Methods for determining a patient's risk score

Abstract

The present disclosure generally relates to a computer-implemented method for updating a patient's treatment model. The present disclosure also relates to a corresponding computer system and computer program product.

Description

[Technical Field]

[0001] The present disclosure generally relates to a computer-implemented method for determining a patient's risk score. The present disclosure also relates to a corresponding computer system and computer program product. [Background technology]

[0002] Over the past few decades, health care costs have grown rapidly, and various plans have been proposed to at least slow the growth of health care costs. Such plans may focus, for example, on implementing thresholds above the threshold for individuals to receive appropriate treatment, while still attempting to maintain a desired level of quality of care within the health care system.

[0003] Alternatively, a doctor, nurse, or any other form of skilled therapist or medical consultant may seek to provide recommendations to an individual with the aim of creating changes in their situation that may have a beneficial impact on the individual's health, thereby reducing the risk that the individual will have to seek treatment within the healthcare system.

[0004] Some form of individual pre-assessment is necessary to be able to determine when to treat a patient and when not to treat them.

[0005] When evaluating an individual, for example, a doctor, nurse, or anyone assisting a patient, they use their experience, guidelines, and best practices to define as objectively as possible the individual's current condition and, in some cases, the recommended changes in treatment status proposed for the individual. For example, a doctor or therapist, while maintaining a sophisticated knowledge base, is human and may not be aware of recent developments in the community or understand the individual's overall situation, including all relevant medical information. Furthermore, currently available best practices may, in some circumstances, be to refrain from providing individualized treatment to an individual.

[0006] Recently, digital solutions have been introduced to assist doctors or therapists, significantly reducing the subjectivity of their decision-making while at the same time improving the "resolution" of the best available practices and allowing them to make more data-based decisions, which also allow the inclusion of all medical information relevant to an individual when defining their current condition.

[0007] One example of a digitalized solution that can be used to recommend status changes is presented in U.S. Patent Application Publication No. 2018 / 0165418. U.S. Patent Application Publication No. 2018 / 0165418 specifically discloses a system that collects data directly characterizing an individual's health, as well as status data related to factors that may potentially affect the individual's health. The collected factor data is used by the system to construct a vector of characteristics ("health vector") that indicates and reflects the individual's health status over time. The system may also evaluate differences in the individual's health vectors as they exist at different points in time to generate changes in the health vector. The system uses the individual's health vector and the changes in the health vector to determine the individual's current health score, which characterizes the individual's overall health at that time (e.g., a spectrum from very healthy to very unhealthy). The system also periodically generates a health score based on more recent health vector information to construct a trend of the individual's health change as the individual's health score changes over time ("health score trend"). The system compares an individual's health score trend data with data reflecting health score trends of people in approximately the same location (i.e., one or more population cohorts) and, based on that comparison and the behavioral patterns of the compared cohorts, generates recommendations for actions or changes that the individual can take that are not only likely to improve the individual's health but are also likely to be adopted by the individual.

[0008] However, the solution presented in US Patent No. 6,299,949 has some general drawbacks: First, the solution presented in US Patent No. 6,299,949 is less accurate in terms of individual assessment, and ultimately, the doctor / therapist may decide to "play it safe" and ensure the individual is satisfied, bypassing the possible recommendations of a digitalized solution.

[0009] Second, the solution presented in Patent Document 1 is only applicable to general recommendations for individuals and does not focus at all on actions that need to be taken when an individual is hospitalized or requires actual treatment within the healthcare system. Therefore, the solution presented in Patent Document 1 does not solve the problem of rising healthcare costs, especially when an individual must receive actual treatment within the healthcare system.

[0010] With the above in mind, there seems to be scope for further improvement of digitalized solutions for doctors, balancing the reliability of assessment and the quality of care, with the overall intention of providing individuals with the type of treatment that best suits their current health / situation. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] US Patent Application Publication No. 2018 / 0165418 Summary of the Invention

[0012] In accordance with one aspect of the present disclosure, the above is alleviated by a computer-implemented method executed by a control unit for determining a patient's risk score, the method comprising: receiving, at the control unit, a first set of personal parameters indicative of the patient's current or previous condition; forming, using the control unit, a personal patient model based on the first set of personal parameters; determining, using the control unit, a degree of match between the personal patient model and each of a plurality of different pre-defined generic patient models, each generic patient model having a pre-defined patient risk score; selecting, using the control unit, at least one generic patient model having a degree of match that exceeds a predetermined threshold; and determining, using the control unit, the patient's risk score based on the at least one selected generic patient model.

