Patent Application
20180322256
Advances in pharmaceutical research have resulted in the availability of specialty drugs that give new hope to many patients who previously had lacked effective treatment options. However, many of these drugs are extremely costly, and leave insurers and other entities responsible for paying for patient care in the unenviable position of facing unsustainable costs, or denying access to powerful and beneficial treatments.
For example, specialty pharmaceutical drugs for use in the treatment of hepatitis C may cost from approximately ten thousand to approximately one hundred thousand dollars for a full course of treatment. Despite their nearly prohibitive costs, however, these specialty drugs can be life saving for some patients. As a result techniques for identifying those patients who may be at the greatest risk of developing advanced liver disease can help to enable access to specialty drug treatment by those most in need.
There are provided systems and methods for cirrhosis forecasting in human subjects, substantially as shown in and/or described in connection with at least one of the figures, and as set forth more completely in the claims.
The following description contains specific information pertaining to implementations in the present disclosure. One skilled in the art will recognize that the present disclosure may be implemented in a manner different from that specifically discussed herein. The drawings in the present application and their accompanying detailed description are directed to merely exemplary implementations. Unless noted otherwise, like or corresponding elements among the figures may be indicated by like or corresponding reference numerals. Moreover, the drawings and illustrations in the present application are generally not to scale, and are not intended to correspond to actual relative dimensions.
As noted above, specialty pharmaceutical drugs for use in the treatment of hepatitis C may cost from approximately ten thousand to approximately one hundred thousand dollars for a full course of treatment. However, and despite their nearly prohibitive costs, these specialty drugs can be life saving for the most vulnerable patients. As a result, techniques for identifying those patients who may be at the greatest risk of developing advanced liver disease can help to enable access to specialty drug treatment by those most in need.
The presence of cirrhosis in a patient having hepatitis C, for example, can foreshadow or accompany progression to advanced liver disease. The present application addresses the serious financial and ethical dilemmas posed by decisions to permit or deny patient access to extremely costly but highly therapeutic specialty drug treatments for hepatitis C by providing systems and methods for training a cirrhosis forecast model. According to various implementations, such a system and method may be used to forecast the likelihood that a patient diagnosed with hepatitis C will subsequently develop cirrhosis, and is thereby likely to progress to advanced liver disease, perhaps requiring liver transplant. Consequently, the systems and methods disclosed in the present application provide powerful tools for identify those patients who might benefit most from specialty drug treatment.
By way of definitions, it is noted that for the purposes of the present application, a “biologic” or “biological medical product” is any pharmaceutical drug manufactured in, extracted from, or at least partially synthesized from biological sources, in contrast to traditional pharmaceutical drugs that are chemically synthesized. In addition, as used herein, a “specialty drug” is a costly prescription medication, which may be chemically synthesized or produced as a biologic, and is used to treat complex, chronic conditions such as hepatitis C, cancer, multiple sclerosis, and rheumatoid arthritis, for example.
It is further noted that, as used in the present application, the term “subject” refers to a human subject, such as a human patient receiving medical evaluation and/or treatment. Moreover, as used herein, the terms “cirrhosis,” “cirrhotic,” and the like, refer specifically to fibrosis of the liver, commonly referred to as cirrhosis of the liver in human subjects.
