Healthcare Network
By maintaining directed graphs and patient models of medical guidelines in the database, identifying treatment-deviated patient groups and generating new directed graphs, the problem that existing medical guidelines cannot fully cover individual differences is solved, and more effective treatment is achieved Path guidance and personalized medical care.
Patent Information
- Application Number
- CN202011042329.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2019-09-30
- Filing Date
- 2020-09-28
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2040-09-28
AI Technical Summary
Existing medical guidelines may not fully cover individual differences in patients during diagnosis and treatment, leading to treatment deviations and lack of effective analytical tools to identify and deal with these deviations.
A system was designed that provides customized treatment pathways by maintaining directed graphs and patient models of medical guidelines in a database, identifying treatment deviated groups and generating new directed graphs to guide decisions. The system uses processor and computer program code to combine patient attributes and clinical step data to identify patient group characteristics and generate directed graphs based on these characteristics to guide medical practitioners to provide more effective treatment plans.
The system can effectively identify patient groups that deviate from medical guidelines, generate customized treatment paths, improve the individualization and effectiveness of treatment, and help medical practitioners provide more suitable health care for specific patient groups.
Smart Images

Figure CN112582071B_ABST
Abstract
Description
Technical Field
[0001] Embodiments described herein generally relate to providing healthcare data to a user device. More specifically, embodiments relate to methods, systems, and computer programs for sending healthcare data to a user device that is configured to analyze medical information. Background Art
[0002] Medical guidelines provide recommendations on how people with specific medical conditions should be treated. Medical guidelines may indicate which diagnostic or therapeutic steps should be taken when treating patients with specific conditions and which follow-up procedures should be performed based on the results of the diagnostic or therapeutic steps. Some medical guidelines provide information about the prevention, prognosis, and risks and / or benefits of certain medical conditions and consider the cost-effectiveness associated with diagnostic and therapeutic steps in the treatment of patients. The information contained in guidelines is often specific to a particular medical field.
[0003] Data related to a patient being treated for a medical condition is often generated during the diagnostic and therapy steps. Such data is often stored in different sources related to the location where the data was generated, such as a clinical center or hospital. Data related to the patient can be encoded to associate the raw data or values with the various clinical steps that generated the data. Clinical coding systems such as SNOMED CT, LOINC, Internal coding systems as well as other coding systems encode the data.
[0004] Patient conditions and diseases do not always conform to the recommendations and clinical pathways provided in medical guidelines. Clinical pathways - also known as disease pathways - can include secondary prevention, screening, diagnostics, diagnosis, therapy decisions, therapy, and subsequent treatment or decisions. Therefore, medical guidelines alone may not always be sufficient to enable adequate analysis.
[0005] Systems for transmitting healthcare data received by a user device are described in European patent application EP18199915 filed on October 11, 2018 and European patent application EP18208021 filed on November 23, 2018, both of which are incorporated herein by reference. Summary of the invention
[0006] According to a first aspect of the present invention, there is provided a system operable to send healthcare data to a user device, the user device being configured to analyze medical information, the system comprising at least one processor and at least one memory comprising computer program code, the at least one memory and the computer program code being configured to, together with the at least one processor, cause the system to: maintain data representing a first directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the first directed graph comprising a plurality of elements representing clinical steps; maintain a plurality of patient models in a second database, each comprising healthcare data associated with a corresponding patient; and select at least one element from the plurality of elements by processing the plurality of patient models and the data representing the first directed graph. factors to identify at least one element where treatment of a subset of patients deviates from at least a portion of the medical guidelines; identifying a first patient group and a second patient group based on a combination of the selected at least one element and a plurality of patient models, wherein treatment of the first patient group deviates from at least a portion of the medical guidelines at the selected at least one element, and treatment of the second patient group complies with at least a portion of the medical guidelines at the selected at least one element; processing a plurality of patient models representing the first patient group and the second patient group to identify at least one patient group characteristic that distinguishes the first patient group from the second patient group; generating a second directed graph based at least on the identified at least one patient group characteristic; and transmitting data representing the second directed graph for receipt by a user device.
[0007] In this way, the system is operable to identify groups of patients who may not be following the treatment specified in at least a portion of the medical guidelines. In these cases, the system can then generate a second directed graph that includes an indication that, for the identified group of patients, the medical guidelines were deviated from. This is used to guide the decision-making of medical practitioners so that they can provide more effective treatment for patients belonging to the following specific group of patients for whom the medical guidelines may not be effective if followed.
[0008] Selecting at least one element may include processing the plurality of patient models and the data representing the first directed graph to identify at least one element at which a subset of patients whose treatment deviates from at least a portion of the medical guideline exceeds a predetermined proportion of patients associated with the plurality of patient models. This may provide appropriate selection of the at least one element in a variety of situations where the total number of available patient models is different.
[0009] The at least one patient cohort characteristic may include at least one of: age; height; weight; gender; body mass index; genetic mutation; associated medical practitioner; and location.
[0010] The memory and computer program code may be configured to, with the at least one processor, cause the system to override selection of at least one element based on user input. In this manner, a user of the system, such as a medical practitioner, may identify elements for which they suspect a deviation may occur for a patient group or for which they suspect a deviation from at least a portion of a medical guideline may be beneficial for a certain patient group.
[0011] The selected at least one element may be associated with at least one condition parameter value, the patient models may each include a plurality of patient attribute values corresponding to respective clinical steps, and the first patient group and the second patient group may be identified based on a comparison of the at least one condition parameter value with respective patient attribute values from the plurality of patient models. This may, for example, provide an efficient and robust way of automatically identifying the first patient group and the second patient group.
[0012] The first patient group and the second patient group can be identified based on the availability of healthcare data in the corresponding patient model corresponding to the selected at least one element. This can allow the system to quickly identify whether the patient meets at least a portion of the medical guidelines.
[0013] Processing a plurality of patient models representing a first patient group and a second patient group to determine at least one patient group characteristic that distinguishes the first patient group from the second patient group may include processing the plurality of patient models using at least one of: principal component analysis; random forest regression; and regularized regression. These methods may provide an efficient and accurate way to identify a patient group characteristic that distinguishes a large number of patient characteristics associated with each patient.
[0014] The second directed graph may include an indication that the clinical step represented by the selected at least one element is not recommended for patients associated with the identified at least one patient group characteristic. This may enable medical practitioners providing treatment to patients based on the second directed graph to be aware of ways in which they can provide more effective treatment to their patients. This may, for example, enable physicians who have no experience treating patients belonging to a particular patient group to provide customized medical care that may not be covered in the associated medical guidelines.
[0015] The second directed graph may include: a plurality of nodes; a set of directed edges; and at least one additional node connected to the selected element, wherein the at least one additional node may include an indication that the clinical step represented by the selected element is not recommended for patients associated with the identified at least one patient group characteristic. In this way, medical practitioners may be made aware of possible deviations in treatment for patients belonging to a particular patient group.
[0016] The second directed graph may include at least one further directed edge connected at a first end to a further node, the further directed edge may indicate at least a partial deviation from the medical guideline for the first patient group.
[0017] The at least one memory and the computer program code may be configured to, with the at least one processor, cause the system to: modify the second directed graph based on user input indicating a decision regarding a patient associated with the identified at least one patient group characteristic. This may allow the system to automatically generate a different treatment pathway including additional clinical steps in the second directed graph based on medical decisions made regarding patients belonging to the first patient group. This may identify new treatment pathways and / or beneficial treatment pathways for a particular patient group that have not been specified in medical guidelines to date.
[0018] At least one memory and computer program code can be configured to, together with at least one processor, cause the system to: maintain in a third database an additional patient model comprising healthcare data associated with a patient, the patient being associated with at least one identified patient group characteristic; determine, based on a combination of the second directed graph and the additional patient model, a state of at least one additional node and / or a state of a directed edge connected to the at least one additional node; and, based on the state of the at least one additional node and / or the state of the directed edge connected to the at least one additional node, send data indicating the state of the at least one additional node and / or the state of the directed edge connected to the at least one additional node for receipt by a user device; the data indicating the state of the at least one additional node and / or the state of the directed edge connected to the at least one additional node is used to determine whether the clinical step represented by the selected node is not recommended for the patient.
[0019] The system may automatically map an additional patient model to the second directed graph and may alert the user if the treatment of a patient represented by the additional patient model is at or near a node where treatment of patients known to belong to a particular cohort deviates.
[0020] At least one memory and computer program code may be configured to, together with at least one processor, cause the system to: send data indicating an input request from a user of a user device for receipt by the user device based on the state of another node and / or the state of a directed edge connected to the other node; receive data indicating another clinical step to be represented by another plurality of nodes from the user device; and modify the second directed graph based on the received data indicating the other clinical step. Once a deviation from at least a portion of the medical guidelines for a particular patient group has been identified, the system may be operable to: generate new nodes representing alternative clinical steps specified by a medical practitioner after the deviation. Physicians may manually change the second directed graph to reflect their preferred treatment path after the deviation.
[0021] According to a second aspect of the present invention, there is provided a computer program comprising a set of instructions, which when executed by a computerized device causes the computerized device to perform a method of sending healthcare data to a user device, the user device being configured to analyze medical information, the method comprising: maintaining data representing a first directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the first directed graph comprising a plurality of elements representing clinical steps; maintaining a plurality of patient models in a second database, each comprising healthcare data associated with a corresponding patient; selecting at least one element from the plurality of elements by processing the plurality of patient models and the data representing the first directed graph to identify treatment deviations for a subset of patients at least one element of at least a portion of a medical guideline; identifying a first patient group and a second patient group based on a combination of the selected at least one element and a plurality of patient models, wherein treatment of the first patient group deviates from at least a portion of the medical guideline at the at least one selected element, and treatment of the second patient group complies with at least a portion of the medical guideline at the at least one selected element; processing a plurality of patient models representing the first patient group and the second patient group to identify at least one patient group characteristic that distinguishes the first patient group from the second patient group; generating a second directed graph based at least on the identified at least one patient group characteristic; and transmitting data representing the second directed graph for receipt by a user device.