[0013] The overall idea of the present disclosure is to determine a patient's risk score, using the patient's prior assessment as a primary input. The risk score may then be used within a healthcare system to provide the patient with the most appropriate treatment. In line with the present disclosure, determining a patient's risk score, compared to prior art, is not simply based on the patient's prior assessment, but includes a process in which data about the patient is compared against multiple different generic patient models. The different generic patient models may have been previously developed in close collaboration with experts in different fields. Here, the different generic patient models may generally be considered to be associated with different patient behaviors and outcomes, such as inadequate treatment. Furthermore, the different generic patient models are typically not based on knowledge about a single patient, but on general (typically anonymized) knowledge about a large number of patients and the (combined) expected outcomes for such patients.

[0014] Thus, in line with the present disclosure, a patient's personal model (which depends on the data collected about the patient) is matched against multiple different generic patient models, and at least one generic patient model with a degree of match above a predetermined threshold is selected. Thus, instead of simply determining a patient's risk score based on a direct assessment of the patient, the present scheme ensures that the patient's assessment is put into the "big picture" by matching the patient's specific behavior with a "population" of patients who appeared / behaved in roughly the same way.

[0015] Thus, the present disclosure allows relying not only on an individual patient but also on general patient behavior to determine a patient's risk score. Thus, advantages of following this scheme include the possibility of reliably predicting a patient's likely future behavior and how to optimally handle this possible behavior to minimize patient complications. Matching to different predefined general patient models may also be considered a way to filter out possible variations in individual parameters for a patient, as such variations may have previously been determined to have little impact on the patient's future.

[0016] Thus, the present disclosure may aim to ensure that the quality of care provided to patients is improved, while also ensuring that "over-treatment" is reduced, thereby reducing the overall burden on the healthcare system. Furthermore, the present disclosure may be implemented in a very flexible manner, ensuring that "new" or "updated" generic patient models may be introduced along the way, possibly taking into account newly identified "best practices."

[0017] Within the context of the present disclosure, the phrase "a set of personal parameters indicative of a patient's current or previous condition" should be interpreted broadly and should include any type of relevant information that has been or has been collected about a patient. Such information may include, but is not limited to, patient clinical data collected over a period of time (e.g., including data collected at different doctor's visits and / or hospitalizations, from a few seconds / hours to the patient's lifetime), including, for example, patient vitals, hospitalizations, test results, and prescription medications. Additional information that may be relevant for use includes, for example, heart rate data, electrocardiogram (EKG / ECG) data, respiratory rate data, patient temperature data, pulse oximetry data, and blood pressure data.

[0018] Of course, other parameters relevant to the patient are possible, such as BMI, continence, incontinence, visual risk areas of skin type, gender and age, malnutrition screening (MTS), mobility, other physical conditions, mental state, activity, sensory perception, hydration of the patient's body parts, nutritional intake, friction and shear of the patient's body parts, temperature, information about previous pressure sores, perfusion (blood flow), diabetes, tissue perfusion and oxygenation, hygiene, hemodynamics, etc.

[0019] Preferably, the set of personal parameters may, in some embodiments, include at least one of patient images and video sequences. However, it may be appropriate to allow the caregiver to input other patient-related information, such as information related to the parameters listed above. The images and / or videos may preferably be collected using, for example, a camera arranged in communication with a control unit. Here, the control unit may apply, for example, an image processing scheme to extract the parameters listed above. The image processing scheme may, in some embodiments, be configured to perform normalization using previously collected patient data.

[0020] Furthermore, in one embodiment, the expression "individual patient model" should also be interpreted broadly to include the determination of a collection of individual parameter sets for a patient, but in another embodiment, the individual patient model may be defined as a "receptacle" of the patient's individual parameters, for example, as a string of individual parameters, possibly organized according to a predefined standard to improve matching with the generic patient model.

[0021] Furthermore, the term "control unit" should be interpreted broadly and may include any means for providing computing power for executing a scheme according to the present disclosure. Thus, as discussed further in the detailed description of the present disclosure, a control unit (corresponding to any means for providing processing power) may, in some cases, be implemented in a server, in a client device (e.g., a computer or a mobile device), or shared between a server and a client device.

[0022] Preferably, in one embodiment of the present disclosure, the selecting step includes selecting the generic patient model with the highest degree of match. Thus, one particular generic patient model may be identified in some embodiments as the most relevant model, and the risk scoring is itself based on this match. Such an implementation may be preferred in some embodiments, for example, when it is desirable to quickly determine a patient's risk score.

