Candidate screening for targeted therapy

Abstract

Techniques for candidate screening for targeted therapy are provided. A plurality of medical codes corresponding to the patient billing data is identified. Generating a candidate classification model for targeted therapy by selecting a set of medical codes from the plurality of medical codes and assigning a weight to each of the selected medical codes; the candidate classification model is applied to patient billing data for a plurality of patients to generate a set of targeted therapy scores. The patient billing data includes an association between one or more of the plurality of patients and one or more of the plurality of medical codes. Each targeted therapy score corresponds to one of the plurality of patients, and each targeted therapy score is calculated based on weights for one or more medical codes associated with the one patient.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of the filing date of German Patent Application No. 102022116219.2, filed on June 29, 2022. Technical Field

[0003] The present disclosure generally relates to systems and methods for analyzing standardized administrative data to identify potential candidates for targeted therapy. Background Art

[0004] Hospitals and clinicians are often faced with the challenge of identifying appropriate treatments for patients. Hospitals and other medical institutions often use large clinical information systems (CIS) to manage patient medical data, which collect and store patient information generated during the treatment process. Although CIS can manage a large amount of data, the data may not be easily accessible across departments. Furthermore, the data generated by different care providers, departments and medical institutions may be inconsistent in content and format, thereby causing challenges when accessing, interpreting or analyzing data across care providers, departments or medical institutions. Further, laws, regulations, standards and other rules may limit the sharing of patient information. Care providers cannot easily access each of the relevant data managed by medical institutions to determine treatment plans for patients. Due to the lack of accessible information, care providers may not be able to identify candidate patients for targeted treatment.

[0005] Therefore, it would be advantageous to have access to available data to identify patients who are candidates for particular treatments in a manner consistent with applicable laws, regulations, standards, or other rules. Summary of the invention

[0006] The following claims may be regarded as the content of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 A block diagram of an exemplary computer system according to an aspect of the present disclosure is illustrated.

[0008] Figure 2 A flow chart of an exemplary method for candidate screening for targeted therapy is illustrated.

[0009] Figure 3 A flow chart of an exemplary method for updating a candidate classification model for targeted therapy is illustrated.

[0010] Figure 4 is a block diagram illustrating a computer system upon which one or more examples may be implemented. DETAILED DESCRIPTION

[0011] In the following description, many specific details are set forth to provide a thorough understanding of the present disclosure. However, it is apparent that these embodiments can be implemented without these specific details. The following detailed description describes exemplary embodiments, and the disclosed features are not intended to be limited to the combinations explicitly disclosed. Therefore, unless otherwise stated, the features disclosed herein can be combined to form additional combinations that are not shown for the sake of brevity.

[0012] It should be further understood that: unless otherwise expressly stated, the term "or" can be inclusive or exclusive; the term "set" can include zero, one, or two or more elements; the terms "first", "second", "some" and "particular" are used as naming conventions to distinguish elements from each other, and unless otherwise stated, do not imply the order, time or any other characteristics of the referenced items; the term "and / or" used herein refers to and covers any and all possible combinations of one or more associated listed items; the terms "comprises" and / or "comprising" specify the presence of the described features, but do not exclude the presence or addition of one or more other features.

[0013] A "computer system" refers to one or more computers, such as one or more physical computers, virtual computers, and / or computing devices. For example, a computer system may be or may include one or more server computers, desktop computers, portable computers, mobile devices, special-purpose computing devices with processors, cloud-based computers, groups of cloud-based computers, virtual machine instances, and / or other computing devices. A computer system may include another computer system, and a computer device may belong to two or more computer systems. Unless explicitly stated otherwise, any reference to a "computer system" may refer to one or more computers. When a computer system performs an action, the action is performed by one or more computers of the computer system.

[0014] A "device" may be a computer system, hardware, and / or software stored in or connected to a memory and / or one or more processors on one or more computers. Alternatively or in addition, a device may include dedicated circuitry. For example, a device may be hardwired or permanently programmed to support a set of instructions to perform the functions discussed herein. A device may be a stand-alone component, work in conjunction with, contain, and / or belong to one or more other devices.

[0015] A "component" may be hardware and / or software stored in or connected to a memory and / or one or more processors on one or more computers. Alternatively or in addition, a component may include a dedicated circuit. For example, a component may be hardwired and / or permanently programmed with a set of instructions to perform the functions discussed herein. A component may be an independent component, work in conjunction with one or more other components, contain one or more other components, and / or belong to one or more other components.

[0016] General Overview

[0017] The present disclosure generally describes a technique for screening candidates for targeted therapy by generating a candidate classification model and applying the candidate classification model to patient billing data. Patient billing data is a form of hospital data that is typically standardized to ensure accurate and consistent medical billing settlement. For example, insurance companies and government entities often require the use of standardized code systems for medical services and surgeries. The candidate classification model is configured based on a set of medical codes related to targeted therapy. For example, the candidate classification model can use selected medical codes to model inclusion criteria and / or exclusion criteria for targeted therapy. The candidate classification model is applied to a set of patient billing data to identify candidate patients who may benefit from targeted therapy. For example, a medical institution may submit a set of data representing a group of patients to screen the group of patients. The care provider of the candidate patient may further evaluate the candidate patient based on the results of the screening. Data and privacy protection techniques can be implemented to manage compliance with privacy and authorization requirements. For example, pseudonymization can be applied during the screening process to protect the privacy of the group of patients.

[0018] One aspect of the present disclosure relates to a computer system, which includes one or more hardware processors and at least one memory storing one or more instructions. When the instructions are executed by the one or more hardware processors, the one or more hardware processors identify a plurality of medical codes corresponding to patient billing data; generate a candidate classification model for targeted therapy by selecting a set of medical codes from the plurality of medical codes and assigning a weight to each selected medical code; receive patient billing data for a plurality of patients, the patient billing data including associations between one or more patients in the plurality of patients and one or more medical codes in the plurality of medical codes; apply the candidate classification model to the patient billing data to generate a set of targeted therapy scores, each of which corresponds to one of the plurality of patients and is calculated based on the weight of one or more medical codes associated with the one patient; and provide targeted therapy candidate data generated based on the set of targeted therapy scores.

[0019] In some examples, the targeted therapy candidate data identifies one or more candidate patients with a targeted therapy score that satisfies a threshold.

[0020] In some examples, the targeted therapy candidate data includes an individual report of at least one patient among a plurality of patients.