[0022] According to a third aspect of the present invention, there is provided a method for sending healthcare data to a user device, the user device being configured to analyze medical information, the method comprising: maintaining data representing a first directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the first directed graph comprising a plurality of elements representing clinical steps; maintaining a plurality of patient models in a second database, each comprising healthcare data associated with a corresponding patient; selecting at least one element from the plurality of elements by processing the plurality of patient models and the data representing the first directed graph to identify at least one element at which treatment of a subset of patients deviates from at least a portion of the medical guideline; identifying a first patient group and a second patient group based on a combination of the selected at least one element and the plurality of patient models, the treatment of the first patient group deviating from at least a portion of the medical guideline at the selected at least one element, and the treatment of the second patient group complying with at least a portion of the medical guideline at the selected at least one element; processing the plurality of patient models representing the first patient group and the second patient group to identify at least one patient group characteristic that distinguishes the first patient group from the second patient group; generating a second directed graph based at least on the identified at least one patient group characteristic; and sending the data representing the second directed graph for reception by the user device.
[0023] According to a fourth aspect of the present invention, there is provided a system operable to send healthcare data to a user device, the user device being configured to analyze medical information, the system comprising at least one processor and at least one memory comprising computer program code, the at least one memory and the computer program code being configured to, together with the at least one processor, cause the system to: maintain data representing a first directed graph and a second directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the second directed graph representing at least a portion of the medical guideline and modifications to at least a portion of the medical guideline, each directed graph comprising a corresponding plurality of elements representing clinical steps; maintain a plurality of patient models, each comprising healthcare data associated with a corresponding patient, in a second database; identify a first group of patient models and a second group of patient models, the first group of patient models representing patients who have been treated based on at least a portion of the medical guideline represented by the first directed graph, the second group of patient models representing patients who have been treated based on at least a portion of the medical guideline represented by the second directed graph; determine which of the first directed graph and the second directed graph is a preferred directed graph based on a comparison of the first group of patient models with the second group of patient models; and in response to the determination, send data representing the preferred directed graph for reception by the user device.
[0024] Directed graphs can be generated in a variety of ways, and in some cases, a customized directed graph including changes to the treatment path can be automatically generated based on the patient's past treatment or manually generated, for example, based on the medical practitioner's assumptions. This aspect of the invention can allow the effectiveness of such directed graphs in the treatment of patients to be compared. Then, the patient's future treatment can use the preferred directed graph to provide medical care to the patient.
[0025] The modification to at least a portion of the medical guideline is represented by at least one of: at least one node in the second directed graph; and at least one directed edge in the second directed graph. This can, for example, provide differences between the first directed graph and the second directed graph that can be directly compared using the patient model.
[0026] The plurality of patient models may each include a plurality of patient entries, at least one of the patient entries may include at least one patient outcome measure, and determining which of the first directed graph and the second directed graph is a preferred directed graph may include comparing a first plurality of patient outcome measures of the first set of patient models with a second plurality of patient outcome measures of the second set of patient models. Using the patient outcome measures to compare the effectiveness of treatment provided based on the first directed graph and the second directed graph may ensure that the directed graph selected as the preferred directed graph provides better patient outcomes than the other directed graphs when used to treat the patient.
[0027] Modification of at least a portion of the medical guideline can be associated with a patient group characteristic, and a first set of patient models and a second set of patient models can be identified based on the patient group characteristic. In this way, the validity of a directed graph generated based on deviations with respect to a specific patient group as described above with respect to the first aspect of the present invention can be tested.
[0028] The at least one memory and the computer program code may be configured to, with the at least one processor, cause the system to identify a first set of patient models based on comparing the plurality of patient models to the first directed graph, wherein comparing the plurality of patient models to the first directed graph may include, for each patient model, determining a state of at least one of the plurality of elements of the first directed graph.
[0029] The at least one memory and the computer program code may be configured to, with the at least one processor, cause the system to identify a second group of patient models based on a comparison of the plurality of patient models to the second directed graph, wherein comparing the plurality of patient models to the second directed graph may include, for each patient model, determining a state of at least one of the plurality of elements of the second directed graph. This may provide a robust and reliable way of identifying patients represented by the second group of patients based on available data. This may alleviate the need to manually select which patients have been treated according to which guideline.
[0030] For the patient model, the state of the element may depend on the availability of data in the patient model associated with the clinical step that is associated with the element. If a patient has not undergone a particular clinical step, then their associated patient model will not include data associated with the clinical step. This allows such patients to be efficiently identified.
[0031] Each patient model may include multiple patient entries, each patient entry includes at least one attribute value, and determining the state of the element may include: maintaining a first association between at least one of the patient entries of the patient model and an identifier from a plurality of identifiers; maintaining a second association between the element and an identifier from a plurality of identifiers; selecting the attribute value associated with the element based on the first association and the second association; and determining whether the conditional parameter value associated with the element is satisfied based on a comparison of the attribute value associated with the element and the conditional parameter value associated with the element.
[0032] Determining which of the directed graphs is a preferred directed graph may include: determining a first measure indicating average consistency of a first group of patient models with the first directed graph; determining a second measure indicating average consistency of a second group of patient models with the second directed graph; and using the first measure and the second measure to perform a comparison of a first plurality of patient outcome measures with a second plurality of patient outcome measures. In this way, the selection of a preferred directed graph may be sensitive to the patient's general adherence to the directed graph. For example, while a given directed graph may have an average or good outcome measure, general adherence to a second directed graph may be poor, and thus treatment according to the directed graph may be more unpredictable and / or more variable. By taking this into account when selecting a directed graph as a preferred directed graph, one can control this characteristic to a certain extent.
[0033] The first measure may depend on the average state of the first set of patient models of a plurality of elements of the first directed graph. This may provide a measure of average compliance which may be used as a variable in the selection of a preferred directed graph.
[0034] Similarly, the second measure may depend on an average state of the second set of patient models over the plurality of elements of the second directed graph.
[0035] Determining which of the first directed graph and the second directed graph is the preferred directed graph may include: sending data indicating the results of the comparison of the first group of patient models with the second group of patient models for reception by a user device; receiving data indicating a decision regarding the results of the comparison from the user device; and selecting one of the first directed graph or the second directed graph based on the received data indicating the decision. In this way, the user is able to select a preferred directed graph based on a comparison of the outcome measures. In some examples, such as where the variables are interrelated and / or nonlinear or where some variables are more important than other variables, comparing the outcome measures may not be a simple process of comparing one set of variables to another set of variables. In this case, presenting the results of the comparison to the user can enable the user to resolve the selection of the preferred directed graph.
[0036] The at least one memory and the computer program code may be configured to, with the at least one processor, cause the system to: if the first directed graph is the preferred directed graph, send data indicating the first plurality of patient outcome measures for reception by a user device; or if the second directed graph is the preferred directed graph, send data indicating the second plurality of patient outcome measures for reception by the user device. In this manner, supplemental medical information may be provided to a user of the user device to support a decision regarding the use of the preferred directed graph for use in, for example, discussing with a patient why a particular treatment pathway is being used and / or demonstrating to a medical practitioner that the preferred directed graph enables better treatment of their patients.
[0037] According to a fifth aspect of the present invention, there is provided a computer program comprising a set of instructions, which, when executed by a computerized device, causes the computerized device to execute a method for sending health care data to a user device, the user device being configured to analyze medical information, the method comprising: maintaining data representing a first directed graph and a second directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the second directed graph representing at least a portion of the medical guideline and modifications to at least a portion of the medical guideline, each directed graph comprising a corresponding plurality of elements representing clinical steps; maintaining a plurality of patient models in a second database, each comprising health care data associated with a corresponding patient; identifying a first group of patient models and a second group of patient models, the first group of patient models representing patients who have been treated based on at least a portion of the medical guideline represented by the first directed graph, the second group of patient models representing patients who have been treated based on at least a portion of the medical guideline represented by the second directed graph; determining which of the first directed graph and the second directed graph is a preferred directed graph based on a comparison of the first group of patient models with the second group of patient models; and in response to the determination, sending data representing the preferred directed graph for reception by the user device.
[0038] According to a sixth aspect of the present invention, there is provided a method for sending healthcare data to a user device, the user device being configured to analyze medical information, the method comprising: maintaining data representing a first directed graph and a second directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the second directed graph representing at least a portion of the medical guideline and modifications to at least a portion of the medical guideline, each directed graph comprising a corresponding plurality of elements representing clinical steps; maintaining a plurality of patient models in a second database, each comprising healthcare data associated with a corresponding patient; identifying a first group of patient models and a second group of patient models, the first group of patient models representing patients who have been treated based on at least a portion of the medical guideline represented by the first directed graph, the second group of patient models representing patients who have been treated based on at least a portion of the medical guideline represented by the second directed graph; determining which of the first directed graph and the second directed graph is a preferred directed graph based on a comparison of the first group of patient models with the second group of patient models; and in response to the determination, sending data representing the preferred directed graph for reception by the user device. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1a shows a schematic block diagram of an example system according to an embodiment;
[0040] Figure 1b A schematic block diagram showing an example system connected to a network according to an embodiment;
[0041] Figure 2 shows an example of a directed graph according to an embodiment;
[0042] Figure 3 A schematic block diagram of an event model according to an embodiment is shown;
[0043] Figure 4 shows a schematic block diagram of a patient model according to an embodiment;
[0044] Figure 5 A flow chart illustrating the operation of a system according to an embodiment;
[0045] FIG. 6A to FIG. 6C An example of a directed graph including instructions according to an embodiment is shown;
[0046] 7A to 7C a diagram showing a patient model in a feature space to graphically represent the determination of at least one patient group characteristic according to an embodiment; and
[0047] Figure 8 A flow chart illustrating the operation of a system according to an embodiment. DETAILED DESCRIPTION
[0048] Embodiments will now be described in the context of systems, methods, and computer programs for providing information to a user of a user device, the device being configured for analyzing medical information. Reference will be made to the accompanying drawings. In the following description, many specific details of certain examples are set forth for illustrative purposes. References in the specification to "examples" or similar language mean that a particular feature, structure, or characteristic described in conjunction with the example is included in at least one example, but not necessarily in other examples. It should also be noted that certain examples are schematically described, and certain features are omitted and / or necessarily simplified in order to facilitate illustration and understanding of the concepts underlying the examples.