[0023] However, it may alternatively be possible to determine the risk score based on the selection of multiple single generic patient models, for example, based on a combination of at least two selected generic patient models. In such an embodiment, it may be desirable to assign a weight to each selected generic patient model, where the weight may depend, for example, on the degree of match. Clearly, such an implementation may provide further improvements in the reliability of the determined risk score, but may also require slightly more processing and therefore be slightly slower than if only a single generic patient model were selected.

[0024] A predetermined threshold may be used in some embodiments to ensure that the match maintains at least a certain basic level. That is, if the match is sufficiently low, i.e., if no actual match is produced when comparing the personal patient model with multiple different pre-defined generic patient models, this information may be used as an indication that the patient's risk score should be manually assessed during a physical examination without relying on this scheme. However, a low degree of match may also be considered as an indicator that the personal parameters indicative of the patient's current or previous condition are incorrect or otherwise unreliable, and that it would be appropriate to gather additional / new information about the patient before proceeding with the risk score determination.

[0025] In one embodiment of the present disclosure, the method further includes using the control unit to define low-risk, medium-risk, and high-risk categories, and using the control unit to assign a risk category to the patient by comparing the patient's determined risk score with predefined risk score ranges for the various categories. Of course, additional categories may be included, and such additional categories are within the scope of the present disclosure. Such additional categories may include, for example, an "elevated risk category" intermediate between the medium-risk category and the high-risk category. The use of risk categories may be useful, for example, to allow caregivers to obtain quick information on how to act in relation to the patient. Here, for example, various categories may have been previously assigned (e.g., during training) to different actions (e.g., between 0 and 100, or defined in other ways) without the need to interpret a "risk score number." Thus, if a patient is determined to be in the high-risk category, the caregiver can act quickly to address the patient.

[0026] Thus, in one embodiment of the present disclosure, the scheme may further include using the control unit to form a suggested treatment for the patient based on the selected patient risk category, where the suggested treatment varies for each risk category. That is, rather than suggesting treatment for all risk categories, the scheme implements exclusions to provide the suggested treatment only when "truly" needed, potentially reducing the overall burden on the healthcare system by providing treatment only to patients who truly need it, which may significantly reduce the overall cost of treatment. It should be understood that the risk score ranges for various categories may be dynamic, i.e., the risk score ranges may change over time or for the purpose of averaging the overall cost of providing optimal treatment to patients.

[0027] Preferably, the proposed treatment may include at least one of a "pre-treatment" for the patient or a treatment product / scheme for the patient. Within the context of the present disclosure, the term "pre-treatment" may refer to, for example, any form of treatment provided to a patient before a difficult situation begins. Such pre-treatment may include anything from nutritional recommendations to hygiene instructions. Similarly, the term "treatment product / scheme" should be interpreted broadly to include any form or means suitable for use in connection with the active treatment of a patient, such as treating a patient's wound. With respect to wound products, by way of example, wound products may include, for example, wound dressings, bandages, topical applications, treatment methods combined with certain types of wound dressings, etc. Additionally, current or future treatment products / schemes are possible and within the scope of the present disclosure.

[0028] In some embodiments, the method further includes receiving, at the control unit, a second set of personal parameters indicative of the patient's condition after receiving the proposed treatment; using the control unit to determine a health improvement for the patient based on the first and second sets of personal parameters; and using the control unit to compare the determined health improvement for the patient to a predefined health improvement defined for at least one selected generic patient model.

[0029] According to the present disclosure, it may be possible to allow a time lag between the collection of the first and second sets of personal parameters, for example, between one hour and ninety days. However, the aforementioned time lag is merely an example, and the time lag may, of course, be shorter or longer. In one embodiment, it may be possible to allow the time lag to depend on the proposed treatment. Furthermore, it should be further understood that more sets of personal parameters than the first and second sets, such as a third set of personal parameters, may be used by the system, and in some cases, the time lag between the times at which data is collected may be fixed or variable.

[0030] Additionally, at least some of the generic patient models may comprise associated health improvements. That is, expectations for the patient may be defined for the generic patient model, with or without associated treatment recommendations / suggestions. In line with this embodiment, it may be possible to compare how the patient actually responded to the proposed treatment compared to the expected response (depending on the selected generic patient model). The comparison may then be used to further develop the scheme according to the present disclosure. That is, in some embodiments, it may be possible to "validate" the selected generic patient model, e.g., if the patient's health improvements substantially correspond to the predefined health improvements of the selected generic patient model. Validation may, in some cases, include updating the selected generic patient model with additional data or fine-tuning.

[0031] However, it may be equally useful to collect and store a patient's progress in situations where the patient's health progress deviates (sufficiently) from the predefined health progress of the selected generic patient model. In such situations, it may be possible, for example, to form a starting point (or a new generic patient model). Here, the deviating health progress may be considered a new situation compared to what was previously expected.

