CROSS REFERENCE TO RELAT ED APPLICATION

jftOfll J This application claims the priority benefit of U.S. Provisional Application No. 61/651 , 1 14 filed May 24, 2012 and entitled "Method and System For Ordering and Arranging a Data Set for a Severity and Heterogeneity Approach to Preventing Events

Including a Disease Strati fication Scheme," the content of which is hereby incorporated by reference in its entirety.

BACKGROUND

|¾002| Many professions rely on the decisions of skilled professionals In order to yield successful results. For example, during almost every patient visit, medical professionals must make decisions regarding whether or not to require a particular course of action for the patient. Based on the resuiis of a test, a physician will typically decide whether to order additional tests, whether to recommend a course of treatment, or whether to maintain the status quo. In highly complex professions such as medicine, professionals may be faced with a vast quantity of data and have difficulty determining which of the data is relevant to a decision,

[ _. 00031 When a patient is suspec ted of having a type of disease, e.g. heart disease or cancer, the clinical issues become fi nding the sub-tvpe of disease and recommending the optimal treatment plan. Examples of a disease sub-type for heart disease might be a patient with valvular disease or a patient with stenotic coronary arteries, each requiring different and unique treatment plans. For each suspected disease type, the patient typically undergoes a series of diagnostic tests, in some instances, the series of tests are designed to eliminate patients from undergoing the most expensive or most invasive test, i.e., if the "lower order" tests do not detect the presence of the disease sub-type, then the patient will not progress through the full series of tests. For a patient suspected of coronary artery disease, coronary artery catheterization is regarded by some a the definitive test (i.e. the "gold standard"), but the patient wil t typically undergo several blood level, tests and a myocardial perfusion test prior to catheterization. Only if the patient exhibits sufficient evidence of disease on these lower order tests will they progress to a catheterization, suite of tests.

|0004j When the catheterization test is performed, the lower order test results are often, discounted as the catheterization test, is regarded as the gold standard in this example _* thus resulting in the wasted expense for the discounted tests incurred by the patient or the patient's insurer. However, there are several drawbacks to this approach: I ) the lower order tests ma miss disease, and thus cause treatment to be withheld; 2) even the gold standard test may misdiagnose patients as having too low a level of disease to treat effectively; 3) the results from the lower order tests often do not contribute to the further management, of the patient after the gold standard test is performed; 4) often, if the suspicion of disease is high, not all the lower order tests are performed, again discounting their ability to contribute to further patient management; and 5} it is quite typical for many patients to have similar values of several lower order tests, e.g. systolic blood pressure and blood cholesterol levels in some mildly unfavorable range, and for it to be observed that the patients experience different outcomes despite these similarities,

{0005J Additionally, treatment of patients Is governed by evidence-based approaches typically overseen by professional bodies. However, the guidelines for initiating treatment are typically based on population averages, and thus some patients are over-treated, and are at risk of experiencing adverse side effects. Conversely, some patients are under- treated, and thus are at risk of suffering a severe adverse outcome that may have been treatable if the patients were properly treated. The typical paradigm for making decisions of when to initiate treatment -are based on the patient attaining a certain extent of disease, e.g., reaching a systolic blood pressure of 140 rami lg, reaching a blood lipid level of 100, or having other similar test .results that may indicate an extent of a disease. However, as discussed above, professionals may be faced with a vast quantity of data and have difficulty determining which of the data is relevant to a decision, related to disease extent, and thus may be prone to misdiagnosing and .fail to deliver a proper level of treatment.

