intelligence-based software as medical devices.

Abstract

A system is provided, including an AI medical device platform for generating artificial intelligence (AI)-based automated healthcare applications, the AI ​​medical device platform comprising at least one processor communicating with a data repository system, the at least one processor configured to receive formatted data associated with multiple medical procedures for generating one or more machine learning models, receive data associated with one or more regulatory guidance and approval processes, generate one or more machine learning models configured to predict one or more aspects associated with medical procedures based on the data associated with multiple medical procedures and the data associated with regulatory guidance and approval processes, and, based on the generation of one or more machine learning models, deploy one or more machine learning models to a production environment. Methods and computer program products are also disclosed.

Description

Technical Field

[0001] Cross - reference to Related Applications This application claims the benefit of U.S. Provisional Application No. 63 / 503,347, filed May 19, 2023, the entire disclosure of which is incorporated herein by reference in its entirety.

Background Art

[0002] 1. Field The present disclosure generally relates to managing health informatics applications, and in some non - limiting embodiments, to systems, methods, and computer program products that provide a platform for the development and deployment of automated healthcare applications, including artificial intelligence (AI) - based software as a medical device.

[0003] 2. Technical Considerations Artificial intelligence (AI) can be used in healthcare as a way to mimic human cognition in the analysis, presentation, and / or understanding of healthcare data. AI can describe the ability of a computer program, such as a computer algorithm, to approximate conclusions based on input data that can be medical data. In some cases, computer algorithms can be used to recognize patterns in data and create logic for identifying such patterns. Such computer algorithms can include machine - learning models trained to perform specific tasks using large amounts of input data. In some cases, healthcare - based computer algorithms can include machine - learning models configured to provide outputs regarding medical treatments.

[0004] Software as a Medical Device (SaMD) may refer to software intended for use in one or more medical purposes that perform these purposes, without necessarily being installed as part of a hardware medical device. In some cases, SaMD may include machine learning models specifically designed to provide outputs related to medical procedures, such as predicting diagnoses. For example, SaMD may range from software that enables a smartphone to view images obtained from a magnetic resonance imaging (MRI) medical device for diagnostic purposes to computer-aided detection (CAD) software that performs image post-processing to help detect breast cancer.

[0005] However, building SaMDs and the machine learning models they contain can require large amounts of data. Therefore, manually programmed processes developed to provide input data to machine learning models may require significant network resources, may require constant manual updates, and may be inaccurate. Furthermore, the datasets used to generate the machine learning models may be stored in various different locations, potentially requiring considerable effort to retrieve and use for model generation. Additionally, SaMDs may require regulatory guidance and approvals at various stages of development and deployment. Determining whether a SaMD meets applicable regulatory guidance and approvals may require extensive resources to retrieve and remember applicable regulations before and during development, as well as extensive resources to track when the SaMD is deployed. [Overview of the project] [Means for solving the problem]

[0006] Accordingly, systems, methods, and computer program products that provide a platform for the development and deployment of automated healthcare applications are disclosed.

[0007] Clause 1: A system for generating artificial intelligence (AI) based automated healthcare applications, comprising: a data repository system storing at least one of formatted data associated with a plurality of medical procedures for the generation of one or more machine learning models, data associated with regulatory guidance and approval processes, or any combination thereof; and an AI medical device platform having at least one processor communicating with the data repository system, wherein the at least one processor is configured to receive formatted data associated with a plurality of medical procedures for the generation of one or more machine learning models, receive data associated with one or more regulatory guidance and approval processes, generate one or more machine learning models configured to predict one or more aspects associated with a medical procedure based on the formatted data associated with a plurality of medical procedures and the data associated with regulatory guidance and approval processes, and, based on the generation of one or more machine learning models, deploy one or more machine learning models to a production environment.

[0008] Clause 2: The data repository system is the system described in Clause 1, comprising: a data ingestion subsystem that receives data from multiple data sources; a data storage subsystem that stores data for the data repository system; a data processing subsystem that transforms data from multiple data sources based on an extract, transform, load (etl) pipeline to provide transformed data used for generating one or more machine learning models; a data analysis subsystem that provides access to data visualization tools and database analysis tools used for generating one or more machine learning models; and a data cloud infrastructure subsystem that provides one or more rules associated with control operations for the data repository system.

[0009] Clause 3: The data repository system is the system described in Clause 1 or 2, configured to receive initial data associated with one or more data records from multiple data sources, including at least one of an electronic medical record (EMR) system, a medical imaging system, a fluid injection system, a pathology information system, a laboratory information system, or any combination thereof, and to perform one or more data curation procedures on the initial data associated with one or more data records in order to provide formatted data associated with multiple medical procedures for the generation of one or more machine learning models.

[0010] Clause 4: The system described in any one of Clauses 1 to 3, wherein at least one processor is configured to train one or more machine learning models based on formatted data associated with multiple medical procedures for the generation of one or more machine learning models in order to provide one or more trained machine learning models, and to validate the output of one or more trained machine learning models based on data associated with regulatory guidance and approval processes.

[0011] Clause 5: The system described in any one of Clauses 1 to 4, further configured to receive a request for access to a particular machine learning model among a plurality of machine learning models, determine whether the request for access conforms to one or more criteria associated with access to the particular machine learning model, and provide access to the particular machine learning model based on the determination that the request for access conforms to one or more criteria associated with access to the particular machine learning model.

[0012] Clause 6: The system described in any one of Clauses 1 to 5, wherein formatted data associated with multiple medical procedures for the generation of one or more machine learning models comprises multiple datasets, each dataset associated with a specific medical procedure among the multiple medical procedures, and at least one processor is further configured to receive requests for access to a specific dataset among the multiple datasets, determine whether the request for access conforms to one or more criteria associated with access to the specific dataset, and, based on the determination that the request for access conforms to one or more criteria associated with access to the specific dataset, provide access to the specific dataset.

[0013] Clause 7: The system described in any one of Clauses 1 to 6, wherein, when providing access to a specific dataset, at least one processor is configured to make application programming interface (API) calls to the data repository system relating to a specific dataset in order to provide access to that specific dataset.

[0014] Clause 8: The system described in any one of Clauses 1 to 7, wherein formatted data associated with multiple medical procedures for the generation of one or more machine learning models comprises multiple datasets, each dataset of the multiple datasets is associated with a specific medical procedure among the multiple medical procedures, and at least one processor is further configured to train one or more machine learning models based on a first dataset of the multiple datasets in order to provide one or more trained machine learning models, assign a unique identifier to the first dataset based on having trained one or more machine learning models, assign a unique identifier to one or more trained machine learning models, validate the output of one or more trained machine learning models based on data associated with regulatory guidance and approval processes, and when validating the output of one or more trained machine learning models, at least one processor is configured to validate the first dataset and one or more trained machine learning models based on the unique identifiers assigned to the first dataset and one or more trained machine learning models.

[0015] Clause 9: The system described in any one of Clauses 1 to 8, wherein, when deploying one or more machine learning models to a production environment, at least one processor is configured to deploy one or more machine learning models to at least one of the following: a production environment of a third-party platform not associated with the AI ​​medical device platform, a production environment of the AI ​​medical device platform, a platform including standardized containers, or any combination thereof.

[0016] Clause 10: A method for generating an artificial intelligence (AI)-based automated healthcare application, comprising: using at least one processor of an AI medical device platform to receive formatted data associated with a plurality of medical procedures from a data repository system for the generation of one or more machine learning models; using at least one processor to receive data associated with one or more regulatory guidance and approval processes; using at least one processor to generate one or more machine learning models configured to predict one or more aspects associated with a medical procedure based on the formatted data associated with a plurality of medical procedures and the data associated with regulatory guidance and approval processes; and using at least one processor to place one or more machine learning models into a production environment based on the generation of one or more machine learning models.

[0017] Clause 11: The method according to Clause 10, further comprising the step of receiving initial data associated with one or more data records from a plurality of data sources, including at least one of an electronic medical record (EMR) system, a medical imaging system, a fluid injection system, a pathology information system, a laboratory information system, or any combination thereof.

[0018] Clause 12: The method according to Clause 10 or 11, further comprising the step of performing one or more data curation procedures on initial data associated with one or more data records in order to provide formatted data associated with multiple medical procedures for the generation of one or more machine learning models.

[0019] Clause 13: The method described in any one of Clauses 10 to 12, wherein the step of generating one or more machine learning models configured to predict one or more aspects associated with a medical procedure includes the steps of training one or more machine learning models on formatted data associated with a plurality of medical procedures in order to provide one or more trained machine learning models, and verifying the output of one or more trained machine learning models on data associated with regulatory guidance and approval processes.

[0020] Clause 14: The method described in any one of Clauses 10 to 13, further comprising the steps of receiving a request for access to a particular machine learning model among a plurality of machine learning models, determining whether the request for access conforms to one or more criteria associated with access to the particular machine learning model, and providing access to the particular machine learning model based on the determination that the request for access conforms to one or more criteria associated with access to the particular machine learning model.

[0021] Clause 15: Formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset of the multiple datasets is associated with a particular medical procedure among the multiple medical procedures, and the method further includes the steps of receiving a request for access to a particular dataset among the multiple datasets, determining whether the request for access conforms to one or more criteria associated with access to the particular dataset, and providing access to the particular dataset based on the determination that the request for access conforms to one or more criteria associated with access to the particular dataset.

[0022] Clause 16: The method described in any one of Clauses 10 to 15, wherein the step of providing access to a particular dataset includes the step of making an application programming interface (API) call to a data repository system related to a particular dataset in order to provide access to a particular dataset.

[0023] Clause 17: Formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset of the multiple datasets is associated with a specific medical procedure among the multiple medical procedures, and the method further includes the steps of training one or more machine learning models on a first dataset of the multiple datasets to provide one or more trained machine learning models, assigning a unique identifier to the first dataset on the basis that one or more machine learning models have been trained, assigning a unique identifier to one or more trained machine learning models, and validating the output of one or more trained machine learning models on data associated with regulatory guidance and approval processes, wherein the step of validating the output of one or more trained machine learning models includes the step of validating the first dataset and one or more trained machine learning models on the first dataset and one or more trained machine learning models on the unique identifiers assigned to the first dataset and one or more trained machine learning models, the method according to any one of Clauses 10 to 16.

[0024] Clause 18: The method described in any one of Clauses 10 to 17, wherein the step of deploying one or more machine learning models to a production environment includes deploying one or more machine learning models to at least one of the following: a production environment of a third-party platform not associated with the AI ​​medical device platform, a production environment of the AI ​​medical device platform, a platform including standardized containers, or any combination thereof.

[0025] Clause 19: A computer program product for generating an artificial intelligence (AI)-based automated healthcare application, comprising at least one non-transitory computer-readable medium containing program instructions, which, when executed by at least one processor, cause the at least one processor to receive formatted data associated with a plurality of medical procedures for generating one or more machine learning models from a data repository system, receive data associated with one or more regulatory guidance and approval processes from the data repository system, generate one or more machine learning models configured to predict one or more aspects associated with a medical procedure based on the data associated with the plurality of medical procedures and the data associated with the regulatory guidance and approval processes, and place the one or more machine learning models in a production environment based on having generated the one or more machine learning models.

[0026] Clause 20: The computer program product according to Clause 19, wherein the program instructions further cause the at least one processor to receive initial data associated with one or more data records from a plurality of data sources including at least one of an electronic medical record (EMR) system, a medical imaging system, a fluid injection system, a pathology information system, a laboratory information system, or any combination thereof.

[0027] Clause 21: The computer program product according to Clause 19 or 20, wherein the program instructions further cause the at least one processor to execute one or more data curation procedures on the initial data associated with one or more data records to provide formatted data associated with a plurality of medical procedures for generating one or more machine learning models.

[0028] Clause 22: A computer program product according to any one of Clauses 19 to 21, wherein program instructions that cause at least one processor to generate one or more machine learning models configured to predict one or more aspects associated with a medical treatment cause the at least one processor to train one or more machine learning models based on formatted data associated with a plurality of medical treatments for generating the one or more machine learning models, and to verify the output of the one or more trained machine learning models based on data associated with a regulatory guidance and an approval process.

[0029] Clause 23: A computer program product according to any one of Clauses 19 to 22, wherein the program instructions further cause the at least one processor to receive an access request to a particular machine learning model among a plurality of machine learning models, determine whether the access request complies with one or more criteria associated with access to the particular machine learning model, and provide access to the particular machine learning model based on a determination that the access request complies with the one or more criteria associated with access to the particular machine learning model.

[0030] Clause 24: A computer program product according to any one of Clauses 19 to 23, wherein the formatted data associated with a plurality of medical treatments for generating one or more machine learning models includes a plurality of data sets, each data set of the plurality of data sets being associated with a particular medical treatment among the plurality of medical treatments, and the program instructions further cause the at least one processor to receive an access request to a particular data set among the plurality of data sets, determine whether the access request complies with one or more criteria associated with access to the particular data set, and provide access to the particular data set based on a determination that the access request complies with the one or more criteria associated with access to the particular data set.

[0031] Clause 25: A computer program product as described in any one of Clauses 19 to 24, which causes at least one processor to provide access to a particular dataset by executing an application programming interface (API) call to a data repository system relating to a particular dataset in order to provide access to that particular dataset.