[0049] Figure 1a A diagram of a system 100 according to an example is shown. The term "system" may refer to any combination of hardware, computer program code, functionality, and virtualized resources embodied in a single device or across multiple devices. For example, system 100 may include a single device housed at one location, such as a computer in a hospital; or system 100 may include multiple devices housed at the one location connected by a local area network, such as at least one other computing device communicatively coupled to a mainframe computer in a hospital. Because multiple devices must fail for a system to become abnormal, a system including multiple computing devices may be more secure than a system including a single computing device. It may be more effective to use a system including multiple connected computing devices that are remotely located from each other, rather than having multiple systems.
[0050] In other examples, the system 100 may include multiple remote devices. The multiple remote devices may be connected via a metropolitan area network, a campus area network, or a wide area network such as the Internet. The system 100 may include multiple servers and / or mainframe computing devices distributed in hospitals within a country. In other examples, the system 100 may be distributed across multiple countries.
[0051] The system 100 includes at least one processor 102a to 102n. The at least one processor 102a to 102n may be a standard central processing unit or a graphics processing unit (CPU or GPU), or the at least one processor 102a to 102n may be a custom processing unit designed for the purposes described herein. Each of the at least one processor 102a to 102n may include a single processing core or multiple cores, such as quad-cores or octa-cores. In an example, where the system 100 includes multiple processors 102a to 102n, the processors 102a to 102n may be embodied in a single device. In other examples, the at least one processor 102a to 102n may include multiple processors remotely distributed in the system 100.
[0052] Figure 1aThe system 100 shown in the embodiment includes at least one memory 104a to 104n. The at least one memory 104a to 104n can be a non-transient computer-readable memory, such as a hard drive, a CD-ROM disk, a USB drive, a solid-state drive, or any other form of magnetic storage device, optical storage device, or flash memory device. The at least one memory 104a to 104n can be referred to as a storage medium or a non-transient computer-readable storage medium. The at least one memory 104a to 104n can be maintained locally in the rest of the system 100, or can be accessed remotely, for example, via the Internet. The at least one memory 104a to 104n can store computer program code suitable for the functions described herein. The computer program code can be distributed on multiple memories or can be stored on a single memory.
[0053] In the example, system 100 refers to a system operable to send healthcare data to a user device. Users analyzing medical information, such as physicians, nurses, clinical assistants, and other personnel analyzing medical information about patients, can operate system 100 via the user device. Medical information can relate to data, such as digital test results from diagnostic tests. Medical information can also relate to qualitative diagnoses and annotations related to one or more patients.
[0054] Figure 1aFour examples of user devices 106a to 106d are shown. The user device can be any combination of hardware and computer program code that can be operated by a user and is suitable for the functions described herein. User device 106a is a tablet computer, user device 106b is a smart phone, user device 106c is a smart watch, and user device 106d is a personal computer such as a desktop PC or a laptop PC. It should be understood that other examples of user devices can also be expected. User devices 106a to 106d can include any number of volatile or non-volatile memories, processors and other electronic components. In some examples, user devices 106a to 106d include multiple components distributed on a network. User devices 106a to 106n can include any number of output terminals, such as displays, speakers, tactile feedback systems, LED indicators, transmitters, or any other output terminals. User devices 106a to 106d can include any number of input terminals, such as microphones, buttons, cameras, receivers, or any number of sensors, etc. In some examples, the inputs and outputs of the user devices 106a to 106d may be considered as user interfaces, such as a touch screen or a combination of a screen and a keyboard. The user devices 106a to 106d may be considered as part of the system 100, or may not be part of the system 100 but may communicate with the system 100. In some examples, the user devices 106a to 106d are local to the system 100 and may be connected to the system 100 via a local area network, such as a personal computer 106d in a hospital connected to a mainframe computer in the same hospital that includes the system 100. In other examples, the user devices 106a to 106d may be connected to the system 100 via a wide area network.
[0055] The user devices 106a to 106d may be configured to analyze medical information. For example, the user devices 106a to 106d may include an application for presenting medical information to a user for analysis. In some examples, there are multiple user devices 106a to 106d. Multiple user devices 106a to 106d may be communicatively coupled to each other. At least one user device 106a to 106d may be used to control the system 100.
[0056] In some examples, user devices 106a-106d may be proprietary devices configured to be used in or with system 100. For example, user devices 106a-106d may be proprietary computing devices that include a combination of firmware, software, and custom applications for providing data and / or other information to a user. For example, user devices 106a-106d may include an application for displaying data received by system 100 in a predetermined manner and / or sent to system 100 in a predetermined manner.
[0057] In other examples, the user devices 106a-106d and the system 100 are included in the same device, such as a desktop computer at a hospital.
[0058] In other examples, the user devices 106a-106d may be commercially available computing devices that include any number of applications operable to access data at, receive data from, or send data to the system 100. For example, the system 100 may maintain at least one web page hosted on a remotely accessible server. The user devices 106a-106d may include a web browser operable to access at least one web page to facilitate communication with the system 100 and / or display data stored by the system 100 on the user devices 106a-106d.
[0059] According to the embodiments described herein, Figure 1a The system 100 shown in includes a database 108 for storing data. The database 108 can be any structured data set stored in a computing device. For example, the database 108 can be structured data stored in at least one memory 104a to 104n. In other examples, the database 108 can be stored elsewhere in the system, for example, on a separate computing device. At least one processor 102a to 102n can be communicatively coupled to the database 108 so that at least one processor 102a to 102n can maintain the database 108. Maintaining the database 108 can include sending data to the database 108, receiving data from the database 108, or reconfiguring the data in the database 108. In an example where the database 108 is stored on a physical memory associated with the system 100, at least one processor 102a to 102n can be configured to read data from the physical memory and / or write data to the physical memory to maintain the database 108. The database 108 may include multiple databases associated with each other. The database 108 can be embodied by any suitable data structure.
[0060] In some examples, the system may be able to communicate with other systems or remote data sources. Figure 1b In the example shown, the system 100 is connected to at least one remote computing device 112a to 112c via a network 110. For example, the system 100 may include a plurality of computers and servers located at a hospital, and the system 100 may communicate with a computing device 112a, which may alternatively be referred to as a computer system, at another hospital via the network 110 to send and / or receive medical data.
[0061] The system 100 may simultaneously communicate with a computing device 112b representing a medical guideline repository storing at least one medical guideline. The medical guideline repository may be embodied in any device storing at least one medical guideline. In some examples, the medical guideline repository may include a remotely accessible database storing at least one guideline, wherein the guideline is stored in a digital format and includes metadata identifying the at least one guideline. For example, the medical guideline may be stored as a PDF, and the metadata may include: an indication of the name of the medical guideline, a date of publication, an indication of the disease to which the medical guideline relates, a country in which the medical guideline was first published or a country in which the medical guideline is designed to apply, and any other information that may be used to identify the guideline. More examples of identifying features of medical guides and their uses will be described later.
[0062] The system may also communicate with a computing device 112c acting as a control device to control the operation of the system 100. For example, a remote computing device such as a personal computer or a server may be used by an administrator to control the system 100.
[0063] Healthcare data as described herein may include data related to: patient records (e.g., results from diagnostic tests or therapy steps), medical guidelines (e.g., directed graphs as will be discussed below), statistics, medical research, scientific articles, or any other medical or clinically relevant information. Healthcare data may be stored in any number of digital formats, and the digital format in which the healthcare data is stored may depend on the type of healthcare data. For example, raw data associated with a diagnostic result may be stored as a plain text file, a CSV file, or any other suitable file format. Some types of healthcare data may be stored in a human-readable format, or alternatively, may be stored in a computer-interpretable language. In some examples, system 100 may send healthcare data to user devices 106a to 106d in a computer-interpretable format, and user devices 106a to 106d may process the data and present the data to the user in a human-readable format.
[0064] Medical guidelines can define clinical pathways for treating patients with medical conditions. In some examples, medical guidelines are divided into a series of treatment stages. Examples of treatment stages can include: staging, initial treatment, active monitoring, recurrence treatment, and others. Clinical pathways can be defined by a series of clinical steps, wherein the selection of which clinical step to be performed at a given time depends at least on the results from at least one previous clinical step. Clinical steps can include observations, decisions, events, diagnostic tests or therapeutic treatments to be given to patients with medical conditions. Medical guidelines can be printed or published online in digital formats such as PDF. Medical guidelines can contain evidence and / or consensus-based recommendations about medical treatment pathways. Medical guidelines can also contain instructions and / or reasons for the clinical pathways defined in the corresponding medical guidelines.