SUMMARY

{0006) In one general respect, a first embodiment discloses a method for ordering and arranging a data set . The method includes accessing a da ta set of patient, data, wherein the data for each patient, includes a set of diagnostic test values and one or more outcomes; retrieving, from the data set, a first patient data set that corresponds to a first outcome;

selecting a first set of the diagnostic test values from the first patient date set; normalizing the first set of diagnostic test values to cover a range; ordering the normalised first set of diagnostic test values such that a low-frequency content plot is derived; creating a first equation describing the ordered data of the first patient data set f r the normalized first set of diagnostic test values; retrieving, -from the data set, a second patient data set that corresponds to a second outcome; selecting, a second set of diagnostic test value from the second patient data set; normalizing the second set of diagnostic test values to cover a range; ordering the normalized second set of diagnostic test values; creating a second equation describing the ordered data of the second patient data set for the normalized second set of diagnostic test values; and comparing the first and second equations,

{00071 In another general respect, a second embodiment discloses a system for ordering and arranging a data set. The system includes a processing device and a non- transitory computer readable medium operahS connected to the processor, the computer readable medium containing a set of instructions. The instructions are configured to cause the processing device to access a data set of patient data, wherein the data for each patient includes a set of diagnostic test values and one or more outcomes; retrieve, from the data set a first patient data set that corresponds to a first outcome; select a first set of the diagnostic test values from ihe first patient data set; normalize the first set of diagnostic test values to cover a range; order the normalized first set of diagnostic test values such thai a low- frequency content plot is derived; create a first .equation describing the ordered data of the first patient data set for the normalized first set of diagnostic test values; retrieve, from the data set, a second patient data set that corresponds to a second outcome; select a second set of diagnostic test value from the second patient data set; normalize the second set of diagnostic test values to cover a. range; order the normalized second set of diagnostic test, values; create a second equation describing the ordered data of the second patient data, set for the normalized second set of diagnostic test values, and compare the first and second equations.

BRIEF DESCRIPTION OF THE DRAWINGS

{0008J FIG. la illustrates a time ordered sampling of a sinusoidal waveform.

(00091 FIG. lb illustrates a random sampling of a sinusoidal waveform,

{0010f FIG, Ic illustrates a time ordered sampling of a sinusoidal waveform including interpolated data points,

{001 If FIG. 2 illustrates an ordered sampling of various disease components according to an embodimen

|0012 FIG. 3 illustrates an ordered sampling of various disease components for two groups overlaid on the same plot.

{0013 f FIG. 4 illustrates a hierarchal representation of disease levels according to an embodiment |0014j FIG. 5 iliustraies a sample radar plot for illustrating various disease components according io an embodiment

|0015! FiG. 6 iliustraies a corresponding set of radar plots for two groups of patients.

{001 ! F G. ? illustrates a corresponding set of normalized radar plot for two groups of patients according to an. embodiment.

|00 7) FIG. 8 illustrates a flowchart showing a process for determining and displaying two or more plots according to an embodiment.

{0018J FIG. 9 iliustraies various elements of a computing device for implementing various methods and processes described herein.

DETAILED DESCRIPTION

{0019{ This disclosure is not limited to the particular systems, devices and methods described, as these may vary. The terminology used in the description is for the purpose of describing the particular versions or embodiments only, and is not intended to limit the scope.

|t)O20| As used in this document, the singular forms *V "atv ^" and "the" include plural references unless the context clearly dictates otherwise. Unless defined otherwise, all technical and scientific terms used herein have the same meanings as commonly understood by one of ordinary skill in the art. As used in this document the term "com i in " means ''including, but not limited to." As used in this document the terms "sum, ^* ' ''product" and similar mathematical terrns are construed broadly to include any method or algorithm in which a single datum is derived or calculated from a plurality of input data.

(0O21f As used herein, the term "modality" refers to a mode, process or method of obtaining a set of data. For example, a modality may include a specific medical test or imaging process wherein one or more sets of patient specific data are obtained. j0022j Typically, when a human makes a decision, Che person relies upon some set of known tacts or concrete evidence. However, this set of known facts may not include all necessary information .required to make the decision. Thus, ihe person must use some amount of judgment when making the decision.

f f)023| One example of an area where a person such as a medical pro vider makes a decision without a full set of data is i determining disease sub-type and an. associated treatment plan. Using the processes and techniques as taught herein, a medical provider may make a more informed decision as to an appropriate treatment plan for a patient.