[0032] 26: A computer program product as described in any one of Clauses 19 to 25, wherein formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset of the multiple datasets is associated with a specific medical procedure among the multiple medical procedures, and the program instructions further cause at least one processor to train one or more machine learning models based on a first dataset of the multiple datasets in order to provide one or more trained machine learning models, assign a unique identifier to the first dataset based on having trained one or more machine learning models, assign a unique identifier to one or more trained machine learning models, validate the output of one or more trained machine learning models based on data associated with regulatory guidance and approval processes, and validate the output of one or more trained machine learning models, thereby causing at least one processor to validate the first dataset and one or more trained machine learning models based on unique identifiers assigned to the first dataset and one or more trained machine learning models.

[0033] Clause 27: A computer program product as described in any one of Clauses 19 to 26, which causes at least one processor to deploy one or more machine learning models to a production environment of a third-party platform not associated with an AI medical device platform, a production environment of an AI medical device platform, a platform including standardized containers, or any combination thereof.

[0034] Clause 28: A system for generating artificial intelligence (AI) based automated healthcare applications, comprising: a data repository system storing at least one of formatted data associated with a plurality of medical procedures for generating one or more non-machine learning automated healthcare applications, data associated with regulatory guidance and approval processes, or any combination thereof; and an AI medical device platform having at least one processor communicating with the data repository system, wherein the at least one processor is configured to receive formatted data associated with a plurality of medical procedures for generating one or more non-machine learning automated healthcare applications, receive data associated with one or more regulatory guidance and approval processes, generate one or more non-machine learning automated healthcare applications configured to predict one or more aspects associated with a medical procedure based on the formatted data associated with the plurality of medical procedures and the data associated with regulatory guidance and approval processes, and deploy one or more non-machine learning automated healthcare applications to a production environment based on the generation of one or more non-machine learning automated healthcare applications.

[0035] Clause 29: The data repository system is the system described in Clause 28, comprising: a data ingestion subsystem that receives data from multiple data sources; a data storage subsystem that stores data for the data repository system; a data processing subsystem that transforms data from multiple data sources based on an extract, transform, load (ETL) pipeline to provide transformed data used for generating one or more non-machine learning automated healthcare applications; a data analysis subsystem that provides access to data visualization tools and database analysis tools used for generating one or more non-machine learning automated healthcare applications; and a data cloud infrastructure subsystem that provides one or more rules associated with control operations for the data repository system.

[0036] Clause 30: The data repository system is configured to receive initial data associated with one or more data records from multiple data sources, including at least one of an electronic medical record (EMR) system, a medical imaging system, a fluid injection system, a pathology information system, a laboratory information system, or any combination thereof, and to perform one or more data curation procedures on the initial data associated with one or more data records in order to provide formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications, as described in Clause 28 or 29.

[0037] Clause 31: A system according to any one of Clauses 28 to 30, wherein at least one processor is configured to generate one or more non-machine learning automated healthcare applications configured to predict one or more aspects associated with a medical procedure, to provide one or more generated non-machine learning automated healthcare applications based on formatted data associated with a plurality of medical procedures for the generation of one or more non-machine learning automated healthcare applications, and to verify the output of one or more generated non-machine learning automated healthcare applications based on data associated with regulatory guidance and approval processes.

[0038] Clause 32: The system described in any one of Clauses 28 to 31, further configured to receive requests for access to a particular machine learning model among a plurality of non-machine learning automated healthcare applications, determine whether the request for access conforms to one or more criteria associated with access to the particular machine learning model, and provide access to the particular machine learning model based on the determination that the request for access conforms to one or more criteria associated with access to the particular machine learning model.

[0039] Clause 33: Formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications, comprising multiple datasets, each dataset associated with a specific medical procedure among the multiple medical procedures, and further configured to receive requests for access to a specific dataset among the multiple datasets, determine whether the request for access conforms to one or more criteria associated with access to the specific dataset, and provide access to the specific dataset based on the determination that the request for access conforms to one or more criteria associated with access to the specific dataset, as described in any one of Clauses 28 to 32.

[0040] Clause 34: The system described in any one of Clauses 28 to 33, wherein, when providing access to a specific dataset, at least one processor is configured to make application programming interface (API) calls to the data repository system relating to the specific dataset in order to provide access to the specific dataset.

[0041] Clause 35: The System as described in any one of Clauses 28 to 34, wherein formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications includes multiple datasets, each dataset of the multiple datasets is associated with a specific medical procedure among the multiple medical procedures, and at least one processor is configured to generate one or more non-machine learning automated healthcare applications based on a first dataset among the multiple datasets in order to provide one or more generated non-machine learning automated healthcare applications, assign a unique identifier to the first dataset based on the fact that one or more non-machine learning automated healthcare applications have been generated, assign a unique identifier to one or more generated non-machine learning automated healthcare applications, and further configure to validate the output of one or more generated non-machine learning automated healthcare applications based on data associated with regulatory guidance and approval processes, and when validating the output of one or more generated non-machine learning automated healthcare applications, at least one processor is configured to validate the first dataset and one or more generated non-machine learning automated healthcare applications based on the unique identifiers assigned to the first dataset and one or more generated non-machine learning automated healthcare applications.

[0042] Clause 36: The System described in any one of Clauses 28 to 35, wherein at least one processor is configured to deploy one or more non-machine learning automated healthcare applications to a production environment of a third-party platform not associated with the AI ​​medical device platform, a production environment of the AI ​​medical device platform, a platform including standardized containers, or any combination thereof.

[0043] Clause 37: A method for generating an artificial intelligence (AI)-based automated healthcare application, comprising: using at least one processor of an AI medical device platform to receive formatted data associated with a plurality of medical procedures for the generation of one or more non-machine learning automated healthcare applications from a data repository system; using at least one processor to receive data associated with one or more regulatory guidance and approval processes; using at least one processor to generate one or more non-machine learning automated healthcare applications configured to predict one or more aspects associated with a medical procedure based on the formatted data associated with a plurality of medical procedures for the generation of one or more non-machine learning automated healthcare applications and the data associated with regulatory guidance and approval processes; and using at least one processor to deploy one or more non-machine learning automated healthcare applications to a production environment based on the generation of one or more non-machine learning automated healthcare applications.

[0044] Clause 38: The method according to Clause 37, further comprising the step of receiving initial data associated with one or more data records from a plurality of data sources, including at least one of an electronic medical record (EMR) system, a medical imaging system, a fluid injection system, a pathology information system, a laboratory information system, or any combination thereof.

[0045] Clause 39: The method according to Clause 37 or 38, further comprising the step of performing one or more data curation procedures on initial data associated with one or more data records in order to provide formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications.

[0046] Clause 40: The method described in any one of Clauses 37 to 39, wherein the step of generating one or more non-machine learning automated healthcare applications configured to predict one or more aspects associated with a medical procedure includes the steps of generating one or more non-machine learning automated healthcare applications based on formatted data associated with a plurality of medical procedures in order to provide one or more generated non-machine learning automated healthcare applications, and verifying the output of one or more generated non-machine learning automated healthcare applications based on data associated with regulatory guidance and approval processes.

[0047] Clause 41: The method described in any one of Clauses 37 to 40, further comprising the steps of receiving a request for access to a particular machine learning model among a group of non-machine learning automated healthcare applications, determining whether the request for access conforms to one or more criteria associated with access to the particular machine learning model, and providing access to the particular machine learning model based on the determination that the request for access conforms to one or more criteria associated with access to the particular machine learning model.

[0048] Clause 42: Formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications includes multiple datasets, each dataset of the multiple datasets is associated with a specific medical procedure among the multiple medical procedures, and the method further includes the steps of receiving a request for access to a specific dataset among the multiple datasets, determining whether the request for access conforms to one or more criteria associated with access to the specific dataset, and providing access to the specific dataset based on the determination that the request for access conforms to one or more criteria associated with access to the specific dataset.

[0049] Clause 43: The method described in any one of Clauses 37 to 42, wherein the step of providing access to a particular dataset includes the step of making an application programming interface (API) call to a data repository system relating to a particular dataset in order to provide access to the particular dataset.

[0050] Clause 44: Formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications includes multiple datasets, each dataset of the multiple datasets is associated with a specific medical procedure of the multiple medical procedures, and the method further includes the steps of generating one or more non-machine learning automated healthcare applications based on a first dataset of the multiple datasets in order to provide one or more generated non-machine learning automated healthcare applications; assigning a unique identifier to the first dataset based on the generation of one or more non-machine learning automated healthcare applications; assigning a unique identifier to one or more generated non-machine learning automated healthcare applications; and validating the output of one or more generated non-machine learning automated healthcare applications based on data associated with regulatory guidance and approval processes, wherein the step of validating the output of one or more generated non-machine learning automated healthcare applications includes validating the first dataset and one or more generated non-machine learning automated healthcare applications based on unique identifiers assigned to the first dataset and one or more generated non-machine learning automated healthcare applications, the method according to any one of Clauses 37 to 43.

[0051] Clause 45: The method described in any one of Clauses 37 to 44, wherein the step of deploying one or more non-machine learning automated healthcare applications to a production environment includes deploying one or more non-machine learning automated healthcare applications to at least one of the following: a production environment of a third-party platform not associated with the AI ​​medical device platform, a production environment of the AI ​​medical device platform, a platform including standardized containers, or any combination thereof.

[0052] Clause 46: A computer program product for generating artificial intelligence (AI)-based automated healthcare applications, comprising at least one non-temporary computer-readable medium including program instructions, wherein, when executed by at least one processor, the program instructions cause at least one processor to receive formatted data associated with a plurality of medical procedures for generating one or more non-machine learning automated healthcare applications from a data repository system, receive data associated with one or more regulatory guidance and approval processes from a data repository system, generate one or more non-machine learning automated healthcare applications configured to predict one or more aspects associated with medical procedures based on the formatted data associated with the plurality of medical procedures and the data associated with regulatory guidance and approval processes, and, based on the generation of one or more non-machine learning automated healthcare applications, deploy one or more non-machine learning automated healthcare applications to a production environment.

[0053] Clause 47: The computer program product described in Clause 46, wherein the program instructions further cause at least one processor to receive initial data associated with one or more data records from a plurality of data sources, including at least one of an electronic medical record (EMR) system, a medical imaging system, a fluid injection system, a pathology information system, a laboratory information system, or any combination thereof.

[0054] Clause 48: The computer program product described in Clause 46 or 47, wherein the program instructions further cause at least one processor to perform one or more data curation procedures on initial data associated with one or more data records in order to provide formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications.

[0055] Clause 49: A computer program product as described in any one of Clauses 46 to 48, which causes at least one processor to generate one or more non-machine learning automated healthcare applications configured to predict one or more aspects associated with a medical procedure, which causes at least one processor to generate one or more non-machine learning automated healthcare applications based on formatted data associated with a plurality of medical procedures for the generation of one or more non-machine learning automated healthcare applications, in order to provide one or more generated non-machine learning automated healthcare applications, and to verify the output of one or more generated non-machine learning automated healthcare applications based on data associated with regulatory guidance and approval processes.

[0056] Clause 50: A computer program product as described in any one of Clauses 46 to 49, wherein the program instructions cause at least one processor to receive a request for access to a particular machine learning model among a plurality of non-machine learning automated healthcare applications, to determine whether the request for access conforms to one or more criteria associated with access to the particular machine learning model, and to provide access to the particular machine learning model based on the determination that the request for access conforms to one or more criteria associated with access to the particular machine learning model.

[0057] Clause 51: A computer program product as described in any one of Clauses 46 to 50, wherein formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications includes multiple datasets, each dataset being associated with a specific medical procedure among the multiple medical procedures, and the program instructions further cause at least one processor to receive a request for access to a specific dataset among the multiple datasets, to determine whether the request for access conforms to one or more criteria associated with access to the specific dataset, and, based on the determination that the request for access conforms to one or more criteria associated with access to the specific dataset, to provide access to the specific dataset.

[0058] Clause 52: A computer program product as described in any one of Clauses 46 to 51, which causes at least one processor to provide access to a particular dataset by executing an application programming interface (API) call to a data repository system relating to a particular dataset in order to provide access to that particular dataset.

[0059] 53: A computer program product as described in any one of Clauses 46 to 52, wherein formatted data associated with multiple medical procedures for the generation of one or more non-machine learning automated healthcare applications includes multiple datasets, each dataset of the multiple datasets is associated with a specific medical procedure among the multiple medical procedures, and the program instructions further cause at least one processor to generate one or more non-machine learning automated healthcare applications based on a first dataset among the multiple datasets in order to provide one or more generated non-machine learning automated healthcare applications, assign a unique identifier to the first dataset based on the generation of one or more non-machine learning automated healthcare applications, assign a unique identifier to one or more generated non-machine learning automated healthcare applications, validate the output of one or more generated non-machine learning automated healthcare applications based on data associated with regulatory guidance and approval processes, and the program instructions cause at least one processor to validate the output of one or more generated non-machine learning automated healthcare applications based on a unique identifier assigned to the first dataset and one or more generated non-machine learning automated healthcare applications.

[0060] Clause 54: A computer program product as described in any one of Clauses 46 to 53, which causes at least one processor to deploy one or more non-machine learning automated healthcare applications to a production environment, which includes at least one of the following: a production environment of a third-party platform not associated with an AI medical device platform, a production environment of an AI medical device platform, a platform including standardized containers, or any combination thereof.