[0021] In some examples, the targeted therapy candidate data includes summary data for one or more targeted therapy classifications in one or more groupings among a plurality of patients.

[0022] In some examples, receiving the patient billing data includes receiving the patient billing data from a medical institution.Alternatively or additionally, providing the targeted therapy candidate data may include providing a set of targeted therapy scores to the medical institution.

[0023] In some examples, the patient billing data is received in a de-identified form.Alternatively or additionally, the targeted therapy candidate data may reference an individual patient among a plurality of patients by a pseudonym.

[0024] In some examples, generating a candidate classification model for a targeted therapy includes assigning one or more negative weights to one or more medical codes corresponding to one or more exclusion criteria for the targeted therapy, and assigning one or more positive weights to one or more medical codes corresponding to one or more inclusion criteria for the targeted therapy.

[0025] In some examples, when the one or more hardware processors execute the instructions, the instructions cause the one or more hardware processors to provide a user interface for adjusting one or more weights assigned to one or more medical codes in a set of medical codes.

[0026] In some examples, generating a candidate classification model includes assigning one or more weights of the candidate classification model based on a training data set using machine learning techniques, the training data set comprising: input data comprising associations between medical codes and a plurality of sample patients; and output data comprising a sample targeted therapy score for each of the plurality of sample patients.

[0027] In some examples, when the one or more hardware processors execute the instructions, the instructions cause the one or more hardware processors to receive care provider review data corresponding to one or more targeted therapy scores in a set of targeted therapy scores and update a candidate classification model by adjusting one or more weights assigned to one or more medical codes in a set of medical codes based on the care provider review data.

[0028] In some examples, when the one or more hardware processors execute the instructions, the instructions cause the one or more hardware processors to identify one or more untreated medical codes in the patient billing data that are not in the plurality of medical codes, and to update a candidate classification model by selecting one or more untreated medical codes from the one or more untreated medical codes and assigning a weight to each of the selected untreated medical codes.

[0029] In some examples, generating the candidate classification model includes grouping two or more medical codes in a set of medical codes into a code group and assigning a single weight to the code group to indirectly assign weights to the two or more medical codes in the code group.

[0030] In some examples, the targeted therapy is left atrial appendage (LAA) occlusion including LAA occluder implantation.

[0031] One aspect of the present disclosure relates to a method, which includes identifying a plurality of medical codes corresponding to patient billing data; generating a candidate classification model for targeted therapy by selecting a set of medical codes from the plurality of medical codes and assigning a weight to each selected medical code; receiving patient billing data for a plurality of patients, the patient billing data including associations between one or more patients in the plurality of patients and one or more medical codes in the plurality of medical codes; applying the candidate classification model to the patient billing data to generate a set of targeted therapy scores, each targeted therapy score in the set of targeted therapy scores corresponding to one of the plurality of patients and calculated based on a weight for one or more medical codes associated with the one patient; and providing targeted therapy candidate data generated based on the set of targeted therapy scores; wherein the method is performed by one or more processors.

[0032] In some examples, receiving the patient billing data includes receiving the patient billing data from a healthcare provider. Alternatively or additionally, providing the targeted therapy candidate data set includes providing a set of targeted therapy scores to the healthcare provider. Alternatively or additionally, the patient billing data is received from the healthcare provider in a de-identified form. Alternatively or additionally, the targeted therapy candidate data references an individual patient from the plurality of patients by a pseudonym.

[0033] In some examples, generating a candidate classification model includes using machine learning techniques to assign one or more weights of the candidate classification model based on a training data set, the training data set comprising: input data including associations between medical codes and a plurality of sample patients; and output data including a sample targeted therapy score for each of the plurality of sample patients.

[0034] In some examples, the method further includes receiving care provider review data corresponding to one or more targeted therapy scores in the set of targeted therapy scores, and updating the candidate classification model by adjusting one or more weights assigned to one or more medical codes in the set of medical codes based on the care provider review data.

[0035] One aspect of the present disclosure relates to a non-transitory computer-readable medium storing instructions. When the instructions are executed by one or more processors in a network system, the instructions cause the network system to identify a plurality of medical codes corresponding to patient billing data; generate a candidate classification model for targeted therapy by selecting a set of medical codes from the plurality of medical codes and assigning a weight to each selected medical code; receive patient billing data for a plurality of patients, the patient billing data including associations between one or more patients in the plurality of patients and one or more medical codes in the plurality of medical codes; apply the candidate classification model to the patient billing data to generate a set of targeted therapy scores, each of which corresponds to one of the plurality of patients and is calculated based on the weight of one or more medical codes associated with the one patient; and provide targeted therapy candidate data generated based on the set of targeted therapy scores.

[0036] In some implementations, the various techniques described herein may achieve one or more of the following advantages: screening for targeted therapy is performed efficiently in a scalable manner; patient billing data from multiple sources is converted into usable data that enables medical institutions and care providers to improve patient care; patient outcomes may be improved; candidate classification models may be updated based on additional data and / or research; data processing and analysis are standardized and simplified; compliance with laws and regulations related to privacy and health is enhanced; generation of statistical data allows for study of population trends; medical institutions may use statistical data about patient populations to develop budgets, staff, build infrastructure, and make other spending decisions; and candidate classification models for targeted therapy may be iteratively improved based on care provider reviews, treatment outcomes of targeted therapies, or other additional data. Additional features and advantages will be apparent from the specification and drawings.

[0037] System Overview

[0038] Figure 1A block diagram of an exemplary computer system according to one aspect of the present disclosure is illustrated. The computer system 100 includes a targeted therapy analysis system 120, a medical institution computer system 110, and one or more care provider devices 112. The targeted therapy analysis system 120 and the medical institution computer system 110 can communicate via one or more networks. The network may include one or more local area networks (LANs) and / or one or more wide area networks (e.g., the Internet).

[0039] The medical institution computer system 110 includes one or more computers and / or devices owned and / or controlled by a medical institution. A medical institution is an entity that provides health care services, such as a hospital, clinic, office, group, or other entity that provides health care services. A medical institution may include a group of medical institutions. The medical institution computer system 110 stores patient information generated in the process of providing treatment to patients of the medical institution. In some examples, the medical institution computer system 110 includes a clinical information system (CIS) that manages clinical data.