{0024J This disclosure is complementary to prior processes DICE and DICIE as taught in the related U.S. Patent. Application No, 1.3/289,335, the disclosure of which is hereb incorporated by reference, but is not dependent on such processes. The approach as taught herein, referred to as Severity and Heterogeneity Approach to Preventing Events (SHAPE) arranges data that forms part of the medical record of a patient in an order that allows characterization of that patient with regard, to disease sub-type and severity, i.e., the ordered data forms a characteristic shape associated with each disease sub-type within the context of a hierarchical scheme or stratification the diseases.

|0G25| By using the characterization scheme of SHAPE, derived from previously evaluated patient groups, a suitable treatment or treatment plan can be recommended for a current patient. Further, since the medical data are arranged in an ordered manner, and the plot of that data forms a characteristic shape, then there is the potential to eliminate some of the data and still adequately characterize the disease sub-type of the patient. This aspect either allows improved characterization of the patient in the absence of a complete set of diagnostic tests, or, more proactively, it allows an. optimal selection of tests, with the aim of omitting the most expensive or invasive tests by design, thereby bringing additional economic benefit. {0026J Typically, a patient is suspected of having a type of disease, e.g., heart disease or cancer, and the clinical issues then become finding the sub-type of disease and recommending the optimal treatment plan. Examples of a disease sub-type for heart disease might be a patient with valvular disease or a patient with stenotic coronar arteries, each requiring quite different treatment strategies. For each suspected disease type, the patient typically undergoes a series of diagnostic tests. In some instances, the series of tests is designed to eliminate patients from undergoing the most expensive or most invasive test, i.e., if the "lower order' tests do not detect the presence of the diseas sub-type, then they will not progress through the full series of tests.

[002T| in the separate area of signal processing there is often the observation that ordering data in some logical manner better represents the data than a random presentation, and that this ordering of dat helps reduce the number of variables needed to characterize the data. F G i a illustrates a sampling 101 of a sinusoidal waveform. If the sampled data are time ordered, then describing the waveform can be done by using the parameters of

frequency and amplitude, thereby fully and uniquely describing the multiple data points using only two parameters. Conversely, if the same data points were presented in a random manner, as shown in FIG l , a resulting sampling 102 would not be easily characterized, and more Importantly , a different arranging of the data would result in a different characterization, effectively invalidating any characterization that sought to simplify the data (i.e. reduce the number of variables describing the data),

{0028 j Referring hack to FIG. I a (I .e. the sequentially time ordered data), if some limited number of data points were omitted, as shown in sample 103 illustrated in FIG, l e, it may a simple matter to interpolate data on either side of the given points and derive the missing values with good accuracy. For example, data point 105 may be derived based upon an interpolation of know data points 104 and 166. Similarly, data point 11.0 may be derived based upon an interpolation of known data points 109 and 1 1 1. The ability to predict values may reflect the fact that the data is band- limited, Le., there is an upper limit to the frequency of the data, which allows a lower sampling rate (i.e., can accommodate missing data) to adequately represent the full data sex.

|f)029J When determining diagnostic medical data, as indicated above, in common practice, the order in which medical tests are performed is governed by the expense and degree of invasiveness of ihe tests. Further, after a series of tests are performed, more clinical weight is attached to some tests over others, and much information is discarded at the stage of deciding a treatment course for the patient. In SHAPE, the medical data may be arranged in a logical order such that the band-width describing the data is limited and optionally minimized. In this way, less data (i.e., fewer tests) may be acquired or missing dat may be accommodated- In one embodiment, the wide variet of data that is entered into SHAPE may be normalized into a common range (e.g. expressing the data using percentages of some physiologic range, such that data typically range .from. 0 to 100%). For instance, one of the input variables could, be a Dl ^' CE-augmented reading of a diagnostic test The ordering of the data may be found at an initialization stage by exploration of historical data sets, especially if the treatment and patient outcome of those tes ts were known. Normalized data from multiple diagnostic tests ma be ordered for each patient, and arranged until the shape described by the ordered plot of data underwent some minimization of bandwid th as described above. The data may be plotted graphically for easy human interpretation, but more importantly, the relationship between ordered data is known to the computer system and can be defined by an equation. For example, FIG. 2 illustrates an exemplary sample 201 showing a SHAPE ordered data set for a series of disease components 202 as normalized on a severity scale 283 ranging from 0 to 0.8. The disease components 202 ma include, for example, myocardial perfusion imaging (MPi), coronary artery disease (CAD), blood lipid levels, diabetes, blood pressure (BP), angina, stress, energy model of the heart (energy ) and ejection function of the heart (EF). It should be noted the disease components 202 as shown in FIG. 2 are by way of example only.