[0032] Subsequent information collected about patients may not only be used to update / adjust / validate the general patient model. Rather, the overall scheme according to the present disclosure may be used to allow different institutions and / or organizations to benchmark against each other. As such, in some embodiments, it may be desirable to ensure that information collected about patients remains strictly anonymous.

[0033] Updating / adjusting the generic patient model may, in one embodiment, include applying a machine learning process. That is, rather than having a physician (or technician) create a new generic patient model, the system itself may create such a model, or a model iteration of an already available generic patient model. For example, in some situations, a single generic patient model may be subdivided into two (or more) sub-models if additional data is provided that suggests a different assessment may be made in different situations. The machine learning process may, in some cases, be an unsupervised machine learning process, a supervised machine learning process, and / or be based on a convolutional neural network (CNN) or a recurrent neural network (RNN). Additional implementations are possible and within the scope of this disclosure.

[0034] According to another aspect of the present disclosure, there is further provided a computer-implemented method executed by a control unit for determining a patient's risk score, wherein the method includes receiving, at the control unit, a first set of personal parameters indicative of the patient's current or previous condition; using the control unit to match the first set of personal parameters to a plurality of different predefined generic patient models, each generic patient model having a predefined patient risk score; using the control unit to select a generic patient model that best suits at least the personal parameters; and determining the patient's risk score based on at least the selected generic patient model. This aspect of the present disclosure provides substantially the same advantages as those discussed above in connection with the previous aspect of the present disclosure. However, according to this aspect of the present disclosure, a slightly different approach is presented in which the personal parameters are directly matched to a plurality of different predefined generic patient models without including an individual patient model. Such an implementation may be preferable in some situations, for example, when the types of personal parameters are expected to be the same / approximately the same across all instances of collection.

[0035] According to yet another aspect of the present disclosure, there is provided a computer-implemented method, executed by a control unit, for reducing medical costs associated with a patient, comprising the steps of: receiving, at the control unit, a first set of personal parameters indicative of a patient's current or previous condition; using the control unit to form a personal patient model based on the first set of personal parameters; using the control unit to determine a degree of match between the personal patient model and each of a plurality of different predefined generic patient models, each generic patient model having a predefined patient risk score; using the control unit to select at least one generic patient model having a degree of match that exceeds a predetermined threshold; using the control unit to determine the patient's risk score based on the at least one selected generic patient model; using the control unit to define low-risk, medium-risk, and high-risk categories; using the control unit to assign the patient a risk category by comparing the patient's determined risk score with predefined risk score ranges for the various categories; and using the control unit to suggest a treatment to the patient only if the patient is assigned to the high-risk category. Additionally, this aspect of the present disclosure provides substantially the same advantages as discussed above in connection with the previous aspect of the present disclosure.

[0036] According to yet another aspect of the present disclosure, there is provided a computer system configured to determine a risk score for a patient, the computer system comprising a control unit configured to perform the following steps: receiving a first set of personal parameters indicative of a current or previous condition of the patient; forming a personal patient model based on the first set of personal parameters; determining a degree of match between the personal patient model and each of a plurality of different pre-defined generic patient models, each generic patient model having a pre-defined patient risk score; selecting at least one generic patient model having a degree of match that exceeds a predetermined threshold; and determining the patient's risk score based on the at least one selected generic patient model. This aspect of the present disclosure provides substantially the same advantages as discussed above in connection with the previous aspect of the present disclosure.

[0037] In possible embodiments of the present disclosure, the computer system is a portable electronic device such as, for example, at least one of a "dedicated electronic device," a mobile phone, a tablet, etc. Alternatively, the computer system may be, for example, a computer (e.g., a laptop) equipped with a camera as discussed above for acquiring images or video sequences of the patient's wound. The computer system may be arranged to be operated, for example, by a caregiver.

[0038] In a preferred embodiment of the present disclosure, the computer system comprises a graphical user interface (GUI) configured to provide instructions to a caregiver for obtaining a first set of patient parameters, which may then be configured to present information indicative of a risk score and / or risk category according to the processing steps discussed above.