[0061] These and other features and characteristics of this disclosure, as well as the operation and function of the relevant elements of the structure, and the economics of the assembly and manufacture of the parts, will become more apparent upon consideration of the following description and the appended claims with reference to the appended drawings, all of which form part of this specification, and similar reference numbers indicate corresponding parts of various figures. However, it should be clearly understood that the drawings are for illustrative and illustrative purposes only and are not intended as definitions of the limitations of this disclosure. Where used herein and in the claims, the singular “a,” “an,” and “the” refer to plural subjects unless otherwise explicitly indicated in the context.

[0062] Further advantages and details of non-limiting embodiments are described in more detail below with reference to the exemplary embodiments shown in the attached schematic diagrams. [Brief explanation of the drawing]

[0063] [Figure 1A] This figure shows non-limiting embodiments of environments in which the systems, methods, and / or computer program products described herein may be implemented in accordance with this disclosure. [Figure 1B] This is a diagram of a non-limiting embodiment of a system for generating artificial intelligence (AI)-based automated healthcare applications. [Figure 2]Figures 1A and 1B are diagrams of non-limiting embodiments of one or more devices and / or components of one or more systems. [Figure 3] This is a flowchart of a non-limiting embodiment of a method for generating AI-based automated healthcare applications using an AI medical device platform. [Figure 4] This is a diagram illustrating a non-limiting embodiment of an AI medical device platform. [Figure 5] This is a diagram of a non-limiting embodiment of a data repository system. [Figure 6A] This is a diagram of a non-limiting embodiment of an implementation of a method for generating AI-based automated healthcare applications using an AI medical device platform. [Figure 6B] This is a diagram of a non-limiting embodiment of an implementation of a method for generating AI-based automated healthcare applications using an AI medical device platform. [Figure 6C] This is a diagram of a non-limiting embodiment of an implementation of a method for generating AI-based automated healthcare applications using an AI medical device platform. [Figure 6D] This is a diagram of a non-limiting embodiment of an implementation of a method for generating AI-based automated healthcare applications using an AI medical device platform. [Modes for carrying out the invention]

[0064] For the purposes of the following description, the terms “end,” “top,” “bottom,” “right,” “left,” “vertical,” “horizontal,” “upper,” “lower,” “lateral,” “longitudinal,” and their derivatives shall be used in reference to the present disclosure as it is oriented in the drawings. However, it should be understood that the present disclosure may envision a variety of alternative variations and sequences of steps unless expressly otherwise specified. It should also be understood that certain devices and methods shown in the accompanying drawings and described in the following specification are merely exemplary embodiments of the present disclosure. Accordingly, certain dimensions and other physical characteristics relating to embodiments or aspects disclosed herein should not be considered limiting unless otherwise indicated.

[0065] The aspects, components, elements, structures, actions, steps, functions, instructions, etc., used herein should not be construed as important or essential unless expressly stated otherwise. Furthermore, as used herein, the articles “a” and “an” are intended to include one or more items and may be used interchangeably with “one or more” and “at least one.” Additionally, as used herein, the term “set” is intended to include one or more items (e.g., related items, unrelated items, combinations of related and unrelated items) and may be used interchangeably with “one or more” or “at least one.” When only one item is intended, the term “one” or a similar term is used. Furthermore, as used herein, terms such as “has,” “have,” and “having” are intended to be open-ended terms. Additionally, the phrase “based on” is intended to mean “at least partially based” unless otherwise specified. Moreover, a reference to an action “based on” a condition may also refer to an action “in response to” a condition. For example, the phrases “based on” and “in response to” may, in some non-limiting embodiments or aspects, refer to conditions for automatically triggering an action (e.g., a specific action of an electronic device such as a computing device or processor).

[0066] As used herein, the terms “communicate” and “communicate” may mean the receiving, acceptance, transmission, transfer, provision, etc., of information (e.g., data, signals, messages, instructions, commands, etc.). Communication between one unit (e.g., a device, a system, a component of a device or system, a combination thereof) and another unit means that one unit can receive information directly or indirectly from and / or transmit information to the other unit. This may mean a direct or indirect connection that is essentially wired and / or wireless. Furthermore, two units can communicate with each other even if the transmitted information is modified, processed, relayed, and / or routed between the first and second units. For example, the first unit can communicate with the second unit even if the first unit passively receives information and does not actively transmit information to the second unit. As another example, the first unit can communicate with the second unit if at least one intermediate unit (for example, a third unit located between the first and second units) processes information received from the first unit and transmits the processed information to the second unit. In some non-limiting embodiments, the information may refer to network packets containing data (for example, data packets).

[0067] Some non-limiting embodiments or aspects may be described herein in relation to thresholds. As used herein, satisfying a threshold may mean values ​​such as greater than the threshold, more than the threshold, higher than the threshold, greater than or equal to the threshold, less than the threshold, less than the threshold, lower than the threshold, less than or equal to the threshold, and equal to the threshold.

[0068] Systems, methods, and computer program products that provide solutions to the above-mentioned problems are disclosed. For example, a system for generating artificial intelligence (AI)-based automated healthcare applications, as disclosed herein, may include a data repository system and an AI medical device platform. In some non-limiting embodiments, the data repository system may store at least one of formatted data associated with a plurality of medical procedures for generating one or more machine learning models, data associated with regulatory guidance and approval processes, or any combination thereof. In some non-limiting embodiments, the AI ​​medical device platform may include at least one processor communicating with the data repository system, the at least one processor being configured to receive formatted data associated with a plurality of medical procedures for generating one or more machine learning models, receive data associated with one or more regulatory guidance and approval processes, generate one or more machine learning models configured to predict one or more aspects associated with a medical procedure based on the data associated with the plurality of medical procedures and the data associated with regulatory guidance and approval processes, and, based on having generated one or more machine learning models, deploy one or more machine learning models to a production environment.

[0069] In some non-limiting embodiments, the data repository system includes a data ingestion subsystem that receives data from multiple data sources; a data storage subsystem that stores data for the data repository system; a data processing subsystem that transforms data from multiple data sources based on an extract, transform, load (ETL) pipeline to provide transformed data used for generating one or more machine learning models; a data analysis subsystem that provides access to data visualization tools and database analysis tools used for generating one or more machine learning models; and a data cloud infrastructure subsystem that provides one or more rules associated with control operations for the data repository system.

[0070] In some non-limiting embodiments, the data repository system may be configured to receive initial data associated with one or more data records from multiple data sources, including at least one of an electronic medical record (EMR) system, a medical imaging system, a fluid injection system, a pathology information system, a laboratory information system, or any combination thereof, and to perform one or more data curation steps on the initial data associated with one or more data records in order to provide formatted data associated with multiple medical procedures for the generation of one or more machine learning models.

[0071] In some non-limiting embodiments, when generating one or more machine learning models configured to predict one or more aspects associated with a medical procedure, at least one processor may be configured to train one or more machine learning models based on formatted data associated with multiple medical procedures for the generation of one or more machine learning models, in order to provide one or more trained machine learning models, and to validate the output of one or more trained machine learning models based on data associated with regulatory guidance and approval processes.

[0072] In some non-limiting embodiments, at least one processor may be further configured to receive a request to access a particular machine learning model among a plurality of machine learning models, determine whether the request conforms to one or more criteria associated with access to the particular machine learning model, and provide access to the particular machine learning model based on the determination that the request conforms to one or more criteria associated with access to the particular machine learning model. In some non-limiting embodiments, the formatted data associated with a plurality of medical procedures for generating one or more machine learning models may include a plurality of datasets, each dataset of the plurality of datasets may be associated with a particular medical procedure among the plurality of medical procedures, and at least one processor may be further configured to receive a request to access a particular dataset among the plurality of datasets, determine whether the request conforms to one or more criteria associated with access to the particular dataset, and provide access to the particular dataset based on the determination that the request conforms to one or more criteria associated with access to the particular dataset.

[0073] In some non-limiting embodiments, when providing access to a particular dataset, at least one processor may be configured to make application programming interface (API) calls to the data repository system related to that particular dataset in order to provide access to that particular dataset.

[0074] In some non-limiting embodiments, the formatted data associated with a plurality of medical procedures for generating one or more machine learning models may include a plurality of datasets, each dataset being associated with a specific medical procedure among the plurality of medical procedures, and at least one processor may be further configured to train one or more machine learning models based on a first dataset among the plurality of datasets in order to provide one or more trained machine learning models, to assign a unique identifier to the first dataset based on having trained one or more machine learning models, to assign a unique identifier to one or more trained machine learning models, and to validate the output of one or more trained machine learning models based on data associated with regulatory guidance and approval processes. In some non-limiting embodiments, when validating the output of one or more trained machine learning models, at least one processor may be configured to validate the first dataset and one or more trained machine learning models based on the unique identifiers assigned to the first dataset and one or more trained machine learning models.

[0075] In some non-limiting embodiments, when deploying one or more machine learning models to a production environment, at least one processor may be configured to deploy one or more machine learning models to at least one of the following: a production environment of a third-party platform not associated with the AI ​​medical device platform, a production environment of the AI ​​medical device platform, a platform including standardized containers, or any combination thereof.

[0076] In this manner, the Disclosure provides a way to efficiently develop and / or deploy automated healthcare applications, including AI-based SaMDs (e.g., automated healthcare applications comprising one or more machine learning models). Furthermore, the Disclosure provides an AI medical device platform and data repository system that enables multiple users (e.g., unrelated users) to access large amounts of data related to multiple medical procedures, as well as data associated with one or more regulatory guidance and approval processes for the generation of automated healthcare applications, and to determine whether an automated healthcare application meets regulatory criteria for regulatory guidance and approval before deployment of the automated healthcare application.

[0077] Referring here to Figure 1A, Figure 1A is a diagram of a non-limiting embodiment of an environment 100A in which the devices, systems, methods, and / or computer program products described herein may be implemented. As shown in Figure 1A, the environment 100A includes an artificial intelligence (AI) medical device platform 102, data sources 104-1 to 104-N (hereinafter referred to individually as data source 104 or collectively as data source 104, as appropriate), user devices 106-1 to 106-N (hereinafter referred to individually as user device 106 or collectively as user device 106, as appropriate), and a data repository system 108.

[0078] The AI ​​medical device platform 102 may interconnect with the data source 104, user device 106, and / or data repository system 108 via a communication network 110 (e.g., by establishing a connection for communication). In some non-limiting embodiments, the AI ​​medical device platform 102 may interconnect with the data source 104, user device 106, and / or data repository system 108 via a wired connection, a wireless connection, or a combination of wired and wireless connections (e.g., by establishing a connection for communication).

[0079] In some non-limiting embodiments, the AI ​​medical device platform 102 may include one or more devices capable of communicating with a data source 104, a user device 106, and / or a data repository system 108 via a communication network 110. For example, the AI ​​medical device platform 102 may include a group of servers, such as a server, a cloud-based solution including multiple servers (e.g., a private cloud solution, a public cloud, a hybrid cloud, a multi-cloud, etc.), and / or other similar devices. Additionally or alternatively, the AI ​​medical device platform 102 may include computing devices such as a desktop computer, a mobile device (e.g., a tablet, a smartphone, a wearable device, etc.). In some non-limiting embodiments, the AI ​​medical device platform 102 may include one or more applications (e.g., software applications) that perform a set of functions on an external application programming interface (API) to send data to an external system such as a user device 106 or a data repository system 108 associated with an external API, and to receive data from an external system associated with an external API. In some non-limiting embodiments, the AI ​​medical device platform 102 may include one or more subsystems. For example, the AI ​​medical device platform 102 may include one or more subsystems related to the generation of one or more machine learning models (e.g., one or more automated healthcare applications including one or more machine learning models). In some non-limiting embodiments, the AI ​​medical device platform 102 may include a data repository system 108.

[0080] In some non-limiting embodiments, the AI ​​medical device platform 102 may generate (e.g., train, validate, retrain, etc.), store, and / or implement (e.g., operate, provide input, and / or provide output, etc.) one or more machine learning models. For example, the AI ​​medical device platform 102 may generate one or more machine learning models by fitting (e.g., validate, test, etc.) one or more machine learning models to data used for training (e.g., training data). In some non-limiting embodiments or aspects, the AI ​​medical device platform 102 may generate, store, and / or implement one or more machine learning models to be provided to a production environment (e.g., runtime environment, real-time environment, environment where software applications and / or services are deployed and made available to end users, etc.) used to provide inference (e.g., secure inference) based on data input in live conditions in the production environment (e.g., real-time conditions). In some non-limiting embodiments, the production environment may include a target information technology (IT) setup including hardware and / or software running on the hardware for reliability and redundancy. Additionally or alternatively, the AI ​​medical device platform 102 may generate, store, and / or implement one or more machine learning models provided for non-production environments (e.g., offline environments, training environments, etc.) used to provide inference based on data input in non-live situations. In some non-limiting embodiments or aspects, the AI ​​medical device platform 102 may communicate with a data storage device (data repository system 108) which may be local or remote to the AI ​​medical device platform 102.