As shown in
It is noted that although
According to the implementation shown in
Although client system 130 is shown as a personal computer (PC) in
It is noted that, in various implementations, cirrhosis forecast model 112 and/or cirrhosis prediction 116 when generated and/or produced using software code 110, may be stored in system memory 106 and/or may be copied to non-volatile storage (not shown in
Network communication link 222, and system 200 including computing platform 202 having hardware processor 204 and system memory 206 correspond in general to network communication link 122, and system 100 including computing platform 102 having hardware processor 104 and system memory 106, in
Client system 230 and display 238 correspond respectively in general to client system 130 and display 138, in
According to the exemplary implementation shown in
Client hardware processor 234 may be the central processing unit (CPU) for client system 230, for example, in which role client hardware processor 234 runs the operating system for client system 230 and executes software code 210b. In the exemplary implementation of
Also shown in
According to the implementation shown in
The systems for training a cirrhosis forecast model discussed above by reference to
Referring to
For example, medical data 160 may include the age, gender, race, general health, history of alcohol use, and genotype of subjects among cirrhotic subject population 152 and non-cirrhotic subject population 154, as well as the results of specific tests performed on those subjects. In addition, medical data 160 can include any other characteristics considered to be relevant to the development of cirrhosis, as well as whether or not a specific subject has been diagnosed as presently cirrhotic or non-cirrhotic. However, it is emphasized that medical data 160 omits any personally identifiable information (PII) of subjects among cirrhotic subject population 152 and non-cirrhotic subject population 154. As a result, all subjects among cirrhotic subject population 152 and non-cirrhotic subject population 154 used to generate and/or train cirrhosis forecast model 112/212 remain anonymous.
Flowchart 470 continues with assigning a subset of subjects among cirrhotic subject population 152 and/or non-cirrhotic subject population 154 as a training group for cirrhosis forecast model 112/212 (action 472). In some implementations, the training group may include subjects from both cirrhotic subject population 152 and non-cirrhotic subject population 154. That is to say, in some implementations, the training group may include cirrhotic subjects and non-cirrhotic subjects. Moreover, in some implementations, the training group may be randomly assigned from cirrhotic subject population 152 and/or non-cirrhotic subject population 154.
According to one exemplary implementation, the training group may include all of the subjects included in cirrhotic subject population 152 and/or non-cirrhotic subject population 154 except one subject. That one omitted subject may subsequently be used as a test subject to test the accuracy of cirrhosis forecast model 112/212. Assigning the subset of subjects among cirrhotic subject population 152 and/or non-cirrhotic subject population 154 as a training group for cirrhosis forecast model 112/212 may be performed by software code 110/210a/210b/310 of system 100/130/200/230/330, executed by hardware processor 104/204/234/334.
Flowchart 470 continues with identifying cirrhosis predictive parameters from medical data 160 for the training group (action 473). Medical data 160 for the training group may include many candidate parameters for use in forecasting cirrhosis, such as dozens, or even hundreds of candidate parameters, for example. In some implementations, action 473 may include determining the cirrhosis predictive parameters through analysis of medical data 160. In those implementations, identification of the cirrhosis predictive parameters from among the dozens or hundreds of candidate parameters provided by medical data 160 may be performed using any suitable statistical technique known in the art, such as receiver operating characteristic (ROC) analysis, for example.
In other implementations, however, the cirrhosis predictive parameters may be predetermined, and action 473 may include extracting those predetermined cirrhosis predictive parameters from medical data 160 for the training group. Identification of the cirrhosis predictive parameters may be performed by software code 110/210a/210b/310 of system 100/130/200/230/330, executed by hardware processor 104/204/234/334. For example, in one implementation, the following parameters may be identified as cirrhosis predictive parameters (pi) for use in training cirrhosis forecast model 112/212:
Flowchart 470 continues with generating cirrhosis forecast model 112/212 including a weighted combination of the cirrhosis predictive parameters (action 474). As shown in
As an example, cirrhosis prediction data structure 114/214 of initially trained cirrhosis forecast model 112/212 may include the following expression for the probability that cirrhosis is present or substantially imminent for a subject:
Where C is a constant, the pi are the cirrhosis predictive parameters identified in action 473, and the wi are their respective weighting factors determined using logistic regression.
As an even more specific example, when the exemplary parameters listed above on page 15 of the present application are a complete set of cirrhosis predictive parameters, K may take the form:
K=C+w
1
p
1
+w
2
p
2
+w
3
p
3
+w
4
p
4
+w
5
p
5
+w
6
p
6
+w
7
p
7
w
8
p
8
Flowchart 470 continues with producing, using cirrhosis forecast model 112/212, cirrhosis prediction 116 for one or more subjects omitted from the training group (action 475). As noted above, a subset of subjects from cirrhotic subject population 152 and/or non-cirrhotic subject population 154 are assigned to a training group for cirrhosis forecast model 112/212 in action 472. Those subjects not assigned to the training group may be used as a testing group for cirrhosis forecast model 112/212.