[0040] The medical institution may include one or more care providers, such as doctors, nurses, therapists, technicians, and / or other health care professionals. A care provider is an individual who can generate patient data in the process of providing treatment to patients of the medical institution. A care provider can operate one or more care provider devices 112 to input patient data. Although the care provider devices 112 are illustrated as being different from the medical institution computer system 110, they may include a computing device that is part of the medical institution computer system 110, such as a device owned by the corresponding medical institution. As an alternative or in addition, one or more care provider devices 112 can be devices owned and / or controlled by one or more care providers.

[0041] The medical organization computer system 110 includes a medical management system 102. The medical management system 102 manages various aspects of the medical billing process, such as billing medical care services provided to patients, maintaining rates for medical care services, insurance submissions, invoicing, payments, and / or other aspects of medical billing.

[0042] The medical institution computer system 110 can maintain medical code data 104. The medical code data 104 may include medical codes organized in a standardized code system. For example, the World Health Organization (WHO) maintains the International Classification of Diseases (ICD), which is a standardized system of diagnostic codes representing states and diseases, related health problems, abnormal findings, signs and symptoms, injuries, external causes of injuries and diseases, and / or social conditions. ICD includes different versions (e.g., ICD-10, ICD-11). Many countries use modified versions of ICD tailored to their health system infrastructure (e.g., ICD-10-CM and ICD-10-PCD in the United States, ICD-10-GM in Germany, and ICD-10-FR in France). Other standardized code systems may also be used (e.g., German surgery and operation codes (Operat ionen-undProzedurenschlüssel) (OPS), and French general classification of medical operations (Classification Communedes Actes Médicaux) (CCAM).

[0043] The medical management system 102 uses the medical code data 104 to manage billing for health care services provided to patients (e.g., physical examinations, tests, imaging, other diagnostic procedures, facilities, surgical procedures, therapeutic procedures, rehabilitation procedures, other procedures, medications, equipment, treatment methods, and / or other medical services or items provided by a medical institution). In some examples, the medical management system 102 uses the medical codes in the medical code data 104 as billing codes. Alternatively or additionally, the medical management system 102 can use one or more custom billing codes, each of which is linked to one or more medical codes in the medical code data 104.

[0044] The medical institution computer system 110 may store patient data (e.g., patient billing data 106 and patient medical data 108). The patient billing data 106 includes an association between one or more patients of the medical institution and one or more medical codes. For example, when a care provider submits a report indicating that a particular patient received a particular treatment for a particular diagnosis from a care provider device 112, the medical management system 102 may use the medical code data 104 to generate and store patient billing data 106 that associates the particular patient with one or more medical codes corresponding to the treatment and / or diagnosis. For example, the medical management system 102 may create patient billing data 106 that directly includes one or more corresponding medical codes from the medical code data 104. Alternatively or in addition, the patient billing data 106 may include one or more custom billing codes that are linked to one or more corresponding medical codes from the medical code data 104.

[0045] In some examples, the medical institution computer system 110 stores patient medical data 108. In the above example, the patient medical data 108 for a particular patient may include reports submitted by a care provider, notes from a care provider regarding the diagnosis or treatment of the particular patient, health data collected for the particular patient during one or more visits, the medical history of the particular patient, personal information for the particular patient, and / or other medical information corresponding to the particular patient.

[0046] The targeted therapy analysis system 120 performs candidate screening for targeted therapy. For example, the targeted therapy may include a therapeutic procedure, a surgical procedure, an inspection, a clinical trial, a patient registry, a diagnostic procedure, or any other procedure that may be performed during a medical treatment. In some examples, the targeted therapy includes procedures and / or medical devices related to repairing heart defects, heart valve replacement and repair, such as cardiac rhythm management, electrophysiology, valve repair, valve replacement, and other procedures and / or medical devices. An exemplary targeted therapy analysis system that performs candidate screening for left atrial appendage (LAA) closure will be described in more detail below.

[0047] The targeted therapy analysis system 120 includes a modeling component 122. The modeling component 122 of the targeted therapy analysis system 120 generates a candidate classification model 140. The candidate classification model 140 is used to perform candidate screening for a targeted therapy, such as by applying the candidate classification model 140 to patient billing data 150 for a plurality of patients to generate a targeted therapy score 152 for one or more patients. The targeted therapy score 152 quantifies the suitability of a particular patient as a candidate for a targeted therapy.

[0048] The modeling component 122 generates a candidate classification model 140 for targeted therapy based on a set of medical codes. The set of medical codes is selected from a plurality of medical codes corresponding to the patient billing data. For example, the medical codes may include medical codes organized in a standardized code system. The plurality of medical codes may include medical codes stored in the medical code data 104 of the medical institution computer system 110. In some examples, the plurality of medical codes used by the modeling component 122 include medical codes from a plurality of standardized code systems to ensure compatibility with a plurality of medical institution computer systems 110.

[0049] The modeling component 122 selects a set of medical codes to be used in the candidate classification model 140 and assigns a weight to each selected medical code. When the patient billing data 150 associates a patient with a particular medical code, the weight assigned to the particular medical code indicates the contribution of the particular medical code in determining the patient's targeted treatment score 152. The medical codes associated with the patient in the patient billing data 150 may correspond to states and diseases, related health problems, abnormal findings, signs and symptoms, injuries, external causes of injuries and diseases, and / or social conditions.

[0050] The targeted treatment analysis system 120 includes a data import component 124. The data import component 124 of the targeted treatment analysis system 120 imports patient billing data 150 for a plurality of patients. For example, the data import component 124 can receive the patient billing data 150 from the medical institution computer system 110. The patient billing data includes an association between one or more patients in the plurality of patients and one or more medical codes. In some examples, the data import component 124 performs a consistency check on the received patient billing data 150 to ensure that the patient billing data 150 is in the correct form to apply the candidate classification model 140.

[0051] In some examples, the medical institution computer system 110 generates the patient billing data 150 received by the targeted therapy analysis system 120 based on the patient billing data 106 stored by the medical institution computer system 110. For example, the medical institution computer system 110 can select a subset of the patient billing data 106 corresponding to specific patients (e.g., patients in a specific department, patients in a specific demographic grouping, or other subset of patients). Alternatively or additionally, the medical institution computer system 110 can use the candidate classification model 140 to select a subset of the patient billing data corresponding to a relevant time period for analysis.