[0065] In the examples described herein, the system 100 maintains in the database 108 a medical guideline represented by at least one directed graph. In some examples, a set of directed graphs are used to represent the medical guideline, for example, where each directed graph represents a treatment stage within the medical guideline. In the discussion that follows, reference may be made to the medical guideline represented by a directed graph. However, it is recognized that the medical guideline may be represented by a set of directed graphs, and reference to a directed graph representing a medical guideline may refer to a directed graph representing at least a portion of the medical guideline. In the example, a set of directed graphs are connected to each other to form a representation of the medical guideline. The system 100 may be pre-installed with a machine-interpretable representation of the latest medical guideline. The machine-interpretable representation of the medical guideline may be provided in the system 100. Figure 2. The directed graph 200 includes a plurality of elements. The plurality of elements include a plurality of nodes 202a to 2021. The nodes 202a to 2021 may represent clinical steps, and in some examples, each of the plurality of nodes 202a to 2021 represents a clinical step described in a corresponding at least a portion of a medical guideline represented by the directed graph 200. The directed graph 200 also includes a set of directed edges 204a to 2041. Each of the plurality of nodes 202a to 2021 is connected to at least one other node of the plurality of nodes 202a to 2021 by at least one of the set of directed edges 204a to 2041. The plurality of nodes 202a to 2021 and the set of directed edges 204a to 2041 may be collectively referred to as elements of the directed graph. The plurality of nodes 202a to 2021 may define a series of diagnostic tests and medical treatments that are recommended to be performed on a patient with a specific medical condition. Nodes 202a to 2021 can also define observation points and / or decisions that occur along the treatment path. Directed edges 204a to 204l can define the direction and / or order in which clinical steps will be executed when treating patients with specific medical conditions. In some examples, directed edges define the following conditions, under which the patient will move from experiencing a specific clinical step represented by a node such as node 202c to experiencing different clinical steps represented by another node such as node 202d. Each node can represent the conditional parameter value obtained from the clinical step associated with the node. Alternatively, each of the group of directed edges can represent the conditional parameter value obtained from the clinical step, and the clinical step is associated with one of the multiple nodes connected thereto. In some examples, at least one of the multiple groups of directed edges can specify the user input to be received before traversing from one node to another node. The directed graph can also include another group of nodes that are not associated with the clinical step.
[0066] In an example, a directed graph is maintained as at least one list in JSON format, the at least one list including unique identifiers of nodes 202a to 2021 and directed edges 204a to 2041 representing the directed graph. The entries in the at least one list may be linked to definitions or references in the corresponding guidelines. In some examples, the link includes other information, such as weighting based on consensus. The entries representing directed edges 204a to 204l may also include associations or links with nodes 202a to 202l, and nodes 202a to 202l are connected to the directed edges in the corresponding directed graph. The entries representing directed edges may also include information about the direction from the first node connected to the directed edge to the second node connected to the directed edge. For example, the entry representing directed edge 204c may include an indication that directed edge 204c is connected to nodes 202b and 202e, and the direction along directed edge 204c is from 202b to 202e. The directed graph may include two layers of elements that allow for modifications at the clinic and / or clinical site without losing information associated with the original directed graph.
[0067] Additional information related to the directed graph may be maintained in the event model. Figure 3 An example of an event model for a directed graph is shown. The event model 300 includes a list of entries 302a to 302n representing events or steps linked to corresponding nodes in the directed graph. The entries 302a to 302n may include any of the following: a unique identifier 304a to 304n; a code 306a to 306n in a medical coding system; a label 308a to 308n; a type of step 310a to 310n (e.g., biopsy); a required patient attribute input 312a to 312n; a required patient attribute output 314a to 314n; and annotations related to the impact 316a to 316n of the event, the effectiveness 318a to 318n of the event, the cost 320a to 320n of the event, the duration 322a to 322n of the event, the invasiveness 324a to 324n of the event, or any other relevant information. This allows important information contained in the medical guide to be stored in an efficient manner that allows the use of the medical guide as described herein. The event model 300 includes an identifier 326 that associates the event model with a corresponding medical guideline. Figure 3 Only data about entry 302a is shown.
[0068] In some examples described herein, the system 100 maintains a plurality of patient models in the database 108, each of the plurality of patient models including healthcare data associated with a corresponding patient. For example, a patient model may include any of the following: data generated during a clinical procedure, such as diagnostic and / or therapeutic steps performed on a patient; reasons for performing a clinical procedure on a patient, for example, particularly where the selection of the clinical procedure deviates from medical guidelines; routine data associated with the patient, such as age, sex, height, body mass index; known conditions, risk factors, patient identifiers; genetic information associated with the patient; or any other information for classifying a patient. In some examples, the patient models may each be stored as a list including a plurality of patient entries. Each patient entry may include data associated with patient attributes. Figure 4 An example of a patient model 400 is shown that includes a list of patient entries 402a-402n, a patient identifier 404, and in some cases other patient data 406. The patient entries 402a-402n may include any of the following: a unique identifier 408a-408n; an encoded identifier 410a-410n, such as an identifier in a medical coding system; a natural language label 412a-412n; a type of clinical step 414a-414n (e.g., biopsy, scan, physical assessment, etc.); a measurement unit 416a-416n; and patient attribute values 418a-418n (e.g., results from a test). The patient entries may include an association between the encoded identifiers 410a-410n and the patient attribute values 418a-418n, also referred to as attribute values. For clarity, the patient entries 402a-402n may include any of the following: a unique identifier 408a-408n; an encoded identifier 410a-410n, such as an identifier in a medical coding system; a natural language label 412a-412n; a type of clinical step 414a-414n (e.g., biopsy, scan, physical assessment, etc.); a measurement unit 416a-416n; and a patient attribute value 418a-418n (e.g., a result from a test). Figure 4 In the illustrated example, only data in the patient entry 402a is shown. The encoded identifiers may identify the clinical steps to which the patient attribute values of the corresponding patient entry relate, wherein each clinical step is associated with a corresponding encoded identifier.
[0069] In some examples, multiple patient models associated with a patient are stored and / or maintained in a central database 108 or computing device. The patient models may be stored in the same database as the data representing the directed graph, or may be stored in a separate database. The patient models may be accessed via the system 100 via a network connection. In other examples, the patient models may be stored locally at a clinical center, such as a physician's operating room or hospital where the patient has been treated in the past or is currently being treated. For example, a patient model stored in one hospital in the system 100 may be accessed at a remote location via the system 100 via the network 110.
[0070] In some examples, the system 100 updates the patient model by accessing remote computing devices 112a to 112c via the network 110. For example, the system 100 can access a medical test device storing data related to the patient through the network 110 to update the corresponding patient model. The system 100 can access a server or other computing device storing medical information related to the patient in the hospital. In some examples, the system 100 can continuously and / or periodically collect data about the patient from various hospital information systems. Data can be collected at predetermined time intervals, such as every hour, every day, etc. The size of the time interval can depend on the size of the system 100 or the clinical center. In some examples, data collection can be triggered by other events and / or messages occurring in the system 100. The retrieval of data about the patient can be based on existing standards, such as HL7, DICOM, FHIR. In other examples, the retrieval of data about the patient can be performed by using proprietary information about the information storage available in the hospital. In some examples, the retrieval of patient data is performed by receiving or accessing data from an external source, including coded identifiers, such as SNOMED CT, LOINC, or Siemens internal coding system. In other examples, the system 100 can use natural language processing techniques to extract information from electronically stored patient-related notes and files. In some examples, a combination of both techniques can be used. The patient model can be represented in the system 100 as a set of resources that conform to the FHIR standard and are stored in a FHIR server.
[0071] Will refer to Figure 1a and Figure 1b The following description of embodiments of the present invention is described with reference to an example system 100 of FIG. Figure 1a and Figure 1b The following implementation methods are implemented in different systems.
[0072] In an embodiment, at least one memory 104a to 104n includes computer program code, and at least one memory 104a to 104n and the computer program code are configured to: together with at least one processor 102a to 102n, cause the system 100 to execute Figure 5The steps indicated by each box of the flowchart in . Wherein, at box 502, the system 100 maintains data representing a first directed graph in the first database 108, the first directed graph representing at least a portion of the medical guideline, the first directed graph including a plurality of elements representing clinical steps. In some implementations, the plurality of elements include a plurality of nodes each representing a clinical step and a set of directed edges, each of the plurality of nodes being connected to at least one other node via one of the set of directed edges, the first directed graph including a main node and terminating at at least one end node. At box 504, the system 100 maintains a plurality of patient models in the second database 108, the plurality of patient models each including healthcare data associated with a corresponding patient. The first database and the second database may be the same or different databases.
[0073] At block 506, the system 100 selects at least one element from the plurality of elements, such as nodes 202a to 2021, directed edges 204a to 2041, or a combination of the two, by processing the plurality of patient models and the data representing the directed graph to identify at least one element for which the treatment of the subset of patients deviates from at least a portion of the medical guideline. This may involve comparing the desired patient attribute inputs 312a to 312n or the desired patient attribute outputs 314a to 314n of each node 202a to 2021 and / or directed edge 204a to 2041 with the corresponding patient attribute values from each patient model. In some examples, the selected element may be an element for which the number of patients whose treatment has deviated from at least a portion of the medical guideline exceeds a predetermined number.
[0074] Alternatively, selecting at least one element may include processing the plurality of patient models and the data representing the first directed graph to identify at least one element for which a subset of patients whose treatment has deviated from at least a portion of the medical guideline exceeds a predetermined proportion of patients associated with the plurality of patient models. For example, there may be a selected ratio or percentage. Wherein a node and / or directed edge may be selected if the number of patients whose treatment deviates at the node and / or directed edge exceeds a given percentage of the total number of patients.
[0075] At block 508, the system 100 identifies, based on the combination of the selected at least one element and the plurality of patient models, a first patient group whose treatment deviates from at least a portion of the medical guidelines at the selected at least one element and a second patient group whose treatment complies with at least a portion of the medical guidelines at the selected at least one element. In some cases, the system 100 may override the selection of at least one element based on user input.
[0076] The treatment of the patient represented by multiple patient models can deviate from the selected node in a variety of ways. The nodes in the first directed graph can be decision nodes or result nodes. A decision node is a node where the decision to perform a specific clinical step such as a treatment protocol, a medical test, or another active clinical step is to be performed. A result node is a node corresponding to the analysis of the result of the active clinical step. For example, a decision node can represent a decision to perform a patient's blood test, and an associated result node can be a node representing the analysis of the blood test, wherein the variable in the result of the blood test has an expected range. If a medical practitioner decides not to perform a recommended blood test, the patient's treatment may deviate from the decision node. If the result of the patient's blood test is not within the expected range associated with the result node, the patient's treatment may deviate from the result node.
[0077] As described above, the selected node can be associated with at least one condition parameter value, each of the patient models includes a plurality of patient attribute values corresponding to a corresponding clinical step, and a first patient group and a second patient group are identified based on a comparison of the at least one condition parameter value with corresponding patient attribute values from a plurality of patient models. In this manner, patients whose attribute values associated with a clinical step are inconsistent with an expected value for the clinical step can be identified as belonging to the first patient group.