ΙΘ03ΘΙ The ordering process may be repeated for similar data from multiple patients, untii a consensus of ordering was achieved- The intention is that the ordering would identify several distinct groups of patients, each group having a distinct shape of the magnitude of ordered variables (as described by an equation obtained for example by a least-squares fitting criteria). From the observation of outcomes of patients, it would be expected that a patient with some characteristic data shape would experience better outcomes for on particular ireaimeni compared to another applied to patients characterised by the same data shape of curve, FIG, 3 illustrates a sample 301 where two patient groups are overlaid on the same sample. In this example, both groups have similar CAD and energy model disease

components. However, when looking at. the entire sample of ordered data 301, each grou is characterized by a distinctly difference shape, and thus ma require distinctly difference treatments despite similarities in key physiologic variables. .Also from this ordered representation of the data, it can be appreciated that to distinguish the two groups of patients, in the tests corresponding to the two points, if the curves cross each other can be omitted from consideration, since they do not provide differentiating data, and further they are not needed to complete the band-limited description of the two groups of patients,

(0031 _. 1 After having made these observations and identified the patient groups and effective treatments, thereafter, the SHAPE ordering of data could be used to guide future therapy and better direct patients as to which diagnostic test would add the most value to characterize their disease sub-type.

}0032j Another aspect of the SHAPE too! for correctl determining a treatment plan may be to arrange diseases in a hierarchical scheme or to stratify the diseases, FIG. 4 illustrates an exemplary stratification 401 of an exemplary disease that allows clinical and patient management decisions to be modified based on the hierarchical ordering. To establish what the hierarchical relationships are may include investigation of clinical data sets during an initialisation step. In a clinical data set, a medical practitioner may identify certain components of this hierarchical structure for cardiovascular disease. Establishing the components of the hierarchical structure may be simplified by use of prior art techniques such as DIC E and DICTE (without use of these prior techniques, constructing a hierarchical ordering would typically require very large populations). The hierarchal scheme as described herein is d irected toward the use of the hierarchical ordering of disease states in decisionmaking. By recognizing that there is a hierarchical structure for a particular disease may be used to alter the disease extent required before a treatment is initiated. For example, a patient displaying an absence of a level 1 disease sta e (e.g. no myocardial perfusion defect) may have medication prescribed only when the lipid levels (a level 2 disease state) reach a high extent, e.g. ! 50. Conversely, a patient with a level 1 disease state (e.g. a myocardial perfusion delect present) may have medication prescribed when the lipid, levels (a level 2 disease state) reach a Sower extent, e.g. 80. Further, the presence of a disease state at the same hierarchical level as elevated lipids (e.g. presence of diabetes) would not modify the threshold at which lipid reduction therapy is initiated.

(0033 The SHAPE and stratification techniques as discussed herein involve the ordering and hierarchal structurin of disease components associated with a major disease category, e.g., ischemic heart disease.. These concepts may be intuitively represented in a one dimensional ( I D) plot or a two-dimensional (2D) radar plot depending on preference. In the radar plot, any essential, features may be retained in a visual representation of the data. The hierarchal nature of the data may be represented as a multiplication effect on the plot.