[0052] In some examples, the medical institution computer system 110 can generate the patient billing data 150 received by the targeted therapy analysis system 120 by processing the patient billing data 106 to ensure compliance with one or more laws or regulations regarding data protection. In some examples, the data import component 124 receives the patient billing data 106 in a de-identified form. For example, the medical institution computer system 110 can ensure that the patient billing data 150 transmitted to the targeted therapy analysis system 120 includes only pseudonymized routine administrative data. The medical institution computer system 110 can create the pseudonymized patient billing data 150 based on its patient billing data 106 and send the pseudonymized patient billing data 150 to the targeted therapy analysis system 120.

[0053] In some examples, the medical institution computer system 110 may maintain information that allows the medical institution computer system 110 to re-identify the targeted therapy candidate data 154 received from the targeted therapy analysis system 120. For example, the reporting component 128 of the targeted therapy analysis system 120 may provide targeted therapy candidate data 154 that references an individual patient through pseudonymization, thereby allowing the medical institution computer system 110 to re-identify the individual patient, enabling care providers at the medical institution to provide medical care to the individual patient based on the targeted therapy candidate data 154. The patient's identity and personal data are protected from disclosure outside the medical institution, and the targeted therapy analysis system 120 only possesses and processes pseudonymized routine administrative data.

[0054] In some examples, the data import component 124 identifies one or more unprocessed medical codes that appear in the patient billing data 150 but were not considered by the modeling component 122 to generate the candidate classification model 140. The unprocessed medical codes can be used to further improve the candidate classification model 140, as described in more detail in one or more examples below.

[0055] The targeted therapy analysis system 120 includes an analysis component 126. The analysis component 126 of the targeted therapy analysis system 120 applies the candidate classification model 140 to the patient billing data 150 to generate a set of targeted therapy scores 152. Each targeted therapy score 152 quantifies the suitability of the patient as a candidate for a targeted therapy. When the candidate classification model 140 is applied to the patient billing data 150, a targeted therapy score 152 is generated for one or more patients based on the weights of the candidate classification model 140 on one or more medical codes associated with each patient in the patient billing data 150. The targeted therapy score 152 may include one or more components. For example, applying the candidate classification model 140 to the patient billing data 150 may cause the candidate classification model 140 to generate a single output score for each patient in the patient billing data 150. Alternatively or additionally, the candidate classification model 140 may generate a single output score for a subset of the patients represented in the patient billing data 150. Alternatively or additionally, the candidate classification model 140 may generate a plurality of output scores for one or more patients represented in the patient billing data 150, each output score representing an aspect of the suitability of the patient as a candidate for a targeted therapy. The targeted therapy score may be a probability value, a Boolean value, an integer, a decimal number, a categorical number, or any other value that may describe suitability for a targeted therapy.

[0056] The targeted therapy analysis system 120 includes a reporting component 128. The reporting component 128 of the targeted therapy analysis system 120 provides targeted therapy candidate data 154 in the form of a report. In some examples, the targeted therapy candidate data 154 includes one or more targeted therapy scores 152 output by the candidate classification model 140. Alternatively or additionally, the targeted therapy candidate data 154 can be generated by further processing the targeted therapy scores 152. For example, the targeted therapy candidate data 154 may include data that classifies one or more patients based on the targeted therapy scores, such as, but not limited to, the following example set of targeted therapy classifications: {"high risk", "medium risk", "low risk"}, and / or {"candidate", "no indication"}. In some examples, the medical institution may specify parameters for one or more targeted therapy classifications used in the targeted therapy candidate data 154.

[0057] In some examples, the targeted therapy candidate data 154 is provided to an operator of the targeted therapy analysis system 120. Alternatively or additionally, the targeted therapy candidate data 154 may be provided to a healthcare facility computer system 110. Alternatively or additionally, the targeted therapy candidate data 154 may be provided to a care provider device 112 of a care provider belonging to a healthcare facility. In some examples, the targeted therapy candidate data 154 is protected, for example, by encryption and / or password protection. Alternatively or additionally, the targeted therapy candidate data 154 may reference individual patients by pseudonyms. The healthcare facility computer system 110 may re-identify individual patients referenced by pseudonyms in the targeted therapy candidate data 154 using re-identification data generated when preparing the patient billing data 150.

[0058] The reporting component 128 can generate the targeted therapy candidate data 154 in the form of one or more files. For example, the files can include one or more of a spreadsheet, a text document, a portable document format (PDF) document, an image, a hypertext markup language (HTML) document, a document in other web page formats, a file with a proprietary format, or any other file. In some examples, the reporting component 128 provides a graphical user interface for display on a computing device operated by a user for the user to access the targeted therapy candidate data 154. The graphical user interface can allow the user to manipulate the targeted therapy candidate data 154, for example, by selecting report parameters for presenting the targeted therapy candidate data 154.

[0059] In some examples, the targeted therapy candidate data 154 includes data for all patients represented in the patient billing data 150. Alternatively or in addition, the targeted therapy candidate data 154 may include data for a subset of patients represented in the patient billing data 150. For example, the targeted therapy candidate data 154 may identify one or more candidate patients having a targeted therapy score that satisfies a targeted therapy score threshold. When the patient's targeted therapy score is above the targeted therapy score threshold, the care provider may further evaluate the patient as a potential candidate for targeted therapy. Alternatively or in addition, the targeted therapy candidate data 154 may include an individual report for at least one patient. The individual report may include additional information about the corresponding patient (e.g., one or more additional medical scores generated based on the patient billing data 150, one or more medical codes associated with the patient in the patient billing data 150, or other information). In some examples, the individual report is provided to a care provider who has provided medical care to the patient.

[0060] In some examples, the targeted therapy candidate data 154 includes summary data for one or more targeted therapy classifications in one or more groupings of a plurality of patients. The summary data may include one or more of counts, percentages, averages, minimums, maximums, means, or other summary data. In some examples, the reporting component 128: (1) groups patients represented in the patient billing data based on a first factor (e.g., department, demographic grouping, and / or association of a particular medical code with the grouping of patients); (2) classifies the patients by targeted therapy classification based on a targeted therapy score associated with each classified patient; and (3) generates summary data for each grouping and targeted therapy classification. Tables 1-3 are examples of summary data generated by the reporting component 128.