[0039] According to yet another aspect of the present disclosure, there is provided a computer program product comprising a non-transitory computer-readable medium having stored thereon computer program means for operating a computer system configured to determine a patient's risk score. The computer system comprises a control unit. The computer program product includes: code for receiving, at the control unit, a first set of personal parameters indicative of the patient's current or previous condition; code for using the control unit to form a personal patient model based on the first set of personal parameters; code for using the control unit to determine a degree of match between the personal patient model and each of a plurality of different predefined generic patient models, each generic patient model having a predefined patient risk score; code for using the control unit to select, using the control unit, at least one generic patient model having a degree of match that exceeds a predetermined threshold; and code for using the control unit to determine the patient's risk score based on the at least one selected generic patient model. Additionally, this aspect of the present disclosure provides substantially the same advantages as discussed above in connection with the previous aspect of the present disclosure.

[0040] The control unit is preferably a microprocessor. Likewise, the computer readable medium may be any type of memory device, including a removable non-volatile random access memory, a hard disk drive, a floppy disk, a CD-ROM, a DVD-ROM, a USB memory, an SD memory card, or one of the similar computer readable media known in the art.

[0041] Additional features and advantages of the present disclosure will become apparent upon review of the appended claims and the following description. Those skilled in the art will appreciate that different features of the present disclosure may be combined to create embodiments other than those described below without departing from the scope of the disclosure.

[0042] Various aspects of the present disclosure, including its particular features and advantages, will be readily understood from the following detailed description and the accompanying drawings. [Brief explanation of the drawings]

[0043] [Figure 1] FIG. 1 conceptually illustrates a computer system in accordance with a presently preferred embodiment of the present disclosure. [Figure 2] FIG. 2 discloses a possible client device with a graphical user interface for applying the present concept. [Figure 3] FIG. 3 is a flow chart illustrating steps for carrying out a method according to a presently preferred embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0044] The present disclosure will now be more generally described below with reference to the accompanying drawings, in which presently preferred embodiments of the disclosure are shown. However, the present disclosure may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, such embodiments are provided for the purposes of wholeness and completeness, and to generally convey the scope of the disclosure to those skilled in the art. Like reference characters refer to like elements throughout.

[0045] Turning now to the drawings, and in particular to FIG. 1, there is conceptually shown a computer system 100 configured to determine a patient's risk score.

[0046] Computer system 100 comprises a server 106 that provides computing power and includes some form of control unit 108 disposed in communication with a database 110, and a client device 112 disposed in network communication with server 106, such as using the Internet. In Figure 1, client device 112 is operated by a caregiver (not shown). However, any user, such as, for example, any form of caregiver, may be authorized to operate client device 112.

[0047] Network communications may be wired or wireless communications, including, for example, wired connections such as a building's LAN, WAN, Ethernet network, IP network, and wireless connections such as WLAN, CDMA, GSM, GPRS, 3G mobile communications, 4G mobile communications, 5G mobile communications, Bluetooth, infrared, and the like.

[0048] 2, shown as a mobile phone, includes a graphical user interface (GUI) and a camera 204. The client device 112 also includes some form of control unit 206 that provides computing power. The GUI is preferably configured to present instructions and information to the caregiver, for example, to acquire images of the patient 106 using the camera 204, to receive additional patient data entered by the caregiver, and to display information regarding the patient's risk score and / or the patient's risk category.

[0049] In addition to the control unit 108, the control unit 206 may include a general-purpose processor, an application-specific processor, a circuit including processing components, a group of distributed processing components, a group of distributed computers configured for processing, etc. The processor may be or include any number of hardware components for performing data processing or signal processing or for executing computer code stored in memory. The memory may be one or more devices for storing data and / or computer code for completing or facilitating the various methods described herein. The memory may include volatile or non-volatile memory. The memory may include a database component, an object code component, a script component, or any other type of information structure for supporting the various activities of the description. According to example embodiments, any distributed or local memory device may be utilized with the systems and methods of the description. According to example embodiments, the memory is communicatively connected to the processor (e.g., via a circuit or any other wired, wireless, or network connection) and includes computer code for performing one or more processes described herein.

[0050] Furthermore, in one embodiment, computer system 100 is preferably implemented as a cloud-based computing system, where server 106 is a cloud server. As such, computing power may be divided among multiple different servers (not shown), and the location of the servers need not be explicitly defined. As noted above, computing power may also be distributed between servers and client devices.

[0051] Another advantage of using a cloud-based solution is the inherent redundancy that is achieved, i.e., by applying a distributed approach to the servers as well as the users / operators, security can be improved, since it would usually not be possible to attack (either physically or computationally) the specific production site that would hold both the servers and the users / operators in prior art solutions.

[0052] During operation of the computer system 100, and with further reference to FIG. 3, the process is initiated by a caregiver, for example, using the GUI of the client device 112, providing a first set of personal parameters indicative of the current or previous patient condition, which are in turn received (S1) by the server control unit 108 and / or the control unit 206 of the client device 112.