[0078] The patient deviates from a decision node at which the patient's treatment does not comply with the decision represented by the node. In the example of a guideline for a therapy for a certain type of cancer, a decision node in an associated directed graph representing at least a portion of the guideline may represent possible therapy options based on a subtype of cancer and / or certain laboratory values. A possible output at a decision node may be an option to recommend chemotherapy followed by radiation therapy. This recommendation may be followed in many cases, and whether this is the case may be determined by identifying data associated with the recommended steps in the respective patient models (e.g., test results from radiation therapy and chemotherapy stored in the patient model). Alternatively, a medical practitioner may input instructions for performing the recommended steps into the system 100 and have the instructions stored with the patient model.
[0079] However, for some patients, chemotherapy may not be the best treatment modality. For example, an elderly patient may not survive the side effects of chemotherapy. The physician may choose to perform only radiation therapy without chemotherapy because the physician knows that this will result in better outcomes for the patient, such as an increased life expectancy and / or a better quality of life for the patient. In this case, a first patient group and a second patient group are identified based on the availability of healthcare data associated with a corresponding patient model corresponding to the selected node. The patient entries stored in the patient model can each be associated with a relative order and / or date that the patient entries were generated. In this way, the system 100 can determine whether a patient has followed the medical guidelines based on the correspondence between the order of clinical steps recommended by the guidelines and the patient entries in the corresponding patient model.
[0080] Similarly, where a directed edge is associated with a conditional parameter value, the patient's treatment deviates at the directed edge if the corresponding patient attribute value does not satisfy the conditional parameter value associated with the directed edge. The patient's treatment may also deviate at a directed edge if the patient's treatment deviates from the clinical step represented by the node connected to the directed edge. In some implementations, the directed edge represents a clinical step.
[0081] Associating each patient entry with a corresponding date and / or relative order can allow the patient model to store medical information related to the entire patient history without interfering with the analysis of the current treatment stage. In this case, the patient model can include patient entries related to the clinical steps to be performed. However, the patient entry may be related to the prior execution of the clinical steps that occurred in the previous treatment stage or at an earlier date. For example, a patient may be treated for a specific cancer, thereby undergoing both radiotherapy and chemotherapy. After a period of improvement, the cancer may recur, or a new cancer may appear in the patient. The patient can be treated again, but in this case, the patient may not follow the instructions at the decision node to perform the guidelines of radiotherapy and chemotherapy, for example, because the patient has become older and this is no longer a recommended option. In this case, it is important to distinguish the patient entries related to the first appearance of cancer in the patient model from the patient entries related to the recurrence of cancer. In this regard, the patient entries can be associated with dates respectively so that the timeline of the patient history can be accurately mapped to a directed graph.
[0082] At block 510, the system 100 processes a plurality of patient models representing a first patient group and a second patient group to determine at least one patient group characteristic that distinguishes the first patient group from the second patient group. At block 512, the system 100 generates a second directed graph based at least on the at least one identified patient group characteristic. At block 514, the system 100 transmits data representing the second directed graph for receipt by the user devices 106a-106d.
[0083] In some cases, the reasons for not following the medical guidelines may be, for example, due to shortcomings in the medical guidelines that have not been identified to date, differences in practice in different medical care facilities, availability of equipment, and other factors that may affect the ability and / or effectiveness of following the medical guidelines. By identifying patient group characteristics that distinguish the first patient group from the second patient group, it is possible to identify the reasons for not following the medical guidelines at the selected at least one element. Many different factors may be considered patient group characteristics, including but not limited to age, height, weight, gender, body mass index, genetic mutations, pre-existing conditions, associated medical practitioners, and location.
[0084] As described above, if the patient attribute value corresponding to the corresponding clinical step deviates from an expected value, such as a condition parameter value, the patient deviates at the result node. The deviation from the expected value can indicate a poor outcome for the patient. By identifying patient group characteristics of patients who have had poor outcomes due to compliance with medical guidelines, it is possible to make treatment for future patients who share the identified patient group characteristics more effective. To this end, the second directed graph can include an indication that the clinical step represented by the selected at least one element is not recommended for patients associated with the identified at least one group characteristic.
[0085] Fig. 6A A first directed graph 600 representing a portion of a medical guideline for treating a particular type of cancer is shown. The directed graph 600 includes a plurality of nodes 602a to 602e and a set of directed edges 603a to 603e. Node 602c represents a decision to perform radiation therapy and chemotherapy. In the example, node 602c is selected, and it can be determined that a first patient group whose treatment deviates from a portion of the medical guideline at node 602c is distinguished from a second patient group because the patients in the first group share the characteristic of being males over 75 years old. Figure 6B A second directed graph 610 is shown that depends at least on the identified characteristic. The second directed graph 610 includes a plurality of nodes 604a to 604d corresponding to corresponding nodes in the directed graph 600, wherein reference numerals having a common letter suffix indicate nodes representing the same clinical step. For example, node 602a represents the same clinical step as node 604a, node 602b represents the same clinical step as node 604b, and so on. The second directed graph 610 also includes a set of directed edges 605a to 605e corresponding to corresponding directed edges in the directed graph 600, wherein reference numerals having a common letter suffix indicate that the edges correspond to each other. The directed graph 610 also includes an indication that the clinical step represented by the selected nodes 602c, 604c is not recommended for patients associated with at least one of the identified characteristics, in this case, including being male and over 75 years of age. Figure 6B In the example of , the indication is implemented as a modification of the appearance of node 604c. The data representing the second directed graph may also cause the user devices 106a to 106d to generate a warning associated with the selected node 604c. For example, when the patient model is mapped to the directed graph 610 during the patient's treatment, if the patient model is mapped to a node before or at the selected node 604c, a warning may be displayed on the user devices 106a to 106d in conjunction with the selected node 604c, indicating that the clinical step represented by the selected node 604c is not recommended for patients who are male and over 75 years old (i.e., patients who share at least one patient group characteristic of the identified first patient group).
[0086] Figure 6C Another example of a second directed graph 620 depending at least on the identified characteristic is shown. Includes a plurality of nodes 606a to 606e corresponding to the corresponding nodes in directed graphs 600 and 610, wherein the reference numerals with a common letter suffix indicate nodes representing the same clinical steps. For example, node 602a represents the same clinical step as node 606a, node 602b represents the same clinical step as node 606b, etc. The second directed graph 620 also includes a group of directed edges 607a to 607h corresponding to the corresponding directed edges in directed graph 600, wherein the reference numerals with a common letter suffix indicate that the edges correspond to each other. Directed graph 620 also includes another node 606f connected to the selected node 606c via the corresponding directed edge 607h. Another node 606f includes an indication that the clinical step represented by the selected node 606c is not recommended for patients associated with at least one patient group characteristic identified. In this manner, a physician or other medical practitioner may be notified of a possible deviation before the patient is at node 606c where a deviation has been identified as likely to occur for a patient having at least one of the identified patient group characteristics.
[0087] In some examples, the directed graph 620 includes more than one additional node 606f, e.g. Figure 6C In this case, when mapping the patient model to the directed graph 620, a possible deviation can be identified at node 606f, and then an additional clinical step of checking whether the analyzed patient model represents a patient associated with the identified at least one patient group characteristic can be performed at node 606g.
[0088] In other examples, the second directed graph 620 includes another directed edge connected at a first end to another node, the other directed edge indicating a deviation from at least a portion of the medical guidelines for the treatment of the first patient group. For example, the directed edge may indicate that the treatment of the first patient group should deviate from the guidelines before the clinical step represented by the selected node. Another directed edge indicates that the physician should decide how to treat the patient, for example, using their knowledge to determine a preferred treatment method.
[0089] The system 100 can modify the second directed graph based on user input indicating a decision about a patient associated with the identified at least one patient group characteristic. In other words, the system 100 can automatically modify the second directed graph to include a change in the treatment phase for a patient in the first patient group. The system 100 can detect where the user makes a treatment decision for a patient in the first group that deviates from the guidelines, and the system 100 can modify the second directed graph to reflect this.
[0090] A second directed graph 610 or 620 may be used to analyze another patient model that is stored in a third database and includes healthcare information associated with a patient that is associated with at least one characteristic of the identified first patient group. The system 100 may determine the state of at least one additional node and / or the state of a directed edge connected thereto based on a combination of the second directed graph and the other patient model. In this case, the state of the node or the directed edge connected to the node refers to the availability and / or characteristics of data stored in the patient model corresponding to the clinical step represented by the corresponding node. For example, the current position of the patient relative to the medical guideline may be determined based on a comparison of a patient entry stored in the patient model with a conditional parameter value corresponding to the node and / or directed edge.
[0091] The system 100 can then send data indicating the state of at least one other node and / or the state of the directed edge connected thereto, for reception by the user devices 106a to 106d, based on the state of at least one other node and / or the state of the directed edge connected thereto. The data is then used to determine whether the clinical step represented by the selected node is not recommended for the patient. In this way, when the second directed graph is used to inform the treatment of the patient, relevant information and / or prompts notifying the physician of updates about at least one medical guideline can be generated based on the identified patient group characteristics. The system 100 can actually prompt the physician to check some medical information related to the patient to determine whether a specific clinical step should be performed.
[0092] Once the medical practitioner is made aware that the clinical steps in the medical guideline are not recommended for the current patient, they can be prompted to update the second directed graph with the clinical steps they did perform. To this end, the system 100 can send data indicating an input request from a user of the user device 106a to 106d based on the state of another node for reception by the user device 106a to 106d. The system 100 can then receive data indicating additional clinical steps to be represented by additional multiple nodes from the user device 106a to 106d, and modify the second directed graph based on the received data indicating the additional clinical steps.
[0093] Processing the plurality of patient models representing the first patient group and the second patient group to determine at least one patient group characteristic may use any number of appropriate statistical and machine learning techniques. For example, the processing may involve at least one of principal component analysis, random forest regression, and regularized regression.