[0061] Table 1: Summary data: counts of patients by age group and targeted therapy

[0062]

[0063]

[0064] Table 2: Summary data: Percentage of patients classified as candidates by department

[0065] Department % Candidates Cardiology 5.2％ Geriatrics 3.6％ Endocrinology 1.2％ Internal Medicine 0.3％ Nephrology 0.8％ Pediatrics 0.1％

[0066] Table 3: Summary data: Average age of patients classified as candidates by department

[0067] Department age Cardiology 42.3 years old Geriatrics 67.8 years old Endocrinology 55.4 years old Internal Medicine 54.2 years old Nephrology 49.5 years old Pediatrics 11.0 years old

[0068] As described herein, the targeted therapy analysis system 120 and / or its components (e.g., modeling component 122, data import component 124, analysis component 126, and / or reporting component 128) are presented as separate components for ease of explanation. Any action performed by one or more components of the targeted therapy analysis system 120 or any action performed on one or more components of the targeted therapy analysis system 120 can be considered to be performed by or on the targeted therapy analysis system 120. The targeted therapy analysis system 120 and / or its components can be implemented as one or more interdependent processes or independent processes and can be implemented on one or more computers. For example, the components can be implemented as a distributed system. As an alternative or in addition, multiple instances of one or more components can be implemented. In some examples, one or more components of the targeted therapy analysis system 120 can be implemented as a cloud service and / or using one or more cloud service providers. Further, the components can be fully and / or partially implemented in one or more programs and / or methods, and the two or more components shown can be fully and / or partially implemented in the same program and / or method.

[0069] Candidate Classification Model Configuration

[0070] In some examples, the modeling component 122 assigns positive weights to medical codes corresponding to one or more inclusion criteria for targeted therapy. When a patient is associated with a medical code corresponding to an inclusion criterion, a positive weight results in a higher targeted therapy score. Alternatively or in addition, the modeling component 122 can assign one or more negative weights to medical codes corresponding to one or more exclusion criteria for targeted therapy. When a patient is associated with a medical code corresponding to an exclusion criterion, a negative weight results in a lower targeted therapy score.

[0071] In some examples, the modeling component 122 generates the candidate classification model 140 based on time information associated with the patient billing data 106. For example, the modeling component 122 can generate the candidate classification model 140 based on the following assumption: the patient billing data 106 includes billing information for medical services provided during a specific time period (e.g., medical services provided during the previous year). Alternatively or in addition, the modeling component 122 can take into account the date or elapsed time for one or more medical codes. For example, if major surgery within the past three months is an exclusion factor for targeted therapy, the modeling component 122 can assign conditional weights to one or more medical codes corresponding to major surgery, and when the date associated with the medical code is less than three months ago, the conditional weight is calculated as a negative value. The conditional weight can be implemented as a formula applied to the time information associated with the medical code in the patient billing data 106.

[0072] The medical code data 104 may include a plurality of medical codes arranged in a hierarchy. For example, the ICD-10-CM code scheme includes codes between three and seven characters that specify a hierarchy. The first three characters indicate the category, the next three characters indicate the cause, anatomical site, severity, or other important details, and the last character is primarily used to record a nursing event or other situation. For example, Table 4 includes a set of ICD-10-CM codes arranged in a hierarchy:

[0073] Table 4: Example ICD-10-CM Codes

[0074] ICD-10-CM Codes describe D30 Benign tumors of urinary organs D30.0 Benign kidney tumor D30.00 Unknown benign renal tumor D30.01 Benign tumor of right kidney D30.02 Benign tumor of left kidney

[0075] In some examples, the modeling component 122 generates a candidate classification model 140 by grouping two or more medical codes into a code group, and assigning a single weight to the code group to indirectly assign weights to two or more medical codes in the code group. For example, the modeling component 122 may create a code group corresponding to a medical code beginning with "D30". The code group includes the ICD-10-CM code "D30" and other ICD-10-CM codes below the code "D30" in the hierarchy of the ICD-10-CM code scheme, including "D30.0", "D30.00", "D30.01", and "D30.02". The modeling component 122 may assign a weight w to the code group so that other ICDM-10 codes below the code "D30" are also associated with the weight w. When the patient billing data 106 of a patient is associated with any medical code in the code group, when the candidate classification model 140 is applied, the weight w will be used to calculate the targeted treatment score for the patient.

[0076] In some examples, modeling component 122 allows manual input to select one or more medical codes and / or generate one or more corresponding weights. For example, modeling component 122 may obtain information about one or more medical codes and / or code groups from researchers, clinicians, or other experts. In some examples, modeling component 122 uses manual input that assigns one or more weights to one or more medical codes and / or code groups. Alternatively or in addition, modeling component 122 may use manual input for modifying one or more weights (e.g., by causing one or more weights to increase or decrease). In some examples, modeling component 122 provides a graphical user interface for display on a computing device operated by a user for the user to enter manual input.

[0077] Alternatively or in addition, the modeling component 122 can use machine learning techniques to generate and / or adjust the candidate classification model 140. For example, the modeling component 122 can use machine learning techniques to generate and / or adjust one or more weights of the candidate classification model 140. In some examples, the modeling component 122 implements a machine learning algorithm to generate one or more weights of the candidate classification model 140 based on a training data set. The training data set includes input data, which includes the association between a medical code and a plurality of sample patients. The training data set includes output data, which includes a sample targeted therapy score for each sample patient. The sample targeted therapy score quantifies the suitability of the corresponding sample patient for targeted therapy. The sample targeted therapy score can be a probability value, a Boolean value, an integer, a decimal number, a classification, or any other value that can describe the suitability of the sample patient for targeted therapy.

[0078] In some examples, the training data set is generated based on clinical data involving actual patients (e.g., a care provider's assessment of the suitability of an actual patient for a targeted therapy). Alternatively or in addition, a training data set may be generated based on retrospective results and / or other retrospective data corresponding to one or more actual patients who received targeted therapy or were determined not to be candidates for targeted therapy. Alternatively or in addition, a training data set may be generated based on clinical trial data. Alternatively or in addition, the training data set may include exemplary patient data designed by researchers, care providers, or other experts. The modeling component 122 may configure the candidate classification model 140 in an automated manner. Alternatively or in addition, the modeling component 122 may configure the candidate classification model 140 by combining automated input and manual input.

[0079] In some examples, the modeling component 122 uses a validation dataset in an improved method to evaluate the candidate classification model 140. Alternatively or additionally, the modeling component 122 can use a test dataset to perform a final evaluation of the candidate classification model 140. The validation dataset and / or the test dataset have the same data type and data structure as the training dataset.