[0053] The control unit 108 / 206 may then form an individual patient model based on the first set of individual parameters (S2). As noted above, the individual patient model may in some embodiments be a pre-assessment of the patient, or in other embodiments may simply be a data string or vector holding the first set of individual parameters.

[0054] The individual patient model is then matched (S3) to a plurality of different pre-defined generic patient models, each of which has a pre-defined patient risk score. The plurality of different pre-defined generic patient models may, in some embodiments, be stored using the database 110 and / or stored using a memory module comprising the client device 112.

[0055] Matching between the individual patient model and multiple different predefined generic patient models determines the degree of match. In some embodiments, matching may be multidimensional matching, where a first set of individual parameters is matched with multiple different parameters associated with multiple different predefined generic patient models. In some cases, the first set of individual parameters may not necessarily correspond to the parameters of the multiple different predefined generic patient models, and the multiple different predefined generic patient models may not necessarily have the same types of parameters. Therefore, finding a match may require matching parameters in multiple dimensions. The degree of match should preferably take this into account, and in some embodiments may include determining the Euclidean distance of the different parameters.

[0056] Once the degree of match has been determined, at least one generic patient model is selected (S4). However, there is a prerequisite for selecting only one or more generic patient models with a degree of match above a predetermined threshold. As noted above, such a threshold is dynamic and may depend on the current implementation. Therefore, the threshold ranges from 0 to 100 (if the degree of match is in approximately the same range).

[0057] After selecting at least one generic patient model, a risk score for the patient can be determined (S5). The risk score determination will be based on at least one selected generic patient model, although a combination of multiple selected generic patient models may also be possible. In such implementations, different generic patient models may have different weights based on their individual degree of agreement.

[0058] The risk score may, in some cases, be normalized between 0 and 100. Of course, other ranges are possible and within the scope of the present disclosure. Additionally, the risk score may, in some cases, be used to determine a patient's risk category by allowing different ranges within the overall range of the risk score to correspond to different risk categories. In some embodiments, a risk score between 0 and 50 may correspond to a low-risk category, a risk score between 51 and 75 may correspond to a medium-risk category, and a risk score between 76 and 100 may correspond to a high-risk category. The ranges provided are for illustrative purposes only. It may be desirable to provide at least some form of treatment to patients in the high-risk category.

[0059] The present disclosure allows for the rapid and effective determination of a patient's risk score in an efficient manner that not only relies on patient-related data but also includes a matching scheme with multiple predefined generic patient models. Advantages of following this scheme include the possibility of reliably predicting a patient's likely future behavior and how to optimally address this potential behavior to minimize patient complications. Matching with different predefined generic patient models may also be considered a way to filter out possible variations in individual parameters for a patient, as such variations may have previously been determined to have little impact on the patient's future.

[0060] The control functions of the present disclosure may be implemented using existing computer processors, or by special-purpose computer processors for appropriate systems, incorporated for this or other purposes, or by hardwired systems. Embodiments within the scope of the present disclosure include program products that include machine-readable media for carrying or having machine-executable instructions or data structures stored thereon. Such machine-readable media can be any available medium that can be accessed by a general-purpose or special-purpose computer or other machine with a processor. By way of example, such machine-readable media can include RAM, ROM, EPROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage, or any other medium that can be used to carry or store desired program code in the form of machine-executable instructions or data structures and that can be accessed by a general-purpose or special-purpose computer or other machine with a processor.

[0061] Although a diagram may indicate a sequence, the order of steps may differ from that shown. Additionally, two or more steps may be performed simultaneously or with partial concurrent execution. Such variations will depend on the software and hardware systems selected and the designer's preferences. All such variations are within the scope of this disclosure. Similarly, software implementations may be achieved using standard programming techniques using rule-based and other logic to perform the various connection, processing, comparison, and decision steps. Furthermore, even though this disclosure has been described with reference to specific exemplary embodiments thereof, many different changes, modifications, and the like will be apparent to those skilled in the art.