Additionally, data may be further compressed by integrating the area in the radar plot. f0034j As shown in FIG. 3, two groups may he represented in a linear plot of the ordered SHAPE variables, the plot showing the- differences between the two groups. As shown i FIG, 5, a 2D radar plot 501 may be used to represent the sam two groups. The radar plot 501 may be used to identity patients who may benefit from medical protection via a treatment plan or other similar treatment technique. The occurrence of a major ad verse cardiovascular event (MACE) may be concentrated in. patients with a large SHAPE area (as determined by an integration of the area of the radar plot) and who are not on protective medicine,

{0O35J Occasionally, a radar plot for a particular group may be inconclusive based solely upon the SHAPE ordering of information. For example, as shown in FIG. 6, radar plot 601 represents a group of patients having no medical protection and who have not experienced a MACE. Conversely, radar plot 602 represents a group of patients that have no medical protection who have experienced a MACE. In both plots 601 and 602, the perfusion nodes are both close to I . Thus, using perfusion s an indicator for these two groups may be inconclusive. Ho wever, and more fundamentally, as shown by the stratification of the medical data, there are hierarchal differences between these two groups. As such, the hierarchal stratification may be incorporated into the radar plots as a multiplication factor, rather than a node on the same plot.

|0O36| For example, to calculate the multiplication factor, the following criteria may be used: ( 1 ) Incorporate the ratio of event rate between patients having a perfusion and those not having a perfusion, in this example, the ratio would be 0.17/0.04 4.6. (2) incorporate the ratio of the mean area of the positive perfusion plots and the negative perfusion plots ^::: .2.6. (3) Determine the hierarchal factor by dividing the event rate ratio by the mean are ratio, 4,6/2.6 ~ 1.7. It should be noted these criteria are shown by way of example only. |O037j By determining a multiplication factor, the process may allow for shifting and personalization of thresholds directed to individual patients for determining a treatment plan. FIG, 7 shows a set of resulting radar plots incorporating the determined multiplication factor. A plot 701 shows patients with a perfusion, de ect present and plot 702 shows patients with no perfusion defect. Patients who experienced a MACE have a high area in the

.hierarchai representation (graphically represented in. the radar plot). Thus, by comparing a patient who has experienced a perfusion defect's information (either det.emi.ined through various tests or through the SHAPE process as described above) to a radar plot such as plot 701., the likelihood of the patient experiencing a MACE may be quickly determined with a inimal set of data.

|0038 By visually presenting the SHAPE data in a radar or I D plot the data may indicate a certain dominant, axis to the disease, as well as allowing overall severity of the disease to he visually identified. More importantly, this axis may be found by analysis of the equation describing the SHAPE ordered data. Further, the area, of the radar plot may be integrated geometrically to identify disease severity. Additionally, the distribution of events among patients may be analyzed to determine high and low risk groups. For example, critical events may be experienced by patients with a hierarchai SHAPE plot area above 0.05. This threshold may indicate the level where, for patients below the threshold, there may be no i mpact in event occurrence for those on or off medication. For those above the threshold, there may be a dramatically higher percentage of those off medication experiencing a critical event than those on medication. Thus, the SHAPE ordering of data, combined with the stratification of the data into a hierarchai structure, summarizes the severity of each individual patient. This information m y be optionally plotted into a 1 D linear or 2D radar plot for visual presentation to die human. SHAPE ordering and stratification also provides a framework to compare groups of patients, such as those that may appear in a clinical study. |0039j FIG. 8 illustrates a flowchart showing a process for performing the SHAPE and stratification processes as discussed above to provide a comparison ordering of data for further analysis. For example, a medical professional may use the process as illustrate in FIG. 8 to determine a treatment plan for an individual patient. The medical professional and/or a processor may access 802 one or .more data sets. The data sets may include patient records, diagnostic values for each patient, treatment plans and one or more outcomes resulting from, those treatment plans. The data sets may be represented as data listings, ID representations such as the SHAPE samples shown in FIGS, 3 and 4, or may be stored in other similar data structures. From the accessed 802 records, the medical professional may retrieve 804 a data set related to a first group of patients that have achieved a first outcome. The medical professional and/or processor may select 806 one or more diagnostic values from the first patient data set and the processor may normalize the set of diagnostic values to cover a specific range (such as values of 0 to 1) and order 808 the normalized test results to derive the values for a low- frequency content plot such as one of those shown in FIG. 7, Optionally, the medical professional may further stratify the data set according to the hierarehal scheme as discussed, above.