[0081] In some non-limiting embodiments, the AI ​​medical device platform 102 may provide platform-level services, including services for isolated software instances, user interfaces for interaction, market-related services for all aspects of automated healthcare applications, services associated with contracts for the use of the AI ​​medical device platform 102, services for consumption monitoring and / or billing associated with the use of the AI ​​medical device platform 102, services for onboarding users of the AI ​​medical device platform 102, procurement services related to the AI ​​medical device platform 102, and / or user management services related to the AI ​​medical device platform 102 (e.g., authentication, authorization, access levels, etc.).

[0082] In some non-limiting embodiments, the data source 104 may include one or more devices that can communicate with the AI ​​medical device platform 102, user devices 106, and data repository system 108 via a communication network 110 and perform fluid infusion procedures. For example, the data source 104 may include computing devices such as servers, desktop computers, and mobile devices (e.g., tablets, smartphones, wearables such as wearable health sensors). In some non-limiting embodiments, the data source 104 may include communication devices associated with hospital information systems, image archiving and communication systems (PACS), EMR systems, medical imaging systems (e.g., imaging scanners), fluid infusion systems (e.g., fluid injectors), medical devices (e.g., handheld medical devices, wearable medical devices such as portable health sensors), devices associated with facilities such as fluid infusion systems, pathology information systems, and laboratory information systems, and / or devices associated with patients (e.g., user devices such as computing devices operated by patients). Additionally or alternatively, data source 104 may include reporting systems (e.g., systems that generate medical reports based on healthcare data), devices associated with facilities (e.g., hospitals), and / or devices associated with patients.

[0083] In some non-limiting embodiments, the user device 106 may include one or more devices that can communicate with the AI ​​medical device platform 102, the data source 104, and / or the data repository system 108 via the communication network 110. For example, the user device 106 may include computing devices such as a server, a group of servers, a desktop computer, or a mobile device (e.g., a tablet, smartphone, wearable device, etc.). In some non-limiting embodiments or aspects, the user device 106 may be associated with a user (e.g., an individual operating the user device 106).

[0084] In some non-limiting embodiments, the data repository system 108 may include one or more devices that can communicate with the AI ​​medical device platform 102, the data source 104, and / or the user device 106 via the communication network 110. For example, the data repository system 108 may include a server, a group of servers, a cloud platform, etc. In some non-limiting embodiments, the data repository system 108 may include a data lake for storing, processing, and / or protecting large amounts of data, such as data records (e.g., a patient's electronic medical record, a patient's electronic health record, etc.). In some examples, the data repository system 108 may receive, store, process, and / or protect multiple data records, such as at least 1,000 data records, 10,000 data records, 100 million data records, 1 billion data records, or 1 trillion data records. In another example, the data repository system 108 may include one or more machine learning models, each stored on a device (e.g., a server).

[0085] In some non-limiting embodiments, the data repository system 108 may include multiple subsystems. In some non-limiting embodiments, one or more subsystems of the data repository system 108 may perform a set of functions on an external API to send data to an external system, such as the AI ​​medical device platform 102 or another data repository associated with the external API, and / or receive data from the AI ​​medical device platform 102 and / or other data repositories associated with the external API.

[0086] In some non-limiting embodiments, the communication network 110 may include one or more wired and / or wireless networks. For example, the communication network 110 may include cellular networks (e.g., Long-Term Evolution (LTE) networks, 3G networks, 4G networks, 5G networks, Code Division Multiple Access (CDMA) networks, etc.), local area networks (LANs), wide area networks (WANs), wireless LANs (WLANs), private networks, ad-hoc networks, intranets, the Internet, fiber optic-based networks, Ethernet networks, Universal Serial Bus (USB) networks, cloud computing networks, and / or some or all of these other types of networks.

[0087] The number and arrangement of systems and / or devices shown in Figure 1A are provided as an example. Additional systems and / or devices, fewer systems and / or devices, different systems and / or devices, or systems and / or devices in different arrangements than those shown in Figure 1A may exist. Furthermore, two or more systems and / or devices shown in Figure 1A may be implemented within a single system or single device, or a single system or single device shown in Figure 1A may be implemented as multiple distributed systems or devices. Additionally or alternatively, a set of systems or devices in environment 100A (e.g., one or more systems, one or more devices) may perform one or more functions that are described as being performed by another set of systems or another set of devices in environment 100A.

[0088] As shown in Figure 1B, System 100B includes an AI medical device platform 102 which includes a data repository system 108, a fluid injection system 112, a workstation device 114 including a display unit 114A, a medical imaging system 116, a digital pathology system 118, a laboratory information system 120, an electronic health record (EHR) system 122, an electronic medical record (EMR) system 124, and a hospital information system 126. In some non-limiting embodiments, the AI ​​medical device platform 102 (e.g., and the data repository system 108) may interconnect the fluid injection system 112, the workstation device 114, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the EHR system 122, the EMR system 124, and the hospital information system 126 via wired connections, wireless connections, or a combination of wired and wireless connections (e.g., by establishing connections for communication). In some non-limiting embodiments, the fluid injection system 112, the workstation device 114 including the display unit 114A, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the electronic health record (EHR) system 122, the electronic medical record (EMR) system 124, and / or the hospital information system 126 may be the same as or similar to the data source 104.

[0089] In some non-limiting embodiments, the fluid injection system 112 may include one or more devices capable of communicating with the AI ​​medical device platform 102, the workstation device 114, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the EHR system 122, the EMR system 124, and / or the hospital information system 126 via a communication network (e.g., communication network 110). For example, the fluid injection system 112 may include one or more computing devices such as one or more computers, one or more servers (e.g., cloud servers, groups of servers, etc.), one or more desktop computers, and one or more mobile devices (e.g., one or more tablets, one or more smartphones, etc.). In some non-limiting embodiments, the fluid injection system 112 may include one or more injection devices (e.g., one or more fluid injection devices, one or more fluid injectors). In some non-limiting embodiments, the fluid infusion system 112 is configured to administer (e.g., by injection, delivery, etc.) a contrast fluid containing a contrast agent to the patient, and / or to administer an aqueous fluid, such as saline, to the patient before, during, and / or after administration of the contrast fluid. For example, the fluid infusion system 112 can inject one or more prescribed doses of the contrast fluid directly into the patient's bloodstream via a subcutaneous needle and syringe. In some non-limiting embodiments, the fluid infusion system 112 may be configured to continuously administer the aqueous fluid to the patient through a peripheral venous line (PIV) and catheter, or one or more prescribed doses of the contrast fluid may be introduced into the PIV and administered to the patient via catheter. In some non-limiting embodiments, the fluid infusion system 112 is configured to inject a certain dose of the contrast fluid, followed by the administration of a specific volume of the aqueous fluid.In some non-limiting embodiments, the fluid injection system 112 is U.S. Patent Application No. 09 / 715,330, filed on November 17, 2000 and issued as U.S. Patent No. 6,643,537; U.S. Patent Application No. 09 / 982,518, filed on October 18, 2001 and issued as U.S. Patent No. 7,094,216; U.S. Patent Application No. 10 / 825,866, filed on April 16, 2004 and issued as U.S. Patent No. 7,556,619; and U.S. Patent Application No. 10 / 825,866, filed on May 7, 2009. This may include one or more fluid injection devices disclosed in U.S. Patent Application No. 12 / 437,011, issued as U.S. Patent No. 8,337,456, U.S. Patent Application No. 12 / 476,513, filed on 2 June 2009 and issued as U.S. Patent No. 8,147,464, and U.S. Patent Application No. 11 / 004,670, filed on 3 December 2004 and issued as U.S. Patent No. 8,540,698, each of which disclosures are incorporated herein by reference in their entirety. In some non-limiting embodiments, the fluid injection system 112 may include the MEDRAD® Stellant CT Injection System, MEDRAD® Stellant FLEX CT Injection System, MEDRAD® MRXperion MR Injection System, MEDRAD® Mark 7 Arterion Injection System, MEDRAD® Intego PET Infusion System, or MEDRAD® Centargo CT Injection System, all of which are supplied by Bayer Healthcare LLC.

[0090] In some non-limiting embodiments, the workstation device 114 includes one or more devices capable of communicating with the AI ​​medical device platform 102, the fluid injection system 112, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the EHR system 122, the EMR system 124, and / or the hospital information system 126 via a communication network (e.g., communication network 110). For example, the workstation device 114 may include one or more computing devices such as computers, including desktop computers, laptops, tablets, etc. In some non-limiting embodiments, the workstation device 114 may provide a control interface for controlling the operation of the fluid injection system 112, including providing input to the fluid injection system 112. Additionally or alternatively, the workstation device 114 may display operating parameters of the fluid injection system 112 while the fluid injection system 112 is operating (e.g., during real-time operation). In some non-limiting embodiments, the workstation device 114 may provide interconnectivity between the fluid injection system 112 and other devices or systems such as the medical imaging system 116. In some non-limiting embodiments, the workstation device 114 may include a Certegra® workstation provided by Bayer.

[0091] In some non-limiting embodiments, the medical imaging system 116 may include one or more devices that can communicate with an AI medical device platform 102, a fluid injection system 112, a workstation device 114, a digital pathology system 118, a laboratory information system 120, an EHR system 122, an EMR system 124, and / or a hospital information system 126 via a communication network (e.g., communication network 110). In some non-limiting embodiments, the medical imaging system 116 may include an ultrasound system, an echocardiography system, a magnetic resonance imaging (MRI) system, an electromagnetic radiation system (e.g., a conventional 2D X-ray, a 3D computed tomography (CT) scanning system, a fluoroscopy system, etc.) that can communicate via the communication network 110 and perform medical imaging procedures, including medical imaging procedures that include the use of radiographic contrast agents. In some non-limiting embodiments, the medical imaging system 116 may provide images of a patient and / or data associated with images of a patient (e.g., images of a patient as a result of imaging examinations). The data associated with the patient's image may include data in the Medical Digital Image Communication (DICOM) format, which may include metadata, pixel data, and / or additional data associated with imaging procedures performed on the patient to provide the image.

[0092] In some non-limiting embodiments, the digital pathology system 118 may include one or more devices capable of communicating with an AI medical device platform 102, a fluid injection system 112, a workstation device 114, a medical imaging system 116, a laboratory information system 120, an EHR system 122, an EMR system 124, and / or a hospital information system 126 via a communication network (e.g., communication network 110). In some non-limiting embodiments, the medical imaging system 116 may include one or more devices that receive, manage, transmit, and / or interpret pathology information, including data analyzed by a microscope, scanner, and / or other similar devices (e.g., image data, slide data, etc.).

[0093] In some non-limiting embodiments, the laboratory information system 120 may include one or more devices capable of communicating with the AI ​​medical device platform 102, the fluid injection system 112, the workstation device 114, the medical imaging system 116, the laboratory information system 120, the EHR system 122, the EMR system 124, and / or the hospital information system 126 via a communication network (e.g., communication network 110). In some non-limiting embodiments, the laboratory information system 120 may include one or more devices that record, manage, update, and / or store patient data and / or laboratory data for a clinical and / or anatomical pathology laboratory, including receiving laboratory instructions, sending instructions to laboratory analyzers, tracking instructions, results, and / or quality control information, and / or sending results to other systems or devices. In some non-limiting embodiments, the laboratory information system 120 may include a system designed to support the operation of a laboratory (e.g., a medical laboratory). The laboratory information system 120 may include features such as workflows, data tracking, a flexible architecture, and / or data exchange interfaces to support the use of laboratories in regulated environments. In some non-limiting embodiments, the laboratory information system 120 may include enterprise resource planning tools designed to manage aspects of laboratory informatics (e.g., patient data associated with laboratory processes and / or laboratory trials).

[0094] In some non-limiting embodiments, the EHR system 122 may include one or more devices capable of communicating with the AI ​​medical device platform 102, the fluid infusion system 112, the workstation device 114, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the EMR system 124, and / or the hospital information system 126 via a communication network (e.g., communication network 110). In some non-limiting embodiments, the EHR system 122 may include one or more devices that receive, manage, store, and / or transmit electronic health records, including medical record data associated with a patient's medical record, such as demographics, medical history, medications and allergies, immune status, laboratory test results, radiographic images, vital signs, personal statistics (e.g., age, weight, height, etc.), and billing information, associated with various providers and / or medical locations (e.g., offices, clinics, hospitals, etc.). Additionally or alternatively, the EHR system 122 may include a patient portal (e.g., a web-based interface) that enables a patient to interact with their respective electronic health records. In some non-limiting embodiments, the EMR system 124 may be a data source for the EHR system 122.

[0095] In some non-limiting embodiments, the EMR system 124 may include one or more devices capable of communicating with the AI ​​medical device platform 102, the fluid injection system 112, the workstation device 114, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the EHR system 122, and / or the hospital information system 126 via a communication network (e.g., communication network 110). In some non-limiting embodiments, the EMR system 124 may include one or more devices that receive, manage, store, and / or transmit electronic medical records (e.g., electronic health records) containing medical record data associated with a patient's medical record, such as demographics, medical history, medications and allergies, immune status, laboratory test results, radiographic images, vital signs, personal statistics (e.g., age, weight, height, etc.), and billing information.