[0062] Furthermore, variations of the disclosed embodiments can be understood and effected by those skilled in the art in practicing the present disclosure, from a study of the drawings, the disclosure, and the appended claims. Moreover, in the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. The inventions disclosed herein include the following: [Aspect 1] 1. A computer-implemented method performed by a control unit for determining a patient's risk score, the method comprising: - receiving at said control unit a first set of personal parameters indicative of a current or previous condition of said patient; - using the control unit to form an individual patient model based on the first set of individual parameters; - using the control unit to determine a degree of match between an individual patient model and each of a plurality of different pre-defined generic patient models, each of the generic patient models having a pre-defined patient risk score; - selecting, using said control unit, at least one generic patient model having a degree of match above a predetermined threshold; - using the control unit to determine the risk score for the patient based on the at least one selected generic patient model. [Aspect 2] 2. The method of aspect 1, wherein the selecting step comprises selecting the generic patient model having the highest degree of match. [Aspect 3] 2. The method of embodiment 1, wherein said risk score is determined based on a combination of at least two selected generic patient models. [Aspect 4] 4. The method of embodiment 3, wherein each of the at least two selected generic patient models has a weighting that is applied when determining the risk score. [Aspect 5] 5. The method of any one of aspects 1 to 4, wherein the individual parameters comprise clinical data of a plurality of the patient collected over a predetermined period of time. [Aspect 6] 6. The method of claim 5, wherein the clinical data includes at least patient vitals, hospitalizations, test results, and prescription medications. [Aspect 7] 7. The method of aspect 6, wherein the patient's vitals include at least one of heart rate data, electrocardiogram (EKG / ECG) data, respiratory rate data, patient temperature data, pulse oximetry data, and blood pressure data. [Aspect 8] - defining low risk, medium risk and high risk categories using said control unit; using the control unit to assign a risk category to the patient by comparing the determined risk score of the patient to predefined risk score ranges for various of the categories. [Aspect 9] The method of aspect 8, further comprising: - using the control unit to form a suggested treatment for the patient based on a selected patient risk category, the suggested treatment being different for each of the risk categories. [Aspect 10] 10. The method of embodiment 9, wherein the treatment for the patient is administered only if the patient is assigned to the high-risk category. [Aspect 11] - receiving, at the control unit, a second set of personal parameters indicative of the patient's condition after receiving the proposed treatment; - determining a health status of a patient based on the first and second sets of personal parameters using the control unit; 11. The method of any one of aspects 9 and 10, further comprising: using the control unit to compare the determined patient health improvement with a predefined health improvement defined for the at least one selected generic patient model. [Aspect 12] 12. The method of claim 11, further comprising updating at least one of the generic patient models based on a combination of the determined individual patient model and the results of the comparison of the health improvement. [Aspect 13] 13. The method of aspect 12, wherein updating the at least one of the generic patient models comprises applying a machine learning process. [Aspect 14] 14. The method of embodiment 13, wherein the machine learning process is an unsupervised machine learning process. [Aspect 15] 14. The method of embodiment 13, wherein the machine learning process is a supervised machine learning process. [Aspect 16] 14. The method of claim 13, wherein the machine learning process is based on a convolutional neural network (CNN) or a recurrent neural network (RNN). [Aspect 17] 1. A computer-implemented method performed by a control unit for determining a patient's risk score, the method comprising: - receiving at said control unit a first set of personal parameters indicative of a current or previous condition of said patient; - using the control unit to match the first set of individual parameters against a plurality of different predefined generic patient models, each of the generic patient models having a predefined patient risk score; - selecting, using said control unit, said generic patient model that best suits at least said individual parameters; - determining the risk score for the patient based at least on the selected generic patient model. [Aspect 18] 1. A computer-implemented method performed by a control unit for reducing patient-related medical costs, the method comprising: - receiving at said control unit a first set of personal parameters indicative of a current or previous condition of said patient; - using the control unit to form an individual patient model based on the first set of individual parameters; - using the control unit to determine a degree of match between the individual patient model and each of a plurality of different pre-defined generic patient models, each of the generic patient models having a pre-defined patient risk score; - selecting, using said control unit, at least one generic patient model having a degree of match above a predetermined threshold; - determining, using the control unit, the risk score for the patient based on the at least one selected generic patient model; - defining low risk, medium risk and high risk categories using said control unit; - using the control unit to assign a risk category to the patient by comparing the determined risk score of the patient with predefined risk score ranges for the various categories; - using the control unit to suggest a treatment to the patient only if the patient is assigned to the high-risk category. [Aspect 19] 1. A computer system configured to determine a risk score for a patient, the computer system comprising: a control unit; - receiving a first set of personal parameters indicative of a current or previous condition of said patient; - forming an individual patient model based on said first set of individual parameters; - determining a degree of match between the individual patient model and each of a plurality of different pre-defined generic patient models, each of the generic patient models having a pre-defined patient risk score; - selecting at least one generic patient model having a degree of match above a predetermined threshold; - determining the risk score for the patient based on the at least one selected generic patient model. [Aspect 20] 1. A computer program product comprising a non-transitory computer readable medium having stored thereon computer program means for operating a computer system configured to determine a patient's risk score, the computer system comprising a control unit, the computer program product comprising: code for receiving, at the control unit, a first set of personal parameters indicative of a current or prior condition of the patient; - code for using the control unit to form an individual patient model based on the first set of individual parameters; - code for using the control unit to determine a degree of match between the individual's patient model and each of a plurality of different pre-defined generic patient models, each of the generic patient models having a pre-defined patient risk score; and - code for using the control unit to select at least one generic patient model having a degree of match above a predetermined threshold; - code for using the control unit to determine the risk score for the patient based on the at least one selected generic patient model.