[0080] In some examples, the modeling component 122 updates the candidate classification model 140 based on care provider review data regarding one or more previously generated targeted therapy scores 152 (e.g., by adjusting one or more weights of the candidate classification model 140). The care provider review data may include feedback provided by one or more care providers regarding one or more targeted therapy scores 152 and / or targeted therapy candidate data 154 for one or more patients. The feedback may include different decisions made by the care provider regarding the suitability of the patient for the targeted therapy (e.g., revisions to the targeted therapy score or reclassification of the targeted therapy classification assigned to the patient). Alternatively or in addition, the care provider review data may be based on retrospective results and / or other retrospective data corresponding to the patient receiving the targeted therapy or being determined not to be a candidate for the targeted therapy. The modeling component 122 may adjust one or more weights of the candidate classification model 140 based on machine learning techniques, manual input, or a combination of both.

[0081] In some examples, the modeling component 122 updates the candidate classification model 140 based on unprocessed medical codes that appear in the patient billing data 150 but were not considered by the modeling component 122 when generating the candidate classification model 140. The modeling component 122 can adjust one or more weights of the candidate classification model 140 based on machine learning techniques, human input, or a combination of both.

[0082] Exemplary targeted therapy – LAA occlusion

[0083] An exemplary targeted therapy analysis system (e.g., targeted therapy analysis system 120) performs candidate screening for left atrial appendage (LAA) closure, including LAA occluder implantation. The LAA is a sac in the muscular wall of the left atrium. In patients with atrial fibrillation, blood can accumulate in the LAA and form clots, causing these clots to embolize and lead to the risk of stroke. The LAA occluder is implanted in the LAA to prevent fluid communication between the LAA and the left atrium of the heart. This sealing or occlusion of the LAA can effectively prevent LAA clots from entering the systemic circulation, reducing the risk of stroke. Another therapy for reducing the risk of stroke is oral anticoagulants. However, some patients have contraindications to oral anticoagulants. LAA closure has become a treatment option for patients with atrial fibrillation who are at moderate to high risk of stroke (especially patients with contraindications to oral anticoagulants and / or patients at high risk of bleeding).

[0084] The targeted therapy analysis system generates a candidate classification model (e.g., candidate classification model 140) for LAA occluder implantation surgery. The candidate classification model may include assigning negative weights to medical codes corresponding to contraindications to LAA closure. Negative weights reduce the targeted therapy scores of patients associated with the corresponding medical codes. The magnitude of the negative weight may depend on the degree of the contraindications. When the presence of exclusion criteria excludes patients from being candidates for LAA closure, the medical codes corresponding to the exclusion criteria may be assigned higher magnitude negative weights. For example, such exclusion criteria may include death, palliative care, limited motor function, cognitive impairment, mechanical heart valve implantation, and dementia. Contraindications assigned lower magnitude negative weights include alcoholism and cirrhosis.

[0085] The candidate classification model includes assigning positive weights to medical codes corresponding to the inclusion criteria for LAA closure. When patient billing data associates a patient with a corresponding medical code, the positive weight increases the targeted treatment score for the patient. The magnitude of the positive weight reflects the potential effectiveness of LAA closure for candidate patients associated with the corresponding medical code. For example, a medical code corresponding to atrial fibrillation has a higher magnitude of positive weight. Medical codes corresponding to the risks of bleeding, dialysis, renal failure, and anemia may also be assigned positive weights.

[0086] The candidate classification model includes code groups for multiple groups of medical codes. For example, the candidate classification model may include a code group for a group of ICD codes beginning with "I61" that correspond to non-traumatic intracerebral hemorrhage listed in Table 5. When the patient billing data associates a patient with a medical code in the code group corresponding to non-traumatic intracerebral hemorrhage, the positive weight of the code group causes the patient's targeted therapy score for LAA occlusion to increase.

[0087] Table 5: Non-traumatic intracerebral hemorrhage code groups

[0088] ICD Code describe I61 Intrahemispheric, subcortical, non-traumatic intracerebral hemorrhage I61.1 Non-traumatic intracerebral hemorrhage in the hemisphere or cortex I61.2 Unidentified non-traumatic intracerebral hemorrhage in the hemisphere I61.3 Non-traumatic intracerebral hemorrhage in the brainstem I61.4 Non-traumatic intracerebral hemorrhage in the cerebellum I61.5 Non-traumatic intraventricular hemorrhage I61.6 Multiple localized non-traumatic intracerebral hemorrhage I61.8 Other non-traumatic intracerebral hemorrhage I61.9 Unknown non-traumatic intracerebral hemorrhage

[0089] A healthcare institution computer system (e.g., healthcare institution computer system 110) submits pseudonymized patient billing data (e.g., patient billing data 150) to a targeted therapy analysis system. A candidate classification model is applied to a sample patient billing data set to generate targeted therapy scores (e.g., targeted therapy scores 152) for 11,526 patients. The targeted therapy analysis system generates targeted therapy candidate data (e.g., targeted therapy candidate data 154) based on the targeted therapy scores. The targeted therapy analysis system provides the targeted therapy candidate data to the healthcare institution computer system. The targeted therapy candidate data references individual patients by pseudonym, allowing the healthcare institution computer system to re-identify individual patients. The targeted therapy candidate data constitutes an initial screen that may be further reviewed by a care provider during the process of providing treatment to a patient.

[0090] In one example, the targeted therapy analysis system generates summary data for the following targeted therapy classifications: LAA closure (previously received targeted therapy), high risk (candidates with a high targeted therapy score above a first threshold), medium risk (candidates with a targeted therapy score between a first threshold and a second threshold), low risk (candidates with a targeted therapy score between a second threshold and a third threshold), and no indication (non-candidates with a targeted therapy score below a third threshold). Table 6 presents summary data for a sample patient billing data set. The summary data shows that: 10 patients were classified as LAA closure; 18 patients were classified as high risk (e.g., stroke or other factors that may benefit from LAA closure); 28 patients were classified as medium risk; 130 patients were classified as low risk; and 11,340 patients were classified as no indication. Care providers can use the screening results to identify patients for further evaluation of targeted therapy.

[0091] Table 6: Summary data for the sample patient billing dataset

[0092] Classification of targeted therapies count LAA occlusion 10 High risk 18 Medium risk 28 Low risk 130 No instructions 11340

[0093] Example Method

[0094] Figure 2 A flow chart of an example method for candidate screening for targeted therapy is illustrated. Method 200 can be performed by one or more computing devices and / or methods thereof. For example, one or more blocks of method 200 can be performed by a computer system (such as, but not limited to, computer system 400). Figure 1 Method 200 is described with reference to targeted therapy analysis system 120 , but is not limited to performance by such a system.