[0096] In some non-limiting embodiments, the hospital information system 126 may include one or more devices that can communicate with the AI ​​medical device platform 102, the fluid injection system 112, the workstation device 114, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the EHR system 122, and / or the EMR system 124 via a communication network (e.g., communication network 110). For example, the hospital information system 126 may include one or more computing devices, such as one or more desktop computers, one or more mobile devices, and one or more servers. In some non-limiting embodiments, the hospital information system 126 may include one or more subsystems such as a patient procedure tracking system (e.g., a system for operating modality worklists, a system for providing patient demographic information for fluid injection procedures and / or medical imaging procedures), a fluid injector management system, an image archiving and communication system (e.g., a picture archiving and communication system (PACS)), a radiology information system (RIS), a radiology analysis system (e.g., Radimetrics® Enterprise Application, commercialized and marketed by Bayer HealthCare LLC), and / or other similar systems or devices.

[0097] As further shown in Figure 1B, the hospital information system 126 may include multiple subsystems. These subsystems may include a patient treatment tracking system 126A, an image archiving and communication system 126B, a radiology information system 126C, and a radiology analysis system 126D. In some non-limiting embodiments, the AI ​​medical device platform 102 may receive data from the hospital information system 126 via a communication network (e.g., communication network 110) according to a communication protocol for communicating data. For example, the AI ​​medical device platform 102 may receive data associated with patient procedures from the hospital information system 126 (e.g., from the patient procedure tracking system 126A) via a communication network in accordance with the Digital Medical Image Communication (DICOM) communication protocol, receive data associated with the operation of fluid injection systems from the hospital information system 126 via a communication network based on API calls (e.g., API calls from the AI ​​medical device platform 102), receive data associated with radiographic images from the hospital information system 126 (e.g., from the image archive and communication system 126B) via a communication network in accordance with the DICOM communication protocol, receive data associated with patient examination procedures from the hospital information system 126 (e.g., from the radiological information system 126C) via a communication network in accordance with the Health Level Seven (HL7) standard communication protocol, and / or receive data associated with radiation doses during medical imaging procedures from the hospital information system 126 (e.g., from the radiological analysis system 126D) via a communication network based on API calls (e.g., API calls from the AI ​​medical device platform 102 to the radiological analysis system 126D).

[0098] In some non-limiting embodiments, the AI ​​medical device platform 102 may include multiple applications, each of which may be associated with an API associated with the respective application that enables other systems and / or devices to interface with the AI ​​medical device platform 102 (e.g., communicate, establish a communication interface, etc.) and / or enable the AI ​​medical device platform 102 to interface with other systems and / or devices (e.g., individual subsystems of the hospital information system 126 such as the patient treatment tracking system 126A, the image archiving and communication system 126B, the radiology information system 126C, and / or the radiology analysis system 126D) (e.g., a first API associated with the first application, a second API associated with the second application, a third API associated with the third application, etc.). In some non-limiting embodiments, the AI ​​medical device platform 102 may provide a user interface (e.g., via an application that includes a user interface, such as a web-based user interface) that enables a user to access information (e.g., medical findings for a patient).

[0099] As further shown in Figure 1B, the workstation device 114 may include a display unit 114A. In some non-limiting embodiments, the display unit 114A may display a user interface (e.g., a web-based user interface) provided by the AI ​​medical device platform 102. In some non-limiting embodiments, the display unit 114A may include a computing device such as a smart display unit, a portable computer such as a tablet, a laptop, etc. In some non-limiting embodiments, the display unit 114A may include a touchscreen for receiving user input. In some non-limiting embodiments, the display unit 114A may include a display device (e.g., a monitor, screen, etc. for displaying visual information).

[0100] Referring here to Figure 2, Figure 2 is a diagram of exemplary components of device 200. Device 200 may correspond to an AI medical device platform 102, a data source 104, a user device 106, a data repository system 108, a fluid injection system 112, a workstation device 114, a medical imaging system 116, a digital pathology system 118, a laboratory information system 120, an EHR system 122, an EMR system 124, and / or a hospital information system 126. In some non-limiting embodiments, the AI ​​medical device platform 102, the data source 104, the user device 106, the data repository system 108, the fluid injection system 112, the workstation device 114, the medical imaging system 116, the digital pathology system 118, the laboratory information system 120, the EHR system 122, the EMR system 124, and / or a hospital information system 126 may include at least one device 200 and / or at least one component of device 200. As shown in Figure 2, the device 200 may include a bus 202, a processor 204, a memory 206, a storage component 208, an input component 210, an output component 212, and a communication component 214.

[0101] Bus 202 may include components that enable communication between components of device 200. In some non-limiting embodiments, the processor 204 may be implemented in hardware or in a combination of hardware and software. For example, the processor 204 may include a processor (e.g., a central processing unit (CPU), a graphics processing unit (GPU), an accelerator processing unit (APU), etc.), a microprocessor, a digital signal processor (DSP), and / or any processing component that can be programmed to perform a function (e.g., a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), etc.). Memory 206 may include random access memory (RAM), read-only memory (ROM), and / or another type of dynamic or static storage device (e.g., flash memory, magnetic memory, optical memory, etc.) that stores information and / or instructions for use by the processor 204.

[0102] The storage component 208 may store information and / or software related to the operation and use of device 200. For example, the storage component 208 may include, together with a corresponding drive, a hard disk (e.g., magnetic disk, optical disk, magneto-optical disk, solid-state disk, etc.), a compact disc (CD), a digital multipurpose disc (DVD), a floppy disk, a cartridge, a magnetic tape, and / or another type of computer-readable media.

[0103] The input component 210 may include components that enable the device 200 to receive information via user input (e.g., a touchscreen display, keyboard, keypad, mouse, buttons, switches, microphone, camera, etc.). Additionally or alternatively, the input component 210 may include sensors for sensing information (e.g., a Global Positioning System (GPS) component, accelerometer, gyroscope, actuator, etc.). The output component 212 may include components that provide output information from the device 200 (e.g., a display, speaker, one or more light-emitting diodes (LEDs), etc.).

[0104] The communication component 214 may include transceiver-like components (e.g., transceivers, separate receivers and transmitters) that enable device 200 to communicate with other devices via wired connections, wireless connections, or a combination of wired and wireless connections. The communication component 214 may also enable device 200 to receive information from and / or provide information to other devices. For example, the communication component 214 may include Ethernet interfaces, optical interfaces, coaxial interfaces, infrared interfaces, radio frequency (RF) interfaces, Universal Serial Bus (USB) interfaces, Wi-Fi® interfaces, cellular network interfaces, and the like.

[0105] Device 200 may perform one or more of the methods described herein. Device 200 may perform these methods based on a processor 204 that executes software instructions stored in a computer-readable medium such as memory 206 and / or storage component 208. A computer-readable medium (e.g., non-temporary computer-readable medium) is defined herein as a non-temporary memory device. A non-temporary memory device includes a memory space located within a single physical storage device or a memory space that extends across multiple physical storage devices.

[0106] Software instructions may be read into memory 206 and / or storage component 208 from another computer-readable medium or from another device via communication component 214. When executed, the software instructions stored in memory 206 and / or storage component 208 may cause the processor 204 to execute one or more of the methods described herein. Additionally or alternatively, embedded circuitry may be used instead of or in combination with software instructions to execute one or more of the methods described herein. Therefore, the embodiments described herein are not limited to any particular combination of hardware circuitry and software.

[0107] The memory 206 and / or storage component 208 may include data storage or one or more data structures (e.g., a database). The device 200 may retrieve information from, store, or retrieve information stored in the data storage devices or one or more data structures within the memory 206 and / or storage component 208.

[0108] The number and arrangement of components shown in Figure 2 are provided as an example. In some non-limiting embodiments, device 200 may include additional components, fewer components, different components, or components in different arrangements than those shown in Figure 2. Additionally or alternatively, a set of components of device 200 (e.g., one or more components) may perform one or more functions described herein as being performed by another set of components of device 200.

[0109] Referring here to Figure 3, Figure 3 is a flowchart of a non-limiting embodiment of Method 300 for generating an automated healthcare application. In some non-limiting embodiments, one or more of the steps of Method 300 may be performed (e.g., entirely, partially, etc.) by the AI ​​medical device platform 102 (e.g., one or more devices of the AI ​​medical device platform 102). In some non-limiting embodiments, one or more of the steps of Method 300 may be performed (e.g., entirely, partially, etc.) by a separate device or group of devices separate from, or including, the AI ​​medical device platform 102, data source 104, user device 106, and / or data repository system 108. The steps shown in Figure 3 are for illustrative purposes only. It will be understood that in some non-limiting embodiments or aspects, additional, fewer, different, and / or different order of steps may be used. In some non-limiting embodiments or aspects, steps may be performed automatically in response to the execution and / or completion of the previous step. As used herein, an automated healthcare application may mean one or more non-machine learning automated healthcare applications (e.g., one or more automated healthcare applications that are programmed and do not include one or more machine learning models) and / or one or more machine learning automated healthcare applications (e.g., one or more automated healthcare applications that include one or more machine learning models).

[0110] As shown in Figure 3, in step 302, method 300 includes the step of receiving formatted data associated with a plurality of medical procedures for the generation of one or more automated healthcare applications. For example, the AI ​​medical device platform 102 may receive formatted data associated with a plurality of medical procedures for the generation of one or more automated healthcare applications from the data repository system 108. In some non-limiting embodiments, the formatted data may include formatted data associated with a plurality of medical procedures for the generation of one or more machine learning models for automated healthcare applications.

[0111] In some non-limiting embodiments, the data repository system 108 may receive initial data (e.g., raw data) associated with one or more data records via one or more data pipelines (e.g., extract, transform, load (etl) pipelines). In some non-limiting embodiments, the data repository system 108 may establish one or more data pipelines to enable the automatic retrieval of data associated with one or more data records from the data source 104.

[0112] In some non-limiting embodiments, the data associated with one or more data records may include data associated with the patient's medical procedures, such as data associated with electronic medical records from an EMR system (e.g., electronic patient records), data associated with images generated by a medical imaging system, data associated with patient biometric measurements, data associated with fluid infusion procedures performed by a fluid infusion system, data associated with records from a pathology information system and / or laboratory information system. In some non-limiting embodiments, the data associated with the patient's medical procedures may include data associated with fluid infusion procedures, which may include data associated with contrast fluids provided during the fluid infusion procedure, the gauge of the catheter used during the fluid infusion procedure, the fluid infusion protocol for the fluid infusion procedure, the configuration of the fluid infusion system, etc. Additionally or alternatively, the data associated with the patient's medical procedures may include data associated with medical imaging procedures, which may include data associated with images (e.g., radiographic images), data associated with radiation dose, etc. Additionally or alternatively, the data may include data associated with patient identification (e.g., metadata), data associated with patient medical procedures (e.g., data about medical procedures performed on the patient), and data associated with the identification of devices involved in patient medical procedures (e.g., data associated with the identification of fluid infusion systems, data associated with the identification of patient devices, etc.).

[0113] In some non-limiting embodiments, the data repository system 108 may receive initial data associated with one or more data records based on a device associated with a data source 104 that performs a medical procedure. For example, the data repository system 108 may receive data after a device associated with the data source 104 has performed a procedure. In some non-limiting embodiments, the data repository system 108 may receive data from devices associated with the hospital (e.g., servers operated by the hospital, such as a field server located in the hospital or a server located remotely from the hospital, or a hospital information system). In some non-limiting embodiments, the data repository system 108 may store the data based on its reception (e.g., using an application, a memory device, etc.). In some non-limiting embodiments, the data repository system 108 may receive data (e.g., from a hospital information system) according to a communication protocol for communicating data over a communication network. For example, the data repository system 108 may receive healthcare data according to the DICOM communication protocol, the Health Level Seven (HL7) standard communication protocol, etc.

[0114] In some non-limiting embodiments, the data repository system 108 may, based on the receipt of data associated with one or more data records from the data source 104, perform one or more data processing procedures to provide formatted data associated with multiple medical procedures for the generation of one or more automated healthcare applications (e.g., one or more machine learning models for automated healthcare applications). In some non-limiting embodiments, the data processing procedures may include one or more data curation procedures. In some non-limiting embodiments, the data curation procedures may include deidentification procedures (e.g., procedures for deidentifying data associated with data records, such as for the removal of Personal Health Information (PHI)), quality check procedures, data bias check procedures, data transformation procedures, deduplication procedures, data enrichment procedures (e.g., procedures for applying preprocessing algorithms to raw data, procedures for creating additional data such as synthetic data such as image rotation, data equilibration and / or data enrichment procedures), data joining procedures, data annotation procedures (e.g., procedures for labeling data, such as drawing bounding boxes around parts in an image), data versioning procedures, and the like. Additionally or alternatively, the data repository system 108 may perform one or more data transformation steps based on the receipt of data associated with one or more data records from the data source 104 in order to provide formatted data associated with multiple medical procedures for the generation of automated healthcare applications. In some non-limiting embodiments, one or more data transformation steps may include feature extraction steps, denoising steps, reduction steps (e.g., text reduction steps in which text is removed from data records, image reduction steps in which at least a portion of an image is removed from a data record), and so on. In some non-limiting embodiments, the formatted data associated with the generation of automated healthcare applications may include training data, validation data, and / or test data for one or more machine learning models.

[0115] In some non-limiting embodiments, the data repository system 108 may segment the data associated with one or more data records and / or formatted data associated with the generation of an automated healthcare application into multiple datasets for the generation of an automated healthcare application. For example, the data repository system 108 may segment the data associated with one or more data records into datasets based on the fact that one or more data processing steps have been performed on the data. In some non-limiting embodiments, each dataset in the multiple datasets may be associated with a specific medical procedure among a plurality of medical procedures.