Claims

1. 1. A computer-implemented method performed by a control unit for determining a patient's risk score, the method comprising: - receiving at said control unit a first set of personal parameters indicative of a current or previous condition of said patient; - using said control unit to create an individual patient model, which is a data string or vector holding said first set of individual parameters; - using the control unit to determine a match between an individual patient model and a number of different parameters associated with each of a plurality of different pre-defined generic patient models, each of the generic patient models having a pre-defined patient risk score; - using said control unit to select at least one generic patient model having a degree of match above a predetermined threshold; - using the control unit to determine the risk score of the patient by matching the first set of personal parameters with the at least one selected generic patient model, each generic patient model being pre-formed and corresponding to a population of similar patients.

2. The method of claim 1 , wherein the selecting step comprises selecting the generic patient model with the highest degree of match.

3. The method of claim 1 , wherein the risk score is determined based on a combination of at least two selected generic patient models.

4. The method described in claim 3, wherein each of the at least two selected generic patient models has a weight that is applied when determining the risk score.

5. The method of claim 1 , wherein the individual parameters comprise clinical data of a plurality of the patients collected over a predetermined period of time.

6. The method of claim 5 , wherein the clinical data includes at least patient vitals, hospitalizations, test results, and prescription medications.

7. 7. The method of claim 6, wherein the patient vitals include at least one of heart rate data, electrocardiogram (EKG / ECG) data, respiratory rate data, patient temperature data, pulse oximetry data, and blood pressure data.

8. - defining low risk, medium risk and high risk categories using said control unit; - using the control unit to assign a risk category to the patient by comparing the determined risk score of the patient with predefined risk score ranges for various of the categories.

9. - receiving, at the control unit, a second set of personal parameters indicative of the patient's condition after receiving a treatment proposed for the patient based on a selected patient risk category; - determining a health status of a patient based on said first and second sets of personal parameters using said control unit; 2. The method of claim 1, further comprising: using the control unit to compare the determined patient health improvement with a predefined health improvement defined for the at least one selected generic patient model.

10. 10. The method of claim 9, further comprising updating at least one of the generic patient models based on a combination of the determined individual patient model and the results of the comparison of health improvement.

11. The method of claim 10 , wherein updating the at least one of the generic patient models comprises applying a machine learning process.

12. The method of claim 11 , wherein the machine learning process is an unsupervised machine learning process.

13. The method of claim 11 , wherein the machine learning process is a supervised machine learning process.

14. The method of claim 11 , wherein the machine learning process is based on a convolutional neural network (CNN) or a recurrent neural network (RNN).

15. 1. A computer system configured to determine a risk score for a patient, the computer system comprising: a control unit; - receiving a first set of personal parameters indicative of a current or previous condition of said patient; - forming an individual patient model, which is a data string or vector holding said first set of individual parameters; - determining a match between the individual patient model and a number of different parameters associated with each of a plurality of different pre-defined generic patient models, each of the generic patient models having a pre-defined patient risk score; - selecting at least one generic patient model having a degree of match above a predetermined threshold; - determining the risk score of the patient by matching the first set of individual parameters with the at least one selected generic patient model, each generic patient model being pre-formed and corresponding to a population of similar patients.

16. 1. A computer program comprising a non-transitory computer readable medium having stored thereon computer program means for operating a computer system configured to determine a patient's risk score, the computer system comprising a control unit, the computer program comprising: - code for receiving, at said control unit, a first set of personal parameters indicative of a current or previous condition of said patient; - code for using said control unit to create an individual patient model, said individual patient model being a data string or vector holding said first set of individual parameters; - code for using the control unit to determine a fit between an individual patient model and a number of different parameters associated with each of a plurality of different pre-defined generic patient models, each of the generic patient models having a pre-defined patient risk score; - code for using said control unit to select at least one generic patient model having a degree of match above a predetermined threshold; - code for using the control unit to determine the risk score of the patient by matching the first set of individual parameters with the at least one selected generic patient model, each generic patient model being pre-formed and corresponding to a population of similar patients.

Images