[0095] In block 202 , the targeted therapy analysis system 120 identifies a plurality of medical codes corresponding to patient billing data.

[0096] In box 204, the targeted therapy analysis system 120 generates a candidate classification model for targeted therapy by selecting a set of medical codes from a plurality of medical codes and assigning a weight to each selected medical code. In some examples, the targeted therapy is a left atrial appendage (LAA) occlusion including LAA occluder implantation. In some examples, the targeted therapy analysis system 120 provides a user interface for adjusting one or more weights assigned to one or more medical codes in a set of medical codes. In some examples, generating a candidate classification model for targeted therapy includes assigning one or more negative weights to one or more medical codes corresponding to one or more exclusion criteria for targeted therapy, and assigning one or more positive weights to one or more medical codes corresponding to one or more inclusion criteria for targeted therapy. Alternatively or in addition, generating a candidate classification model includes grouping two or more medical codes in a set of medical codes into a code group and assigning a single weight to the code group to indirectly assign weights to two or more medical codes in the code group. Alternatively or additionally, generating the candidate classification model includes assigning one or more weights of the candidate classification model based on a training data set using machine learning techniques, the training data set including input data and output data, wherein the input data includes associations between medical codes and a plurality of sample patients, and the output data includes a sample targeted therapy score for each sample patient.

[0097] In block 206, the targeted therapy analysis system 120 receives patient billing data for a plurality of patients. The patient billing data includes an association between one or more of the plurality of patients and one or more of the plurality of medical codes. In some examples, the patient billing data is received from a medical institution. The patient billing data may be received in a de-identified form.

[0098] In block 208, the targeted therapy analysis system 120 applies the candidate classification model to the patient billing data to generate a set of targeted therapy scores. Each targeted therapy score in the set of targeted therapy scores corresponds to one of the plurality of patients. Each targeted therapy score is calculated based on a weight for one or more medical codes associated with the corresponding patient.

[0099] In block 210, the targeted therapy analysis system 120 provides targeted therapy candidate data generated based on the set of targeted therapy scores. For example, the targeted therapy candidate data can be provided to a medical institution (e.g., a medical institution from which patient billing data is received). The targeted therapy candidate data can reference an individual patient from a plurality of patients by a pseudonym.

[0100] In some examples, providing the targeted therapy candidate data includes providing at least a subset of the set of targeted therapy scores. Alternatively or additionally, the targeted therapy candidate data may identify one or more candidate patients having a targeted therapy score that satisfies a targeted therapy score threshold. Alternatively or additionally, the targeted therapy candidate data may include an individual report of at least one patient in the plurality of patients. Alternatively or additionally, the targeted therapy candidate data may include aggregated data of one or more targeted therapy classifications for one or more groupings in the plurality of patients.

[0101] Figure 3 A flowchart of an example method for updating a candidate classification model for targeted therapy is illustrated. Method 300 may be performed by one or more computing devices and / or methods thereof. For example, one or more blocks of method 300 may be performed by a computer system (such as, but not limited to, computer system 400). Figure 1 Method 300 is described with reference to targeted therapy analysis system 120 , but is not limited to performance by such a system.

[0102] In block 302 , the targeted therapy analysis system 120 applies a candidate classification model to patient billing data to generate a set of targeted therapy scores.

[0103] In block 304 , the targeted therapy analysis system 120 identifies one or more unprocessed medical codes in the patient billing data that are not in the plurality of medical codes.

[0104] In block 306 , the targeted therapy analysis system 120 updates the candidate classification model by selecting one or more untreated medical codes from the one or more untreated medical codes and assigning a weight to each of the selected untreated medical codes.

[0105] In block 308 , the targeted therapy analysis system 120 receives care provider review data corresponding to one or more targeted therapy scores in the set of targeted therapy scores.

[0106] In block 310 , the targeted therapy analysis system 120 updates the candidate classification model by adjusting one or more weights assigned to one or more medical codes in the set of medical codes based on the care provider review data.

[0107] Implementation mechanism - Hardware overview

[0108] The technology described herein can be implemented by one or more special-purpose computing devices. Special-purpose computing devices can be hardwired to perform one or more technologies described herein, including combinations of these technologies. Alternatively and / or additionally, the one or more special-purpose computing devices may include digital electronic devices, such as one or more application-specific integrated circuits (ASICs) or field programmable gate arrays (FPGAs) that are continuously programmed to perform technology. Alternatively and / or additionally, the one or more special-purpose computing devices may include one or more general-purpose hardware processors that are programmed to perform technology according to program instructions in firmware, memory, other storage or combinations. Such special-purpose computing devices can also combine custom programs with custom hard-wired logic, ASICs or FPGAs to implement technology. Special-purpose computing devices can be desktop computer systems, portable computer systems, handheld devices, networking devices, and / or any other devices that incorporate hard-wired and / or program logic to implement technology.

[0109] Figure 4 4 is a block diagram illustrating a computer system on which examples may be implemented. Computer system 400 includes a bus 402 or other communication mechanism for communicating information, and one or more hardware processors 404 coupled to bus 402 for processing information (e.g., computing instructions and data). Processor 404 may include one or more general-purpose microprocessors, graphics processing units (GPUs), coprocessors, central processing units (CPUs), and / or other hardware processing units.

[0110] The computer system 400 also includes one or more units of a main memory 406 (e.g., random access memory (RAM) or other dynamic storage) coupled to the bus 402 for storing information and instructions to be executed by the processor 404. The main memory 406 may also be used to store temporary variables or other intermediate information during the execution of instructions to be executed by the processor 404. When stored in a non-transitory storage medium accessible by the processor 404, such instructions make the computer system 400 a special-purpose machine customized to perform the operations specified in the instructions. In some embodiments, the main memory 406 may include dynamic random access memory (DRAM) (including but not limited to double data rate synchronous dynamic random access memory (DDRSDRAM), thyristor random access memory (T-RAM), zero capacitor (Z-RAM) TM ) and / or non-volatile random access memory (NVRAM).

[0111] Computer system 400 may further include one or more units of read-only memory (ROM) 408 or other static storage coupled to bus 402 for storing information and instructions for processor 404 that are either always static or static during normal operation but reprogrammable. For example, ROM 408 may store firmware for computer system 400. ROM 408 may include a mask ROM (MROM) or other hardwired ROM that stores purely static information, a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electronically erasable programmable read-only memory (EEPROM), another hardware memory chip or cartridge, or any other read-only memory unit.