[0116] In some non-limiting embodiments, the AI ​​medical device platform 102 and / or data repository system 108 may receive access requests from a user associated with a user device 106 to a specific dataset of multiple datasets, determine whether the access request conforms to one or more criteria associated with access to the specific dataset, and, based on the determination that the access request conforms to one or more criteria associated with access to the specific dataset, provide the user device 106 with access to the specific dataset. In some non-limiting embodiments, the AI ​​medical device platform 102 and / or data repository system 108 may determine whether the access request conforms to criteria based on user identification information (e.g., a device identifier associated with the user device 106, such as a username, password, identification number, Internet Protocol (IP) address, or Media Access Control (MAC) address). For example, the AI ​​medical device platform 102 and / or data repository system 108 may determine whether the user identification information provided as part of the access request matches one or more criteria associated with access to a specific dataset. The AI ​​medical device platform 102 and / or data repository system 108 may determine whether the user identification information matches additional user identification information stored in a data structure. In some non-limiting embodiments, the AI ​​medical device platform 102 and / or data repository system 108 may provide the user with access to a specific dataset if it determines that the access request conforms to one or more criteria. The AI ​​medical device platform 102 and / or data repository system 108 may not provide the user with access to a specific dataset if it determines that the access request does not conform to one or more criteria. In some non-limiting embodiments, one or more criteria associated with access to a specific dataset may include whether the user has access to a subscription associated with that specific dataset.In some non-limiting embodiments, when providing access to a specific dataset, the AI ​​medical device platform 102 may make application programming interface (API) calls to the data repository system 108 related to the specific dataset in order to provide access to the specific dataset.

[0117] As shown in Figure 3, in step 304, method 300 includes the step of receiving data associated with one or more regulatory guidance and approval processes. In some non-limiting embodiments, the AI ​​medical device platform 102 may receive data associated with one or more regulatory guidance and approval processes from the data repository system 108. In some non-limiting embodiments, the data associated with one or more regulatory guidance and approval processes may include standards published by government agencies such as the U.S. Food and Drug Administration (FDA) and the European Union Medical Devices Coordination Group (MDCG) (e.g., regulatory standards and / or non-regulatory standards), and standards published by non-governmental agencies such as the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). In some non-limiting embodiments, the AI ​​medical device platform 102 may receive data associated with one or more regulatory guidance and approval processes from the data repository system 108.

[0118] As shown in Figure 3, in step 306, method 300 includes the step of generating one or more automated healthcare applications configured to provide outputs associated with a medical procedure. In some non-limiting embodiments, the AI ​​medical device platform 102 may generate one or more automated healthcare applications based on formatted data associated with a plurality of medical procedures for the generation of one or more automated healthcare applications, and / or data associated with one or more regulatory guidance and approval processes. In some non-limiting embodiments, the AI ​​medical device platform 102 may generate one or more machine learning models for one or more automated healthcare applications. In some non-limiting embodiments, one or more automated healthcare applications (e.g., one or more machine learning models for one or more automated healthcare applications) may be configured to provide outputs associated with a manner of medical procedure (e.g., outcomes). In some non-limiting embodiments, one or more automated healthcare applications may include computer-aided detection applications and / or computer-aided diagnostic applications.

[0119] In some non-limiting embodiments, the AI ​​medical device platform 102 may generate one or more machine learning models. In some non-limiting embodiments, the machine learning models may include models designed to receive data as input and provide as output predictions of outcomes associated with medical procedures (e.g., predictions of image classification, predictions of the presence of disease states, predictions of treatment procedures, etc.). For example, a machine learning model may receive data (e.g., data associated with a patient's procedure) and provide output including predicted classifications for multiple automated healthcare data analysis applications. The predicted classifications for automated healthcare data analysis applications may include classifications for automated healthcare data analysis applications that can provide further analysis of healthcare data. In some non-limiting embodiments, the AI ​​medical device platform 102 may store the machine learning models (e.g., for later use).

[0120] In some non-limiting embodiments, as described herein, the AI ​​medical device platform 102 may process data (e.g., data associated with multiple medical procedures) to obtain training data (e.g., training dataset) for a machine learning model. For example, the AI ​​medical device platform 102 may process data to transform it into a format that can be analyzed (e.g., by the AI ​​medical device platform 102) to generate a machine learning model. The transformed data (e.g., data resulting from the transformation) may be called training data. In some non-limiting embodiments, the AI ​​medical device platform 102 may process data to obtain training data based on the fact that it has received data. Additionally or alternatively, the AI ​​medical device platform 102 may process data to obtain training data based on the fact that the AI ​​medical device platform 102 has received instructions from a user of the AI ​​medical device platform 102 (e.g., a user associated with user device 106) that the AI ​​medical device platform 102 should process data, such as when the AI ​​medical device platform 102 receives instructions to generate a machine learning model (e.g., about time intervals corresponding to the data, locations corresponding to the data, devices corresponding to the data, etc.).

[0121] In some non-limiting embodiments, the AI ​​medical device platform 102 may process data by determining predictor variables based on the data. The predictor variables may include metrics associated with one or more medical procedures from a plurality of medical procedures that can be derived based on the data. The predictor variables may be analyzed to generate a machine learning model. For example, the predictor variables may include variables associated with a patient's procedure (e.g., the duration of the medical procedure, the protocol used during the medical procedure, the amount of substance used during the medical procedure, the outcome of the medical procedure, etc.), variables associated with the aspect of an image (e.g., radiographic image), variables associated with the patient's demographics, variables associated with the patient's condition, etc.

[0122] In some non-limiting embodiments, the AI ​​medical device platform 102 may analyze training data to generate a machine learning model. For example, the AI ​​medical device platform 102 may use machine learning techniques to analyze training data and generate a machine learning model. In some non-limiting embodiments, generating a machine learning model may be referred to as training a machine learning model. Machine learning techniques may include, for example, supervised and / or unsupervised techniques such as decision trees, random forests, logistic regression, linear regression, gradient boosting, support vector machines, extra trees (e.g., extensions of random forests), Bayesian statistics, learning automata, hidden Markov modeling, linear classifiers, quadratic classifiers, association rule learning, and computer vision. In some non-limiting embodiments, the machine learning model may include models specific to particular characteristics, e.g., models specific to a particular location (e.g., a particular location in a hospital), specific patient demographics, specific systems used during patient treatment, etc. Additionally or alternatively, the machine learning model may be specific to a particular entity providing healthcare services (e.g., a healthcare provider such as a hospital, a clinical facility, or a group of physicians). In some non-limiting embodiments, the AI ​​medical device platform 102 may generate one or more machine learning models for one or more entities, a particular group of entities, and / or one or more users of one or more entities.

[0123] Additionally or alternatively, when analyzing training data, the AI ​​medical device platform 102 may identify one or more variables (e.g., one or more independent variables) as predictor variables (e.g., features) that can be used to make predictions when analyzing the training data. In some non-limiting embodiments, the values ​​of the predictor variables may be inputs to a machine learning model. For example, the AI ​​medical device platform 102 may identify a subset of variables (e.g., a proper subset) as predictor variables that can be used to accurately predict outcomes associated with medical procedures. In some non-limiting embodiments, the predictor variables may include one or more predictor variables that have a significant influence (e.g., an influence that satisfies a threshold) on the prediction determined by the AI ​​medical device platform 102, as described above.

[0124] In some non-limiting embodiments, the AI ​​medical device platform 102 may validate the machine learning model. For example, the AI ​​medical device platform 102 may validate the machine learning model after it has generated it. In some non-limiting embodiments, the AI ​​medical device platform 102 may validate the machine learning model based on a portion of the training data used for validation. For example, the AI ​​medical device platform 102 may divide the training data into a first portion and a second portion, the first portion of which may be used to generate the machine learning model as described above. In this example, the second portion of the training data (e.g., validation data) may be used to validate the machine learning model.

[0125] In some non-limiting embodiments, the AI ​​medical device platform 102 may validate the machine learning model by providing validation data as input to the machine learning model and determining, based on the output of the machine learning model, whether the machine learning model correctly or incorrectly predicted the outcomes associated with the medical procedure. In some non-limiting embodiments, the AI ​​medical device platform 102 may validate the machine learning model based on validation thresholds. For example, the AI ​​medical device platform 102 may be configured to validate the machine learning model when the machine learning model correctly predicts the outcomes associated with the medical procedure (identified by the validation data) (e.g., when the machine learning model correctly predicts 50% of the outcomes, 70% of the outcomes, a threshold amount of the outcomes, etc.).

[0126] In some non-limiting embodiments, if the AI ​​medical device platform 102 does not validate the machine learning models (for example, when the percentage of accurately predicted outcomes associated with a medical procedure does not meet the validation threshold), the AI ​​medical device platform 102 may generate one or more additional machine learning models.

[0127] In some non-limiting embodiments, once a machine learning model has been validated, the AI ​​medical device platform 102 may further train the machine learning model and / or generate a new machine learning model based on the receipt of new training data. The new training data may include additional data associated with multiple medical procedures. In some non-limiting embodiments, the new training data may include new healthcare data associated with multiple patient procedures (e.g., data associated with multiple patient procedures performed after a specified time interval). The AI ​​medical device platform 102 may use the machine learning model to predict outcomes associated with medical procedures and compare the output of the machine learning model with the new training data, which includes the data associated with multiple medical procedures. In such an example, the AI ​​medical device platform 102 may update one or more machine learning models based on the new training data.

[0128] In some non-limiting embodiments, the AI ​​medical device platform 102 may store machine learning models. For example, the AI ​​medical device platform 102 may store machine learning models in a data structure (e.g., a database, linked list, tree, etc.). The data structure may be located within the AI ​​medical device platform 102 or outside of the AI ​​medical device platform 102 (e.g., remotely), such as in a data repository system 108.

[0129] In some non-limiting embodiments, when generating an automated healthcare application configured to predict one or more aspects associated with a medical procedure, the AI ​​medical device platform 102 may provide one or more trained machine learning models for the automated healthcare application based on data associated with multiple medical procedures, or the AI ​​medical device platform 102 may validate the output of one or more trained machine learning models based on data associated with regulatory guidance and approval processes. In some non-limiting embodiments, the AI ​​medical device platform 102 may validate the output by comparing it to standards included in the data associated with regulatory guidance and approval processes. If the output meets the standards, the AI ​​medical device platform 102 may validate one or more trained machine learning models. If the output does not meet the standards, the AI ​​medical device platform 102 may refrain from validating one or more trained machine learning models.

[0130] In some non-limiting embodiments, the AI ​​medical device platform 102 may train one or more machine learning models on a first dataset of multiple datasets and provide one or more trained machine learning models, or the AI ​​medical device platform 102 may assign a unique identifier to the first dataset and a unique identifier to the one or more trained machine learning models based on the fact that it has trained one or more machine learning models. In some non-limiting embodiments, when validating the output of one or more trained machine learning models, the AI ​​medical device platform 102 may validate the first dataset and one or more trained machine learning models based on the unique identifiers assigned to the first dataset and one or more trained machine learning models. For example, the AI ​​medical device platform 102 may validate the first dataset and one or more trained machine learning models by comparing the unique identifier assigned to the first dataset with the unique identifier assigned to one or more trained machine learning models. The AI ​​medical device platform 102 may validate one or more trained machine learning models if the unique identifier assigned to the first dataset and the unique identifier assigned to one or more trained machine learning models correspond. If the unique identifier assigned to the first dataset does not correspond to the unique identifier assigned to one or more trained machine learning models, the AI ​​medical device platform 102 may discontinue validating the one or more trained machine learning models.

[0131] As shown in Figure 3, in step 308, process 300 includes the step of deploying one or more automated healthcare applications to the production environment. For example, the AI ​​medical device platform 102 may deploy one or more automated healthcare applications to the production environment based on the fact that it has generated one or more automated healthcare applications.

[0132] In some non-limiting embodiments, the AI ​​medical device platform 102 may deploy one or more automated healthcare applications on a production environment of a third-party platform not associated with the AI ​​medical device platform 102 (e.g., a third-party platform operated by an entity not associated with the entity operating the AI ​​medical device platform 102), a production environment of the AI ​​medical device platform, a platform including standardized containers, or any combination thereof.

[0133] In some non-limiting embodiments, before deploying the automated healthcare application, the AI ​​medical device platform 102 may perform a clinical evaluation of the automated healthcare application based on data associated with one or more regulatory guidance and approval processes. For example, the AI ​​medical device platform 102 may perform a performance evaluation of the automated healthcare application using clinical trial data to determine whether the automated healthcare application meets regulatory standards. In some non-limiting embodiments, the performance evaluation may include retrospective read-through studies (e.g., studies including clinical trial data) and / or prospective read-through studies (e.g., studies including evaluations based on data acquired during the trial). In some non-limiting embodiments, the AI ​​medical device platform 102 may generate one or more reports based on the results of the performance evaluation.