[0094] 7A to 7C The figure shows how to use statistical methods to identify distinctive group characteristics. Let us consider a plurality of patient models, each of which includes N characteristics or features (f1, f2, ..., f N ). These characteristics can include demographics (gender, age, etc.), clinical findings and observations, previous examinations, current medications, allergies, smoking history, etc.
[0095] Fig. 7A A diagram of a patient model based on two such features (f1, f2) is shown, wherein a first subgroup of patients is shown using circles and a second subgroup is shown with crosses. Although a diagram of a patient model is shown using two such features, it should be understood that differences between patients may be analyzed based on a greater number of variables. Statistical methods may involve analyzing the feature space to identify features that allow one to distinguish between the two groups. Fig. 7A and Figure 7B In the example shown, it is possible to differentiate between the groups based on a single feature, in this case f1. An automated analysis can be performed to identify a threshold that separates the two subgroups.
[0096] A classification and regression tree model (CART) can be trained to classify the patient model into one of two classes, thereby internally performing feature selection. The selected features generated as a byproduct of CART training can be used as discriminative features and form at least one patient group characteristic. As discussed above, other classification techniques can also be used, including linear discriminant analysis (LDA), random forests, and support vector machines (SVM).
[0097] In some examples, such as Figure 7CIn the example shown in , the groups cannot be distinguished by a single feature. Instead, a linear or nonlinear combination of features can be used to distinguish between the two subgroups. In this case, considering two subgroups of patient models, a modified version of the above technique can be used to generate a hyperplane for any number of features N. In the example of FIG. 7 c, the hyperplane is implemented as a classification line.
[0098] Where multiple directed graphs are available, representing at least a portion of the guidelines, for example where one directed graph includes manually or automatically generated modifications to the guidelines, the system 100 can test the suitability of each directed graph and select the most advantageous one.
[0099] In an embodiment, at least one memory 104a to 104n includes computer program code, and at least one memory 104a to 104n and the computer program code are configured to: together with at least one processor 102a to 102n, cause the system 100 to execute Figure 8 Each box in the flowchart indicates a step.
[0100] At block 802 , the system 100 maintains data in the first database 108 representing a first directed graph representing at least a portion of a medical guideline and a second directed graph representing at least a portion of the medical guideline and modifications to at least a portion of the medical guideline.
[0101] Modifications to at least a portion of the medical guideline may include actual deviations from the recommended clinical steps indicated in the guideline. For example, as described above, certain clinical steps may be avoided when treating certain types of patients, and the second directed graph may represent such deviations in treatment. The second directed graph representing the modifications may be manually created, for example, based on assumptions by a medical practitioner, or may be automatically generated based on the patient history represented by the patient model.
[0102] Each of the first directed graph and the second directed graph includes a corresponding plurality of elements representing clinical steps. The plurality of elements may include a plurality of nodes and a set of directed edges, wherein the plurality of nodes represent corresponding clinical steps.
[0103] At block 804, the system 100 maintains a plurality of patient models each including healthcare data associated with a respective patient in the second database 108. The first database and the second database may be the same database or different databases.
[0104] At block 806, the system 100 identifies a first set of patient models and a second set of patient models. The first set of patient models represents patients that have been treated based on at least a portion of the medical guideline represented by the first directed graph. The second set of patient models represents patients that have been treated based on at least a portion of the medical guideline represented by the second directed graph. Where the modification depends on a patient group characteristic, the first set of patient models and the second set of patient models may represent patients associated with the patient group characteristic. In other examples, the patient models are a random or near-random sample of patient models stored in the second database. As previously described with respect to FIG. 6A to FIG. 6C As discussed, the modification to at least a portion of the medical guideline can be represented in the second directed graph by at least one of at least one node and at least one directed edge. Similar to the above example, the modification to at least a portion of the medical guideline is associated with a patient group characteristic, and a first group of patient models and a second group of patient models can be identified based on the patient group characteristic.
[0105] At block 808, the system 100 determines which of the first directed graph and the second directed graph is a preferred directed graph based on a comparison of the first set of patient models with the second set of patient models. At block 810, the system 100 transmits data representing the preferred directed graph in response to the determination for receipt by the user devices 106a to 106d.
[0106] Preferred directed graphs are generally directed graphs that result in better outcomes for patients when treatments are administered to those patients by a physician. A better outcome for a patient is one that provides any of the following: improved health status, better quality of life, longer life expectancy, fewer complications following treatment, and, in some cases, a reduced likelihood of disease recurrence.
[0107] The system 100 can determine which of the first directed graph or the second directed graph is a preferred directed graph based on a plurality of patient models. In this embodiment, the patient models each include at least one patient outcome measure, for example, in a corresponding one or more patient entries. The system 100 can compare a first plurality of patient outcome measures of a first set of patient models with a second plurality of patient outcome measures of a second set of patient models.
[0108] Outcome measures in the patient model may be automatically generated, for example, based on data including test results generated from devices related to the treatment phase. In other examples, outcome measures may be collected and manually entered based on input from a user. In this case, a physician treating a patient may generate data representing outcome measures during the patient's treatment and / or during a follow-up procedure after treatment. These data may be input to user devices 106a to 106d and sent for storage in a corresponding patient model.
[0109] The system 100 may identify a first set of patient models based on a comparison of the plurality of patient models to the first directed graph. The comparison of the patient models to the first directed graph may include determining, for each patient model, a state of each of the elements of the first directed graph. Similarly, a second set of patient models may be identified based on a comparison of the plurality of patient models to the second directed graph. Alternatively, the second set of patient models may be determined based on a process of elimination following identification of the first set of patient models.
[0110] To determine the state of the element, the system 100 can maintain a first association between at least one of the patient entries and an identifier from a plurality of identifiers and a second association between the element and the identifier. An attribute value associated with the element is selected based on the first association and the second association. Then, the system 100 determines whether the conditional parameter value associated with the element is satisfied based on a comparison of the attribute value associated with the element and the conditional parameter value associated with the element.
[0111] Determining which directed graph is the preferred directed graph may also be sensitive to the degree to which patients follow each directed graph. For example, patients treated according to the first directed graph may be more likely to deviate than patients treated according to the second directed graph, which may be an indication that the second directed graph is not suitable. To this end, the system 100 can determine a first measure and a second measure, the first measure indicating the average consistency of the first group of patient models with the first directed graph, and the second measure indicating the average consistency of the second group of patient models with the second directed graph. The first measure and the second measure of average consistency can be used to perform a comparison of the first plurality of patient outcome measures with the second plurality of patient outcome measures. The first measure and the second measure of consistency depend on average states with respect to: (i) the first group of patient models and multiple elements of the first directed graph; and (ii) the second group of patient models and multiple elements of the second directed graph.
[0112] The user input may be used to influence the selection of the first directed graph or the second directed graph as the preferred directed graph. To this end, the results of the comparison of the first patient model with the second patient model - i.e., the comparison of the respective outcome measures - may be sent for receipt by the user devices 106a to 106d. The system 100 may then receive data from the user devices 106a to 106d indicating a decision regarding the results of the comparison, and select one of the first directed graph or the second directed graph based on the received data.
[0113] The system 100 may also provide the user devices 106a to 106d with supplemental data and data representing the preferred directed graph. If the first directed graph is preferred, the system 100 may send data corresponding to the first plurality of result measurements for reception by the user devices 106a to 106d; or if the second directed graph is preferred, the system 100 may send data corresponding to the second plurality of result measurements for reception by the user devices 106a to 106d.
[0114] The above examples are given for illustrative purposes.The above described system may be used for staging and management of any patient suffering from a disease for which at least one medical guideline is available.
[0115] Numbered clauses
[0116] The following numbered clauses describe various embodiments of the invention.
[0117] 1. A system operable to send healthcare data to a user device, the user device being configured for analyzing medical information, the system comprising at least one processor and at least one memory containing computer program code, the at least one memory and the computer program code being configured to, together with the at least one processor, cause the system to:
[0118] maintaining, in a first database, data representing a first directed graph, the first directed graph representing at least a portion of a medical guideline, the first directed graph comprising a plurality of elements representing clinical steps;
[0119] maintaining in a second database a plurality of patient models each including healthcare data associated with a respective patient;
[0120] selecting at least one element from the plurality of elements by processing the plurality of patient models and data representing the first directed graph to identify at least one element for which treatment of a subset of patients deviates from at least a portion of the medical guideline;
[0121] identifying, based on a combination of the selected at least one element and the plurality of patient models, a first patient group whose treatment deviates from at least a portion of the medical guideline at the selected at least one element and a second patient group whose treatment complies with at least a portion of the medical guideline at the selected at least one element;
[0122] processing the plurality of patient models representing the first patient group and the second patient group to identify at least one patient group characteristic that distinguishes the first patient group from the second patient group;
[0123] generating a second directed graph based at least on the identified at least one patient group characteristic; and
[0124] Data representing the second directed graph is transmitted for reception by the user device.
[0125] 2. A system according to claim 1, wherein selecting the at least one element includes: processing the multiple patient models and data representing the first directed graph to identify at least one element at which a subset of patients whose treatment deviates from at least a portion of the medical guidelines exceeds a predetermined proportion of patients associated with the multiple patient models.
[0126] 3. A system according to clause 1 or clause 2, wherein the at least one patient group characteristic comprises at least one of the following:
[0127] age;
[0128] height;
[0129] weight;
[0130] gender;
[0131] Body mass index;
[0132] Gene mutations;
[0133] Associated medical practitioners; and
[0134] Location.
[0135] 4. A system according to any preceding clause, wherein the memory and computer program code are configured to: with the at least one processor, cause the system to override selection of the at least one element based on user input.
[0136] 5. A system according to any of the preceding clauses, wherein at least one selected element is associated with at least one condition parameter value, each of the patient models includes a plurality of patient attribute values corresponding to respective clinical steps, and the first patient group and the second patient group are identified based on a comparison of the at least one condition parameter value with corresponding patient attribute values from the plurality of patient models.