As further noted above, in one exemplary implementation, the training group may include all of the subjects included in cirrhotic subject population 152 and/or non-cirrhotic subject population 154 except one subject. Consequently, in that implementation, cirrhosis forecast model 112/212 may be tested using the one subject omitted from the training group. Cirrhosis prediction 116 may be produced by software code 110/210a/210b/310 of system 100/130/200/230/330, executed by hardware processor 104/204/234/334, and using cirrhosis forecast model 112/212 including cirrhosis prediction data structure 114/214.
Flowchart 470 continues with determining the accuracy of cirrhosis prediction 116 (action 476). In some implementations, the accuracy of cirrhosis prediction 116 may be determined through comparison of cirrhosis prediction 116 with medical data 160, which may include the cirrhosis status of the one or more subjects on which cirrhosis forecast model is tested in action 475. However, in other implementations, cirrhosis prediction 116 may be displayed to system user 140 for review and evaluation. For example, in those implementations, cirrhosis prediction 116 may be displayed to system user 140 through use of display 138/238/338. Moreover, in those implementations, system user 140 may provide accuracy evaluation data 124 rating the accuracy of cirrhosis prediction 116.
Consequently, in some implementations, determining the accuracy of cirrhosis prediction 116 may include receiving accuracy evaluation data 124 from system user 140. It is noted that accuracy evaluation data 124 may be received by system 130/230/330 as a direct input from system user 140. Alternatively, in some implementations, accuracy evaluation data 124 may be received by system 100/200 via communication network 120 and network communication links 122/222. The accuracy of cirrhosis prediction 116 may be determined by software code 110/210a/210b/310 of system 100/130/200/230/330, executed by hardware processor 104/204/234/334.
Flowchart 470 can conclude with adapting cirrhosis forecast model 112/212 based on the accuracy of cirrhosis prediction 116 (action 477). For example, software code 110/210a/210b/310 may be configured to engage in machine learning to adapt cirrhosis prediction data structure 114/214 in response to accuracy evaluation data 124 and/or its own determination of the accuracy of cirrhosis prediction 116. In some implementations, for example, cirrhosis prediction data structure 114/214, and thus cirrhosis forecast model 112/214 may be adapted by modifying one or more of the weighting factors wi appearing in Equation 2, above. That is to say, cirrhosis prediction data structure 114/214 may be an adaptive data structure. Cirrhosis forecast model 112/212 including cirrhosis prediction data structure 114/214 may be adapted by software code 110/210a/210b/310 of system 100/130/200/230/330, executed by hardware processor 104/204/234/334.
Thus, the systems and methods disclosed in the present application address serious financial and ethical dilemmas posed by decisions to permit or deny patient access to extremely costly but highly therapeutic specialty drug treatments for hepatitis C. By forecasting the likelihood that a patient having hepatitis C will develop cirrhosis, the disclosed systems and methods may be used to advantageously identify those patients who might benefit most from specialty drug treatment. Consequently, the systems and methods disclosed in the present application can play an important role in enabling seriously and/or chronically ill patients to have greater access to necessary treatments, thereby improving clinical outcomes for insurers, healthcare providers, and patients alike.
From the above description it is manifest that various techniques can be used for implementing the concepts described in the present application without departing from the scope of those concepts. Moreover, while the concepts have been described with specific reference to certain implementations, a person of ordinary skill in the art would recognize that changes can be made in form and detail without departing from the scope of those concepts. As such, the described implementations are to be considered in all respects as illustrative and not restrictive. It should also be understood that the present application is not limited to the particular implementations described herein, but many rearrangements, modifications, and substitutions are possible without departing from the scope of the present disclosure.