[0134] In some non-limiting embodiments, the AI ​​medical device platform 102 may receive an access request from a user associated with a user device 106 to a specific automated healthcare application among a plurality of automated healthcare applications (e.g., one or more machine learning models of an automated healthcare application), determine whether the access request conforms to one or more criteria associated with access to the specific automated healthcare application, and, based on the determination that the access request conforms to one or more criteria associated with access to the specific automated healthcare application, provide the user device 106 with access to the specific automated healthcare application. In some non-limiting embodiments, the AI ​​medical device platform 102 may determine whether the access request conforms to criteria based on user identification information (e.g., a device identifier associated with the user device 106, such as a username, password, identification number, Internet Protocol (IP) address, or Media Access Control (MAC) address). For example, the AI ​​medical device platform 102 may determine whether the user identification information provided as part of the access request matches one or more criteria associated with access to a specific automated healthcare application. The AI ​​medical device platform 102 may determine whether the user identification information matches additional user identification information stored in a data structure. In some non-limiting embodiments, the AI ​​medical device platform 102 may grant the user access to a specific automated healthcare application if it determines that the access request conforms to one or more criteria. The AI ​​medical device platform 102 may not grant the user access to a specific automated healthcare application if it determines that the access request does not conform to one or more criteria.In some non-limiting embodiments, one or more criteria associated with access to a particular automated healthcare application may include whether the user has access to a subscription related to a particular dataset.

[0135] Referring now to Figure 4, which is a diagram of a non-limiting embodiment of the AI ​​medical device platform 102. In some non-limiting embodiments, the AI ​​medical device platform 102 may include a plurality of subsystems (e.g., hardware components, modules, software components such as applications, extensions, and units configured to run on one or more devices of the AI ​​medical device platform 102 or any combination thereof). In some non-limiting embodiments, the plurality of subsystems of the AI ​​medical device platform 102 may be interconnected with one another via wired connections, wireless connections, or a combination of wired and wireless connections.

[0136] As shown in Figure 4, the subsystems may include an experimental subsystem 402, a research workflow subsystem 404, a data science subsystem 406, a clinical trial subsystem 408, a product development subsystem 410, a validation subsystem 412, a deployment subsystem 414, and a monitoring subsystem 416. In some non-limiting embodiments, one or more subsystems of the AI ​​medical device platform 102 may be hosted on a network device (e.g., a network edge device) closest to the equipment interacting with the AI ​​medical device platform 102. In this way, results from one or more subsystems of the AI ​​medical device platform 102 may be received in a shorter time compared to a situation where one or more subsystems are hosted on different network devices. In some non-limiting embodiments, the AI ​​medical device platform 102 may include more or fewer subsystems than those shown in Figure 4. In some non-limiting embodiments, the AI ​​medical device platform 102 may include one or more subsystems configured to operate based on a manually provided configuration (e.g., a configuration provided by a user such as a physician or technician).

[0137] In some non-limiting embodiments, the experimental subsystem 402 may be configured to receive and / or provide specification data associated with one or more aspects of a medical procedure in order to enable the development (e.g., development of the concept) of one or more machine learning models associated with a medical procedure (e.g., one or more automated healthcare applications comprising one or more machine learning models) and / or to enable the generation of proof of concept of one or more machine learning models associated with a medical procedure. In some non-limiting embodiments, the experimental subsystem 402 may provide recommendations for one or more machine learning models associated with a medical procedure. For example, the experimental subsystem 402 may provide recommendations for the concept of one or more machine learning models associated with a medical procedure based on user input including specification data.

[0138] In some non-limiting embodiments, the experimental subsystem 402 may receive data associated with a concept for an automated healthcare application. For example, the experimental subsystem 402 may receive data associated with a concept for a machine learning model that provides an output which is a prediction of treatment outcomes. In some non-limiting embodiments, the experimental subsystem 402 may generate data associated with multiple machine learning models (e.g., multiple types of machine learning models, multiple machine learning model architectures, multiple parameters associated with the machine learning models, etc.) based on the data associated with the concept for an automated healthcare application. For example, the experimental subsystem 402 may provide a list of multiple machine learning models. In some non-limiting embodiments, the experimental subsystem 402 may receive a selection of one or more machine learning models (e.g., a user selection), or the experimental subsystem 402 may generate data associated with a proof of concept for an automated healthcare application.

[0139] In some non-limiting embodiments, the research workflow subsystem 404 may be configured to provide automation assistance and / or data enhancement procedures. In some non-limiting embodiments, the research workflow subsystem 404 may provide data enhancement procedures such as image quality modifications, including the identification of anatomical regions of the body in an image, noise reduction in an image, blurring in an image, and / or text detection in an image.

[0140] In some non-limiting embodiments, the data science subsystem 406 may be configured to provide standard software tools and / or libraries, procedures for automating sequential integration and sequential delivery / deployment (CI / CD), procedures for abstraction from infrastructure, medical image processing capabilities, model training capabilities, procedures for machine learning paradigms including federative learning, procedures for machine learning model cataloging and / or machine learning model sharing, and / or procedures for collaboration for machine learning model development.

[0141] In some non-limiting embodiments, the clinical trial subsystem 408 may be configured to receive, store, and / or provide data related to clinical trials of an automated healthcare application. In some examples, the clinical trial subsystem 408 may receive, store, and / or provide data associated with evaluations of the automated healthcare application (e.g., product evaluation, product evaluation design, etc.), data associated with clinical site settings, data associated with evaluation system settings, and / or data associated with inputs and / or outputs provided to and / or received from the automated healthcare application (e.g., image data deidentification, image data transfer, image data quality check, image data curation, and / or preparation). In some non-limiting embodiments, the clinical trial subsystem 408 may be accessible separately from other subsystems of the AI ​​medical device platform 102. For example, the clinical trial subsystem 408 may be isolated from other subsystems by a data firewall.

[0142] In some non-limiting embodiments, the product development subsystem 410 may be configured to provide data associated with software development tools, an integrated and validated toolchain, data associated with user feedback, data associated with issue tracking, data associated with regulatory documents, data associated with technical documents, a code repository, an artifact repository, tools for data tracking, tools for model tracking, tools for experiment tracking, and / or data pipelined.

[0143] In some non-limiting embodiments, the verification subsystem 412 may be configured to provide the ability to determine whether the automated healthcare application meets regulatory requirements associated with the output of the automated healthcare application (e.g., the output of a machine learning model, which includes predictions about one or more aspects associated with a healthcare procedure, such as predictive treatment, predictive diagnosis, or predictive images of an imaging procedure).

[0144] In some non-limiting embodiments, the verification subsystem 412 may be configured to provide the ability to determine whether the automated healthcare application meets regulatory requirements associated with the output of the automated healthcare application (e.g., the output of a machine learning model, which includes predictions about one or more aspects associated with a medical procedure, such as predictive treatment, predictive diagnosis, or predictive images of an imaging procedure).

[0145] In some non-limiting embodiments, the deployment subsystem 414 may be configured to provide the ability to deploy automated healthcare applications to a production environment. In some examples, the deployment subsystem 414 may provide containerization of automated healthcare applications, container management of containers for automated healthcare applications, continuous deployment procedures for automated healthcare applications, data integration with other services (including other AI-based services such as Bayer Calantic® Digital Solutions), and / or integration with a developer portal for other services (including other AI-based services such as Bayer Calantic® Digital Solutions).

[0146] In some non-limiting embodiments, the monitoring subsystem 416 may be configured to provide the ability to monitor automated healthcare applications deployed in a production environment. For example, the monitoring subsystem 416 may be configured to receive, store, and / or provide an analysis of the output of automated healthcare applications deployed in a production environment. In some non-limiting embodiments, the monitoring subsystem 416 may be configured to monitor the automated healthcare application for issues associated with data drift, issues associated with model drift, usage information of the automated healthcare application, and / or information associated with the lifecycle management of the automated healthcare application. In some non-limiting embodiments, the monitoring subsystem 416 may be configured to monitor the automated healthcare application based on predetermined time intervals (e.g., hourly time intervals, daily time intervals, weekly time intervals, monthly time intervals, etc.). In some non-limiting embodiments, the monitoring subsystem 416 may be configured to determine the accuracy of the automated healthcare application and compare its accuracy to an accuracy threshold (e.g., an accuracy threshold based on government regulations, an accuracy threshold based on key performance indicators (KPIs)). If the monitoring subsystem 416 determines that the accuracy of the automated healthcare application meets the accuracy threshold, the monitoring subsystem 416 may allow the automated healthcare application to remain online (for example, to remain online in the production environment). If the monitoring subsystem 416 determines that the accuracy of the automated healthcare application does not meet the accuracy threshold, the monitoring subsystem 416 may take the automated healthcare application offline (for example, to remove the automated healthcare application from the production environment).

[0147] Referring now to Figure 5, which is a diagram of a non-limiting embodiment of the data repository system 108. In some non-limiting embodiments, the data repository system 108 may include a plurality of subsystems (e.g., software components such as hardware components, modules, applications, extensions, and units configured to run on one or more devices of the data repository system 108 or any combination thereof). In some non-limiting embodiments, the plurality of subsystems of the data repository system 108 may be interconnected with one another via wired connections, wireless connections, or a combination of wired and wireless connections.

[0148] As shown in Figure 5, the subsystems may include a data ingestion subsystem 502, a data storage subsystem 504, a data processing subsystem 506, a data services subsystem 508, a visualization and analysis subsystem 510, and a cloud infrastructure subsystem 512. In some non-limiting embodiments, one or more subsystems of the subsystems of the data repository system 108 may be hosted on a network device (e.g., a network edge device) closest to the equipment interacting with the data repository system 108. In this way, results from one or more subsystems of the data repository system 108 can be received in a shorter time compared to a situation where one or more subsystems are hosted on different network devices. In some non-limiting embodiments, the data repository system 108 may include more or fewer subsystems than those shown in Figure 5. In some non-limiting embodiments, the data repository system 108 may include one or more subsystems configured to operate based on a manually provided configuration (e.g., a configuration provided by a user such as a physician or technician).

[0149] In some non-limiting embodiments, the data ingestion subsystem 502 may be configured to receive initial data (e.g., raw data) associated with one or more data records from multiple data sources (e.g., data source 104). In some non-limiting embodiments, the data ingestion subsystem 502 may create temporary virtual machine instances for ingesting data from on-premises servers (e.g., via Secure File Transfer Protocol (SFTP) and / or File Transfer Protocol (FTP)) and / or cloud storage platforms. The jobs performed by the data ingestion subsystem 502 may be repeated according to predetermined time intervals.

[0150] In some non-limiting embodiments, the data storage subsystem 504 may be configured to function as a primary source of data for the data repository system 108. In some non-limiting embodiments, the data storage subsystem 504 may distinguish between data from different data sources and / or data in different formats and store the data in a memory device.

[0151] In some non-limiting embodiments, the data processing subsystem 506 may be configured to process initial data associated with one or more data records received from multiple data sources received through one or more data pipelines. In some non-limiting embodiments, the data processing subsystem 506 may transform the initial data based on an extract, transform, load (ETL) pipeline and provide transformed data used for generating one or more automated healthcare applications (e.g., one or more machine learning models for one or more automated healthcare applications).

[0152] In some non-limiting embodiments, the data service subsystem 508 may be configured to provide manual and / or automated access to data stored in the data repository system 108. In some non-limiting embodiments, the data service subsystem 508 may be configured to provide services related to viewing data in image form, including an image viewer, a service for generating ground truth information for the data, a service for defining protocols for collecting the data, and services related to data generation, including generative AI algorithms. In some non-limiting embodiments, the services provided by the data service subsystem 508 may be based on clinical procedures.

[0153] In some non-limiting embodiments, the visualization and analysis subsystem 510 may be configured to provide access to data visualization tools and / or database analysis tools used for generating one or more automated healthcare applications (e.g., one or more machine learning models for one or more automated healthcare applications). In some non-limiting embodiments, the visualization and analysis subsystem 510 may also include services for performing online analytical processing (OLAP) on data and for searching and retrieving specific data of interest to the user.

[0154] In some non-limiting embodiments, the cloud infrastructure subsystem 512 may be configured to control operations related to the data repository system 108. For example, the cloud infrastructure subsystem 512 may provide one or more rules associated with control operations of the data repository system 108 (e.g., authentication, authorization, network and security, data encryption, monitoring and logging modes, etc.).

[0155] Referring now to Figures 6A to 6D, Figures 6A to 6D are diagrams of non-limiting embodiments of an implementation 600 of a method (e.g., method 300) for generating an AI-based automated healthcare application using an AI medical device platform.

[0156] As shown by reference numeral 605 in Figure 6A, the AI ​​medical device platform 102 may receive initial data from the data source 104. In some non-limiting embodiments, the initial data may include initial data (e.g., raw data) associated with one or more data records of one or more medical procedures.

[0157] As shown by reference numeral 610 in Figure 6B, the AI ​​medical device platform 102 may process the initial data to provide formatted data. In some non-limiting embodiments, the AI ​​medical device platform 102 may process the initial data by performing one or more data curation steps on the initial data to provide formatted data. In some non-limiting embodiments, the formatted data may include formatted data associated with multiple medical procedures for the generation of one or more machine learning models. As further shown by reference numeral 615 in Figure 6B, the AI ​​medical device platform 102 may train one or more machine learning models based on the formatted data.

[0158] As shown by reference numeral 620 in Figure 6C, the AI ​​medical device platform 102 may receive data associated with one or more regulatory guidance and approval processes. In some non-limiting embodiments, the AI ​​medical device platform 102 may receive data associated with one or more regulatory guidance and approval processes from the data repository system 108. In some non-limiting embodiments, the data associated with one or more regulatory guidance and approval processes may include standards published by government agencies (e.g., regulatory standards and / or non-regulatory standards) and / or standards published by non-governmental agencies.