[0137] 6. The system of any of clauses 1 to 4, wherein the first patient group and the second patient group are identified based on availability of healthcare data in respective patient models corresponding to the selected at least one element.
[0138] 7. The system of any preceding clause, wherein processing the plurality of patient models representing the first patient group and the second patient group to determine at least one patient group characteristic that distinguishes the first patient group from the second patient group comprises processing the plurality of patient models using at least one of:
[0139] Principal component analysis;
[0140] Random Forest Regression; and
[0141] Regularized regression.
[0142] 8. A system according to any preceding clause, wherein the second directed graph includes an indication that the clinical step represented by the selected at least one element is not recommended for patients associated with the identified at least one patient group characteristic.
[0143] 9. The system of clause 8, wherein the second directed graph comprises:
[0144] Multiple nodes;
[0145] a set of directed edges; and
[0146] at least one other node connected to the selected element,
[0147] Therein, the at least one further node comprises an indication that the clinical step represented by the selected element is not recommended for patients associated with the identified at least one patient group characteristic.
[0148] 10. The system of clause 9, wherein the second directed graph comprises at least one further directed edge connected at a first end to the further node, the further directed edge indicating a deviation from at least a portion of the medical guideline for the first patient group.
[0149] 11. A system according to any preceding clause, wherein the at least one memory and computer program code are configured to, together with the at least one processor, cause the system to:
[0150] The second directed graph is modified based on user input indicating a decision regarding a patient associated with the identified at least one patient group characteristic.
[0151] 12. A system according to clause 9 or clause 10, wherein the at least one memory and computer program code are configured to, together with the at least one processor, cause the system to:
[0152] maintaining in a third database an additional patient model including healthcare data associated with a patient associated with the identified at least one patient group characteristic;
[0153] determining, based on a combination of the second directed graph and the further patient model, a state of the at least one further node and / or a state of a directed edge connected to the at least one further node; and
[0154] Based on the state of the at least one additional node and / or the state of the directed edge connected to the at least one additional node, data indicating the state of the at least one additional node and / or the state of the directed edge connected to the at least one additional node is sent for reception by the user device, and the data indicating the state of the at least one additional node and / or the state of the directed edge connected to the at least one additional node is used to determine whether the clinical step represented by the selected node is not recommended for the patient.
[0155] 13. The system of clause 12, wherein the at least one memory and the computer program code are configured to, together with the at least one processor, cause the system to:
[0156] sending data indicative of an input request from a user of the user device for receipt by the user device based on a state of the further node and / or a state of a directed edge connected to the further node;
[0157] receiving data from the user device indicating additional clinical steps to be represented by an additional plurality of nodes; and
[0158] The second directed graph is modified based on the received data indicative of the additional clinical steps.
[0159] 14. A computer program comprising a set of instructions which, when executed by a computerized device, cause the computerized device to perform a method of sending healthcare data to a user device, the user device being configured for analyzing medical information, the method comprising:
[0160] maintaining, in a first database, data representing a first directed graph, the first directed graph representing at least a portion of a medical guideline, the first directed graph comprising a plurality of elements representing clinical steps;
[0161] maintaining in a second database a plurality of patient models each including healthcare data associated with a respective patient;
[0162] selecting at least one element from the plurality of elements by processing the plurality of patient models and data representing the first directed graph to identify at least one element for which treatment of a subset of patients deviates from at least a portion of the medical guideline;
[0163] identifying, based on a combination of the selected at least one element and the plurality of patient models, a first patient group whose treatment deviates from at least a portion of the medical guideline at the selected at least one element and a second patient group whose treatment complies with at least a portion of the medical guideline at the selected at least one element;
[0164] processing the plurality of patient models representing the first patient group and the second patient group to identify at least one patient group characteristic that distinguishes the first patient group from the second patient group;
[0165] generating a second directed graph based at least on the identified at least one patient group characteristic; and
[0166] Data representing the second directed graph is transmitted for reception by the user device.
[0167] 15. A method for transmitting healthcare data to a user device, the user device being configured to analyze medical information, the method comprising:
[0168] maintaining, in a first database, data representing a first directed graph, the first directed graph representing at least a portion of a medical guideline, the first directed graph comprising a plurality of elements representing clinical steps;
[0169] maintaining in a second database a plurality of patient models each including healthcare data associated with a respective patient;
[0170] selecting at least one element from the plurality of elements by processing the plurality of patient models and data representing the first directed graph to identify at least one element for which treatment of a subset of patients deviates from at least a portion of the medical guideline;
[0171] identifying, based on a combination of the selected at least one element and the plurality of patient models, a first patient group whose treatment deviates from at least a portion of the medical guideline at the selected at least one element and a second patient group whose treatment complies with at least a portion of the medical guideline at the selected at least one element;
[0172] processing the plurality of patient models representing the first patient group and the second patient group to identify at least one patient group characteristic that distinguishes the first patient group from the second patient group;
[0173] generating a second directed graph based at least on the identified at least one patient group characteristic; and
[0174] Data representing the second directed graph is transmitted for reception by the user device.
[0175] 16. A system operable to send healthcare data to a user device, the user device being configured for analyzing medical information, the system comprising at least one processor and at least one memory containing computer program code, the at least one memory and computer program code being configured to, with the at least one processor, cause the system to:
[0176] maintaining data representing a first directed graph and a second directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the second directed graph representing at least a portion of the medical guideline and modifications to at least a portion of the medical guideline, each directed graph including a respective plurality of elements representing clinical steps;
[0177] maintaining in a second database a plurality of patient models each including healthcare data associated with a respective patient;
[0178] identifying a first set of patient models representing patients who have been treated based on at least a portion of the medical guidelines represented by the first directed graph and a second set of patient models representing patients who have been treated based on at least a portion of the medical guidelines represented by the second directed graph;
[0179] determining which of the first directed graph and the second directed graph is a preferred directed graph based on a comparison of the first set of patient models and the second set of patient models; and
[0180] In response to the determination, data representing the preferred directed graph is transmitted for receipt by the user device.
[0181] 17. The system of clause 16, wherein the modification to at least a portion of the medical guideline is represented by at least one of:
[0182] at least one node in the second directed graph; and
[0183] At least one directed edge in the second directed graph.
[0184] 18. A system according to clause 16 or clause 17, wherein the multiple patient models each include multiple patient entries, at least one of the patient entries includes at least one patient outcome measurement, and determining which of the first directed graph and the second directed graph is the preferred directed graph includes: comparing a first plurality of patient outcome measurements of the first group of patient models with a second plurality of patient outcome measurements of the second group of patient models.
[0185] 19. The system of any of clauses 16 to 18, wherein modification of at least a portion of the medical guideline is associated with a patient group characteristic, and the first and second sets of patient models are identified based on the patient group characteristic.
[0186] 20. A system according to any one of clauses 16 to 19, wherein the at least one memory and computer program code are configured to: together with the at least one processor, cause the system to identify the first group of patient models based on comparing the multiple patient models with the first directed graph, wherein comparing the multiple patient models with the first directed graph includes: for each patient model, determining a state of at least one element of the multiple elements of the first directed graph.
[0187] 21. A system according to any one of clauses 16 to 20, wherein the at least one memory and computer program code are configured to: together with the at least one processor, cause the system to identify the second group of patient models based on comparing the multiple patient models with the second directed graph, wherein comparing the multiple patient models with the second directed graph includes: for each patient model, determining a state of at least one element of the multiple elements of the second directed graph.
[0188] 22. A system according to clause 20 or clause 21, wherein, for the patient model, the state of the element depends on the availability of data in the patient model associated with a clinical step associated with the element.
[0189] 23. The system of clause 20 or clause 21, wherein each patient model comprises a plurality of patient entries, each patient entry comprises at least one attribute value, and determining the state of the element comprises:
[0190] maintaining a first association between at least one of the patient entries of the patient model and an identifier from a plurality of identifiers;
[0191] maintaining a second association between the element and an identifier from the plurality of identifiers;
[0192] selecting the attribute value associated with the element based on the first association and the second association; and
[0193] Based on a comparison of the attribute value associated with the element and the conditional parameter value associated with the element, it is determined whether the conditional parameter value associated with the element is satisfied.
[0194] 24. The system of any of clauses 16 to 23, wherein determining which of the directed graphs is a preferred directed graph comprises:
[0195] determining a first measure indicative of average consistency of the first set of patient models with the first directed graph;
[0196] determining a second measure indicative of average consistency of the second set of patient models with the second directed graph; and
[0197] A comparison of the first plurality of patient outcome measures to the second plurality of patient outcome measures is performed using the first measure and the second measure.
[0198] 25. The system of clause 24, wherein the first measure depends on an average state of a first set of patient models of the plurality of elements of the first directed graph.
[0199] 26. A system according to clause 24 or clause 25, wherein the second measure depends on an average state of a second set of patient models of the plurality of elements of the second directed graph.
[0200] 27. The system of any of clauses 16 to 26, wherein determining which of the first directed graph and the second directed graph is a preferred directed graph comprises:
[0201] transmitting data indicative of a result of a comparison of the first set of patient models with the second set of patient models for receipt by the user device;
[0202] receiving data from the user device indicative of a decision regarding an outcome of the comparison; and
[0203] Based on the received data indicative of the decision, one of the first directed graph or the second directed graph is selected.
[0204] 28. The system of clause 18, wherein the at least one memory and the computer program code are configured to, with the at least one processor, cause the system to:
[0205] If the first directed graph is the preferred directed graph, sending data indicative of the first plurality of patient outcome measures for receipt by the user device; or
[0206] If the second directed graph is the preferred directed graph, data indicative of the second plurality of patient outcome measures is transmitted for receipt by the user device.