(00 01 The processor may then create 810 a first plot for the first patient based upon the normalized and ordered diagnostic test results. As shown above, the first plot may be a 2D radar plot including a multiplication factor for normalizing the data shown therein,

(0041 _. 1 In order to provide a means for comparing the first patient's information, the medical professional or processor may create 818 a similar second plot for a second patient's data set. Like before, the processor raay retrieve 812 a data set related to the second patient and the second patient's present course of treatment. The medical professional and/or the processor may select 814 one or more diagnostic test results r m: the second patient data set and normalize and order 816 Che second set of test results. Optionally, the system may further stratify the data set according to the hierarchal scheme as discussed above.

|0042j The processor may the create MS the second plot for the second patient based upon the second patient set's ordered, normalised test results. The two plots may then be displayed 820 for comparison purposes, allowing the medical professional to evaluate and analyze the current treatment plan for the first patient based upon the historic records associated with the second patient,

0043| It should be noted that while the process as illustrated in FIG, 8 is shown in a linear path, several of the steps may be performed simultaneously. For example, the medical professional may create both the first plot (81.0) and the second plot (818) simultaneously.

|004 FIG. depicts a block diagram of internal hardware that may be used to contain or implement various components to perform the processes illustrated in the previous figures. A bus 900 serves as the main information highway interconnecting the other illustrated components of the hardware, CPU 90S is the central processing unit of the system, performing calculations and logic operations required to execute a program. CPU 905, alone or in conjunction with one or more of the other elements disclosed in FIG. 9, is an illustration of a processing device, computing device or processor as such terms are used within this disclosure. Read only memory (ROM) 910 and random access memory (RAM) 915

constitute examples of memory devices.

{0045J A controller 920 interfaces with one or more optional memory devices 925 to the system bus 900. These memory devices 925 may include, for example, an external or internal DVD drive, a CD ROM drive, a hard drive, flash memory, a USB drive or the like. As indicated previously, these various drives and controllers are optional devices.

Additionally, the memory devices 925 may be configured to include individual flies for storing any software modules or instructions, auxiliary data, common files for storing groups of .results or auxiliary, or one or more databases for storing the result information, auxiliary data, and related information as discussed above.

|0046| Program -instructions, software or interacti ve modules for performing the processes as discussed above may be stored in the ROM 910 and/or the RAM 915.

Optionally, the program instructions may be stored on a tangible computer readable medium such as a compact disk, a digital, disk, flash, memory, a memory card, a USB drive, an optical disc storage medium, and/or other recording medium,

£©047f An optional display interface 930 ma permit information from the bus 900 to be displayed on the display 935 in audio, visual, graphic or alphanumeric format, The information may include information related various data. sets. Communication with external devices may occur using various communication ports 940. A communication port 940 may be attached to a. communications network, such as the Internet or an intranet,

jiHMSj The hardware may also include an interface 945 which allows for receipt of data from input devices such as a. keyboard 950 or other input device 955 such as a mouse, a joystick, a touch screen, a remote control, a pointing device, a video input device and/or an audio input device.

(00 1 Several of the above-disclosed and other features and functions, or

alternatives thereof, may be combined into many other different systems or applications. Various presently unforeseen or unanticipated alternatives, modifications, variations or improvements therein may be subsequently made by those skilled in the art, each of which is also intended to be encompassed by the disclosed embodiments.