[0159] As further shown by reference no. 625 in Figure 6C, the AI ​​medical device platform 102 may validate one or more trained machine learning models based on data associated with one or more regulatory guidance and approval processes. As further shown by reference no. 630 in Figure 6C, the AI ​​medical device platform 102 may deploy one or more trained machine learning models to a production environment.

[0160] As shown by reference no. 635 in Figure 6D, the AI ​​medical device platform 102 may receive requests to access a specific machine learning model among multiple machine learning models. As shown by reference no. 640 in Figure 6D, the AI ​​medical device platform 102 may provide access to a specific machine learning model.

[0161] The systems, methods, and computer program products described above are described in detail for illustrative purposes based on what is considered to be the most practical and preferred embodiments at present; however, such details are for that purpose only, and it should be understood that this disclosure is not limited to the embodiments described, but rather intended to cover modifications and equivalent configurations that fall within the spirit and scope of the appended claims. For example, it should be understood that this disclosure is intended to allow, wherever possible, at least one feature of any embodiment or aspect to be combined with at least one feature of any other embodiment or aspect. [Explanation of symbols]

[0162] 100A environment 100B System 102 Artificial Intelligence (AI) Medical Device Platform 104 Data Sources 106 User Devices 108 Data Repository System 110 Communication Network 112 Fluid injection system 114 Workstation Devices 114A Display Unit 116 Medical Imaging Systems 118 Digital Pathology System 120 Laboratory Information System 122 Electronic Health Record (EHR) Systems 124 Electronic Medical Record (EMR) Systems 126 Hospital Information Systems 126A Patient Treatment Tracking System 126B Image archiving and communication system 126C Radiation Information System 126D Radiation Analysis System 200 devices Bus 202 204 Processors 206 memory 208 Memory Components 210 Input Components 212 Output Components 214 Communication Components 300 ways 402 Experimental Subsystem 404 Research Workflow Subsystem 406 Data Science Subsystem 408 Clinical Trial Subsystem 410 Product Development Subsystem 412 Verification subsystem 414 Deployment subsystems 416 Monitoring Subsystem 502 Data Ingestion Subsystem 504 Data Storage Subsystem 506 Data Processing Subsystem 508 Data Services Subsystem 510 Visualization and Analysis Subsystem 512 Cloud Infrastructure Subsystem

Claims

1. A system for generating AI-based automated healthcare applications, A data repository system that stores at least one of the following: formatted data associated with multiple medical procedures for the generation of one or more machine learning models, data associated with regulatory guidance and approval processes, or any combination thereof; An AI medical device platform comprising at least one processor that communicates with the aforementioned data repository system The at least one processor is The formatted data associated with multiple medical procedures for the generation of one or more machine learning models is received, The data associated with one or more regulatory guidance and approval processes is received, Based on the formatted data associated with the plurality of medical procedures and the data associated with the regulatory guidance and approval process, one or more machine learning models are generated that are configured to predict one or more aspects associated with the medical procedures. A system configured to deploy the one or more machine learning models to a production environment based on the generation of the one or more machine learning models.

2. The aforementioned data repository system is A data ingestion subsystem that receives data from multiple data sources, A data storage subsystem for storing data for the aforementioned data repository system, A data processing subsystem that transforms the data from the plurality of data sources based on an extract, transform, load (ETL) pipeline in order to provide transformed data used for generating the one or more machine learning models, A data analysis subsystem that provides access to data visualization tools and database analysis tools used for generating one or more machine learning models, A data cloud infrastructure subsystem that provides one or more rules associated with the control operations for the data repository system. The system according to claim 1, comprising:

3. The aforementioned data repository system is Electronic Medical Record (EMR) system, Medical imaging systems, Fluid injection system, Pathology information system, Laboratory information system, or Any combination of them Initial data associated with one or more data records is received from multiple data sources, including at least one of the following: The system according to claim 1, configured to perform one or more data curation steps on the initial data associated with one or more data records in order to provide the formatted data associated with a plurality of medical procedures for the generation of one or more machine learning models.

4. When generating one or more machine learning models configured to predict one or more aspects associated with a medical procedure, the at least one processor, To provide one or more trained machine learning models, the one or more machine learning models are trained based on the formatted data associated with multiple medical procedures for the generation of the one or more machine learning models. The system according to claim 1, configured to validate the output of one or more trained machine learning models based on the data associated with the regulatory guidance and approval process.

5. The aforementioned at least one processor is When a request is received to access a specific machine learning model among multiple machine learning models, Determine whether the access request complies with one or more criteria associated with accessing the specific machine learning model. The system according to claim 1, further configured to provide access to the particular machine learning model based on the determination that the access request conforms to one or more criteria associated with access to the particular machine learning model.

6. The formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset of which is associated with a specific medical procedure among the multiple medical procedures, and the at least one processor, We receive a request to access a specific dataset among multiple datasets. Determine whether the access request complies with one or more criteria associated with accessing the specific dataset. The system according to claim 1, further configured to provide access to a particular dataset based on the determination that the access request conforms to one or more criteria associated with access to the particular dataset.

7. When providing access to the aforementioned specific dataset, the at least one processor, To provide access to the aforementioned specific dataset, the system executes application programming interface (API) calls to the data repository system related to the aforementioned specific dataset. The system according to claim 6, configured as follows.

8. The formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset of which is associated with a specific medical procedure among the multiple medical procedures, and the at least one processor, To provide one or more trained machine learning models, the one or more machine learning models are trained based on a first dataset among the multiple datasets, Based on having trained one or more machine learning models, assign a unique identifier to the first dataset. Assign the unique identifier to one or more trained machine learning models. Further configured to validate the output of one or more trained machine learning models based on the data associated with the regulatory guidance and approval process, When verifying the output of the one or more trained machine learning models, the at least one processor shall The system according to claim 1, configured to validate the first dataset and the one or more trained machine learning models based on the unique identifier assigned to the first dataset and the one or more trained machine learning models.

9. When deploying the one or more machine learning models to the production environment, the at least one processor is: Production environment of a third-party platform not associated with the aforementioned AI medical device platform, Production environment for the aforementioned AI medical device platform, Platforms that include standardized containers, or Any combination of them The system according to claim 1, configured to place one or more machine learning models in at least one of the following.

10. A method for generating an AI-based automated healthcare application, A step of using at least one processor of an AI medical device platform to receive formatted data associated with multiple medical procedures for generating one or more machine learning models from a data repository system, The steps include using the at least one processor to receive data associated with one or more regulatory guidance and approval processes, A step of using at least one processor to generate one or more machine learning models configured to predict one or more aspects associated with a medical procedure, based on the formatted data associated with a plurality of medical procedures for the generation of one or more machine learning models and the data associated with the regulatory guidance and approval process, The steps include: deploying the one or more machine learning models to the production environment based on the generation of the one or more machine learning models using the at least one processor; Methods that include...

11. Electronic Medical Record (EMR) system, Medical imaging systems, Fluid injection system, Pathology information system, Laboratory information system, Any combination of them Steps to receive initial data associated with one or more data records from multiple data sources, including at least one of the following: The method according to claim 10, further comprising:

12. A step of performing one or more data curation procedures on the initial data associated with one or more data records in order to provide the formatted data associated with multiple medical procedures for generating one or more machine learning models. The method according to claim 11, further comprising:

13. The step of generating one or more machine learning models configured to predict one or more aspects associated with a medical procedure is: The steps include training one or more machine learning models based on the formatted data associated with the plurality of medical procedures in order to provide one or more trained machine learning models, A step of verifying the output of one or more trained machine learning models based on the data associated with the regulatory guidance and approval process. The method according to claim 10, including the method described in claim 10.

14. The steps include receiving a request to access a specific machine learning model among multiple machine learning models, The steps include determining whether the access request conforms to one or more criteria associated with accessing the particular machine learning model, The steps include: providing access to the specific machine learning model based on the determination that the access request conforms to one or more criteria associated with access to the specific machine learning model; The method according to claim 10, further comprising:

15. The formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset of which is associated with a specific medical procedure among the multiple medical procedures, and the method is The steps include receiving a request to access a specific dataset among multiple datasets, The steps include determining whether the access request complies with one or more criteria associated with accessing the specific dataset, The steps include: providing access to the specific dataset based on the determination that the access request complies with one or more criteria associated with access to the specific dataset; The method according to claim 10, further comprising:

16. The step of providing access to the aforementioned specific dataset is: The step of making an application programming interface (API) call to the data repository system related to the specific dataset in order to provide access to the specific dataset. The method according to claim 15, including the method described in claim 15.

17. The formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset of which is associated with a specific medical procedure among the multiple medical procedures, and the method is To provide one or more trained machine learning models, the steps include: training one or more machine learning models based on a first dataset among the plurality of datasets; The steps include assigning a unique identifier to the first dataset based on having trained one or more machine learning models, The steps include assigning the unique identifier to one or more trained machine learning models, A step of verifying the output of one or more trained machine learning models based on the data associated with the regulatory guidance and approval process. The step of verifying the output of the one or more trained machine learning models further includes, A step of validating the first dataset and the one or more trained machine learning models based on the unique identifier assigned to the first dataset and the one or more trained machine learning models. The method according to claim 10, including the method described in claim 10.

18. The step of deploying one or more machine learning models to the production environment is: Production environment of a third-party platform not associated with the aforementioned AI medical device platform, Production environment for the aforementioned AI medical device platform, Platforms that include standardized containers, or Any combination of them The step of placing one or more machine learning models in at least one of the following locations. The method according to claim 10, including the method described in claim 10.

19. A computer program product for generating artificial intelligence (AI) based automated healthcare applications, comprising at least one non-temporary computer-readable medium containing program instructions, wherein, when the program instructions are executed by at least one processor, the at least one processor, The system receives formatted data associated with multiple medical procedures for the generation of one or more machine learning models from a data repository system. The data repository system receives one or more regulatory guidance and data associated with approval processes. Based on the data associated with the plurality of medical procedures and the data associated with the regulatory guidance and approval process, one or more machine learning models are generated that are configured to predict one or more aspects associated with the medical procedures. A computer program product that, based on the generation of one or more machine learning models, causes one or more machine learning models to be deployed in a production environment.

20. The program instruction further provides the at least one processor, Electronic Medical Record (EMR) system, Medical imaging systems, Fluid injection system, Pathology information system, Laboratory information system, Any combination of them The computer program product according to claim 19, which causes to receive initial data associated with one or more data records from a plurality of data sources, including at least one of the following.

21. The program instruction further provides the at least one processor, The computer program product according to claim 20, which causes one or more data curation procedures to be performed on the initial data associated with one or more data records in order to provide the formatted data associated with a plurality of medical procedures for the generation of one or more machine learning models.

22. The program instruction causing the at least one processor to generate one or more machine learning models configured to predict one or more aspects associated with a medical procedure, To provide one or more trained machine learning models, the one or more machine learning models are trained based on the formatted data associated with multiple medical procedures for the generation of the one or more machine learning models. The computer program product according to claim 19, which causes the output of one or more trained machine learning models to be validated based on the data associated with the regulatory guidance and approval process.

23. The program instruction further provides the at least one processor, It receives requests to access a specific machine learning model among multiple machine learning models. Determine whether the access request conforms to one or more criteria associated with accessing the specific machine learning model. The computer program product according to claim 19, which provides access to the particular machine learning model based on the determination that the access request conforms to one or more criteria associated with access to the particular machine learning model.

24. The formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset associated with a specific medical procedure among the multiple medical procedures, and the program instructions further provide the at least one processor, To receive access requests to a specific dataset among multiple datasets, Determine whether the access request complies with one or more criteria associated with accessing the specific dataset. The computer program product according to claim 19, which provides access to the specific dataset based on the determination that the access request conforms to one or more criteria associated with access to the specific dataset.

25. The program instruction that causes the at least one processor to provide access to the specific dataset is, The computer program product according to claim 24, which causes the system to make application programming interface (API) calls to the data repository system related to the specific dataset in order to provide access to the specific dataset.

26. The formatted data associated with multiple medical procedures for the generation of one or more machine learning models includes multiple datasets, each dataset associated with a specific medical procedure among the multiple medical procedures, and the program instructions further provide the at least one processor, To provide one or more trained machine learning models, the one or more machine learning models are trained based on a first dataset among the multiple datasets, Based on having trained one or more machine learning models, assign a unique identifier to the first dataset. The one or more trained machine learning models are assigned the unique identifier. The output of one or more trained machine learning models is validated based on the regulatory guidance and the data associated with the approval process. The program instruction causing the at least one processor to verify the output of the one or more trained machine learning models is, The computer program product according to claim 19, which causes the first dataset and the one or more trained machine learning models to be validated based on the unique identifier assigned to the first dataset and the one or more trained machine learning models.

27. The program instruction that causes the at least one processor to deploy the one or more machine learning models to the production environment is, Production environments for third-party platforms not associated with the AI ​​medical device platform, Production environment for the aforementioned AI medical device platform, Platforms that include standardized containers, or Any combination of them The computer program product according to claim 19, wherein at least one of the above is configured to house one or more machine learning models.

Images