[0207] 29. A computer program comprising a set of instructions which, when executed by a computerized device, cause the computerized device to perform a method of sending healthcare data to a user device, the user device being configured for analyzing medical information, the method comprising:
[0208] maintaining data representing a first directed graph and a second directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the second directed graph representing at least a portion of the medical guideline and modifications to at least a portion of the medical guideline, each directed graph including a respective plurality of elements representing clinical steps;
[0209] maintaining in a second database a plurality of patient models each including healthcare data associated with a respective patient;
[0210] identifying a first set of patient models representing patients who have been treated based on at least a portion of the medical guidelines represented by the first directed graph and a second set of patient models representing patients who have been treated based on at least a portion of the medical guidelines represented by the second directed graph;
[0211] determining which of the first directed graph and the second directed graph is a preferred directed graph based on a comparison of the first set of patient models and the second set of patient models; and
[0212] In response to the determination, data representing the preferred directed graph is transmitted for receipt by the user device.
[0213] 30. A method for transmitting healthcare data to a user device, the user device being configured to analyze medical information, the method comprising:
[0214] maintaining data representing a first directed graph and a second directed graph in a first database, the first directed graph representing at least a portion of a medical guideline, the second directed graph representing at least a portion of the medical guideline and modifications to at least a portion of the medical guideline, each directed graph including a respective plurality of elements representing clinical steps;
[0215] maintaining in a second database a plurality of patient models each including healthcare data associated with a respective patient;
[0216] identifying a first set of patient models representing patients who have been treated based on at least a portion of the medical guidelines represented by the first directed graph and a second set of patient models representing patients who have been treated based on at least a portion of the medical guidelines represented by the second directed graph;
[0217] determining which of the first directed graph and the second directed graph is a preferred directed graph based on a comparison of the first set of patient models and the second set of patient models; and
[0218] In response to the determination, data representing the preferred directed graph is transmitted for receipt by the user device.
Claims
1. A system (100) operable to send healthcare data to a user device (106a to 106d), the user device (106a to 106d) being configured for analyzing medical information, the system (100) comprising at least one processor (102a to 102n) and at least one memory (104a to 104n) containing computer program code, the at least one memory (104a to 104n) and the computer program code being configured to, together with the at least one processor (102a to 102n), cause the system (100) to: maintaining data representing a first directed graph (600) in a first database (108), the first directed graph (600) representing at least a portion of a medical guideline, the first directed graph (600) including a plurality of elements representing clinical steps; maintaining in a second database (108) a plurality of patient models each including healthcare data associated with a respective patient; selecting at least one element from the plurality of elements by processing the plurality of patient models and data representing the first directed graph (600) to identify at least one element for which treatment of a subset of patients deviates from at least a portion of the medical guideline; identifying, based on a combination of the selected at least one element and the plurality of patient models, a first patient group whose treatment deviates from at least a portion of the medical guideline at the selected at least one element and a second patient group whose treatment complies with at least a portion of the medical guideline at the selected at least one element; processing the plurality of patient models representing the first patient group and the second patient group to identify at least one patient group characteristic that distinguishes the first patient group from the second patient group; generating a second directed graph (610, 620) based at least on the identified at least one patient group characteristic, the second directed graph (610, 620) including an indication of deviations from at least a portion of the medical guideline for the first patient group; as well as transmitting data representing the second directed graph (610, 620) for reception by the user device (106a to 106d), Wherein, selecting the at least one element includes: processing the multiple patient models and data representing the first directed graph (600) to identify the following at least one element, at which a subset of patients whose treatment deviates from at least a portion of the medical guideline exceeds a predetermined proportion of patients associated with the multiple patient models.
2. The system (100) according to claim 1, in, The at least one patient group characteristic comprises at least one of the following: age; height; weight; gender; Body mass index; Gene mutations; Associated medical practitioners; and Location.
3. The system (100) according to any of the preceding claims, in, The memory (104a to 104n) and computer program code are configured to, with the at least one processor (102a to 102n), cause the system (100) to override selection of the at least one element based on user input.
4. The system (100) according to any of the preceding claims, in, The selected at least one element is associated with at least one condition parameter value, the patient models each include a plurality of patient attribute values corresponding to respective clinical steps, and the first patient group and the second patient group are identified based on a comparison of the at least one condition parameter value with respective patient attribute values from the plurality of patient models.
5. The system (100) according to any one of claims 1 to 3, in, The first patient group and the second patient group are identified based on availability of healthcare data in respective patient models corresponding to the selected at least one element.
6. The system (100) according to any preceding claim, in, Processing the plurality of patient models representing the first patient group and the second patient group to determine at least one patient group characteristic that distinguishes the first patient group from the second patient group includes processing the plurality of patient models using at least one of: Principal component analysis; Random Forest Regression; and Regularized regression.
7. The system (100) according to any preceding claim, in, The second directed graph (610, 620) includes an indication that the clinical step represented by the selected at least one element is not recommended for patients associated with the identified at least one patient group characteristic.
8. The system (100) according to claim 7, in, The second directed graph (620) includes: A plurality of nodes (606a to 606e); a set of directed edges; and at least one additional node (606f) connected to the selected element, Therein, the at least one additional node (606f) includes an indication that the clinical step represented by the selected element is not recommended for patients associated with the identified at least one patient group characteristic.
9. The system (100) according to claim 8, in, The second directed graph (620) includes at least one further directed edge connected at a first end to the further node (606f), the further directed edge indicating a deviation from at least a portion of the medical guideline for the first patient group.
10. The system (100) according to any preceding claim, in, The at least one memory (104a to 104n) and the computer program code are configured to, together with the at least one processor (102a to 102n), cause the system (100) to: The second directed graph is modified based on user input indicating a decision regarding a patient associated with the identified at least one patient group characteristic (610, 620).
11. The system (100) according to claim 8 or 9, in, The at least one memory (104a to 104n) and the computer program code are configured to, together with the at least one processor (102a to 102n), cause the system (100) to: maintaining in a third database (108) an additional patient model including healthcare data associated with a patient associated with the identified at least one patient group characteristic; Determining a state of the at least one further node (606f) and / or a state of a directed edge (607h) connected to the at least one further node (606f) based on a combination of the second directed graph (620) and the further patient model; as well as Based on the state of the at least one additional node (606f) and / or the state of the directed edge (607h) connected to the at least one additional node (606f), data indicating the state of the at least one additional node (606f) and / or the state of the directed edge (607h) connected to the at least one additional node (606f) is sent for reception by the user device (106a to 106d), and the data indicating the state of the at least one additional node (606f) and / or the state of the directed edge (607h) connected to the at least one additional node (606f) is used to determine whether the clinical step represented by the selected node (606c) is not recommended for the patient.
12. The system (100) according to claim 11, in, The at least one memory (104a to 104n) and the computer program code are configured to, together with the at least one processor (102a to 102n), cause the system (100) to: sending data indicative of an input request from a user of the user device (106a to 106d) for receipt by the user device (106a to 106d) based on a state of the further node (606f) and / or a state of a directed edge connected to the further node (606f); receiving data from the user device (106a to 106d) indicating additional clinical steps to be represented by an additional plurality of nodes; as well as The second directed graph is modified based on the received data indicative of the additional clinical steps (610, 620).
13. A computer program comprising a set of instructions which, when executed by a computerized device, cause the computerized device to perform a method of sending healthcare data to a user device (106a to 106d), the user device (106a to 106d) being configured to analyze medical information, the method include: maintaining data representing a first directed graph (600) in a first database (108), the first directed graph (600) representing at least a portion of a medical guideline, the first directed graph (600) including a plurality of elements representing clinical steps; maintaining in a second database (108) a plurality of patient models each including healthcare data associated with a respective patient; selecting at least one element from the plurality of elements by processing the plurality of patient models and data representing the first directed graph (600) to identify at least one element for which treatment of a subset of patients deviates from at least a portion of the medical guideline; identifying, based on a combination of the selected at least one element and the plurality of patient models, a first patient group whose treatment deviates from at least a portion of the medical guideline at the selected at least one element and a second patient group whose treatment complies with at least a portion of the medical guideline at the selected at least one element; processing the plurality of patient models representing the first patient group and the second patient group to determine at least one patient group characteristic that distinguishes the first patient group from the second patient group; generating a second directed graph (610, 620) based at least on the identified at least one patient group characteristic, the second directed graph (610, 620) including an indication of deviations from at least a portion of the medical guideline for the first patient group; as well as transmitting data representing the second directed graph (610, 620) for reception by the user device (106a to 106d), Wherein, selecting the at least one element includes: processing the multiple patient models and data representing the first directed graph (600) to identify the following at least one element, at which a subset of patients whose treatment deviates from at least a portion of the medical guideline exceeds a predetermined proportion of patients associated with the multiple patient models.
14. A method of transmitting healthcare data to a user device (106a to 106d), the user device (106a to 106d) being configured to analyze medical information, the method include: maintaining data representing a first directed graph (600) in a first database (108), the first directed graph (600) representing at least a portion of a medical guideline, the first directed graph (600) including a plurality of elements representing clinical steps; maintaining in a second database (108) a plurality of patient models each including healthcare data associated with a respective patient; selecting at least one element from the plurality of elements by processing the plurality of patient models and data representing the first directed graph (600) to identify at least one element for which treatment of a subset of patients deviates from at least a portion of the medical guideline; identifying, based on a combination of the selected at least one element and the plurality of patient models, a first patient group whose treatment deviates from at least a portion of the medical guideline at the selected at least one element and a second patient group whose treatment complies with at least a portion of the medical guideline at the selected at least one element; processing the plurality of patient models representing the first patient group and the second patient group to determine at least one patient group characteristic that distinguishes the first patient group from the second patient group; generating a second directed graph (610, 620) based at least on the identified at least one patient group characteristic, the second directed graph (610, 620) including an indication of deviations from at least a portion of the medical guideline for the first patient group; as well as transmitting data representing the second directed graph (610, 620) for reception by the user device (106a to 106d), Wherein, selecting the at least one element includes: processing the multiple patient models and data representing the first directed graph (600) to identify the following at least one element, at which a subset of patients whose treatment deviates from at least a portion of the medical guideline exceeds a predetermined proportion of patients associated with the multiple patient models.