[0112] One or more storage devices 410 (e.g., magnetic disks or optical disks) are provided and coupled to bus 402 for storing information and / or instructions. Storage device 410 may include non-volatile storage media (e.g., read-only memory, optical disks (e.g., but not limited to compact disks (CDs), digital video disks (DVDs), Blu-ray disks (BDs)), magnetic disks, other magnetic media (e.g., floppy disks and tapes), solid-state drives, flash memory, optical disks, one or more forms of non-volatile random access memory (NVRAM), and / or other non-volatile storage media).

[0113] Computer system 400 may be coupled to one or more input / output (I / O) devices 412 via bus 402. For example, I / O devices 412 may include one or more displays (e.g., cathode ray tube (CRT) displays, liquid crystal display (LCD) displays, light emitting diode (LED) displays, projectors, and / or any other type of display) for displaying information to a computer user.

[0114] The I / O device 412 may also include one or more input devices (e.g., an alphanumeric keyboard and / or any other keyboard device). The one or more input devices may also include one or more cursor control devices, such as a mouse, a trackball, a touch input device, or a cursor direction key, for passing direction information and command selections to the processor 404, and for controlling the movement of a cursor on another I / O device (e.g., a display). The cursor control device typically has degrees of freedom on two or more axes (e.g., a first axis x, a second axis y, and optionally one or more additional axes z), which allow the device to specify a position in a plane. In some embodiments, the one or more I / O devices 412 may include a device with a combined I / O function, such as a display that supports touch functionality.

[0115] Other I / O devices 412 may include a fingerprint reader, a scanner, an infrared (IR) device, an imaging device (e.g., a camera or video recording device), a microphone, a speaker, an ambient light sensor, a pressure sensor, an accelerometer, a gyroscope, a magnetometer, another motion sensor, or any other device that can communicate signals, commands and / or other information with the processor 404 via the bus 402.

[0116] Computer system 400 can implement the techniques described herein using custom hardwired logic, one or more ASICs or FPGAs, firmware, and / or program logic, which in combination with the computer system results in or is programmed to make computer system 400 a special purpose machine. In some examples, computer system 400 performs the techniques herein in response to processor 404 executing one or more sequences of one or more instructions contained in main memory 406. Such instructions may be read into main memory 406 from another storage medium (e.g., one or more storage devices 410). Executing the sequences of instructions contained in main memory 406 causes processor 404 to perform the method steps described herein. In alternative embodiments, hardwired circuits may be used in place of or in combination with software instructions.

[0117] The computer system 400 also includes one or more communication interfaces 418 coupled to the bus 402. The communication interface 418 provides two-way data communication over one or more physical or wireless network links 420 that are connected to a local network 422 and / or a wide area network (WAN), such as the Internet. For example, the communication interface 418 may include an integrated services digital network (ISDN) card, a cable modem, a satellite modem, or a modem for providing a data communication connection to a corresponding type of telephone line. Alternatively and / or in addition, the communication interface 418 may include one or more of the following: a local area network (LAN) device that provides a data communication connection to a compatible local network 422; a wireless local area network (WLAN) device that sends and receives wireless signals (e.g., electrical signals, electromagnetic signals, optical signals, or other wireless signals representing various types of information) to a compatible LAN; a wireless wide area network (WWAN) device that sends and receives such signals over a cellular network; and other network devices that establish a communication channel between the computer system 400 and one or more LANs 422 and / or WANs.

[0118] The network link 420 typically provides data communication across one or more networks to other data devices. For example, the network link 420 can provide a connection to one or more host computers 424 or to data devices operated by an Internet service provider (ISP) 426 through one or more local area networks 422 (LANs). The ISP 426 provides a connection to one or more wide area networks 428 (e.g., the Internet). The LAN 422 and WAN 428 use electronic signals, electromagnetic signals, or optical signals that carry digital data streams. The signals through the various networks and the signals on the network link 420 and through the communication interface 418 are exemplary forms of transmission media or transient media.

[0119] The term "storage medium" as used herein refers to any non-transitory medium that stores data and / or instructions that cause a machine to operate in a specific manner. Such storage media may include volatile and / or non-volatile media. Storage media are distinct from, but may be used with, transmission media. Transmission media participate in the transmission of information between storage media. For example, transmission media include coaxial cables, copper wires, and optical fibers, including traces and / or other physical conductive components that make up bus 402. Transmission media may also take the form of sound waves or light waves, such as those generated in radio wave and infrared data communications.

[0120] Various forms of media may be utilized to carry one or more sequences of one or more instructions to the processor 404 for execution. For example, these instructions may initially be carried on a disk or solid state drive of a remote computer. The remote computer may load the instructions into its main memory 406 and send the instructions via a telephone line using a modem. The local modem of the computer system 400 may receive data on the telephone line and convert the data into an infrared signal using an infrared transmitter. An infrared detector may receive the data carried in the infrared signal and appropriate circuitry may place the data on the bus 402. The bus 402 carries the data to the main memory 406, from which the processor 404 retrieves and executes the instructions. Optionally, the instructions received by the main memory 406 may be stored on the storage device 410 before or after the processor 404 executes.

[0121] Computer system 400 may send messages and receive data, including program code, through the network, network link 420, and communication interface 418. In the Internet example, one or more servers 430 may transmit signals corresponding to data or instructions requested by an application program executed by computer system 400 through Internet 428, ISP 426, local network 422, and communication interface 418. Received signals may include instructions and / or information that are executed and / or processed by processor 404. Processor 404 may execute and / or process the instructions and / or information when the signals are received by accessing main memory 406, or execute and / or process the instructions and / or information later by storing them and then accessing them from storage device 410.

[0122] Other aspects of the present disclosure

[0123] Although the concepts herein have been described with respect to specific embodiments, it should be understood that these embodiments are merely exemplary of the principles and applications of the present disclosure. Therefore, it should be understood that various modifications may be made to the illustrative embodiments and other arrangements may be designed without departing from the spirit and scope of the present disclosure as defined in the appended claims.

Claims

1. A computer system comprising: one or more hardware processors; at least one memory storing one or more instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to: identifying a plurality of medical codes corresponding to patient billing data; generating a candidate classification model for targeted treatment by selecting a set of medical codes from the plurality of medical codes and assigning a weight to each of the selected medical codes; receiving patient billing data for a plurality of patients, the patient billing data comprising an association between one or more of the plurality of patients and one or more of the plurality of medical codes; applying the candidate classification model to the patient billing data to generate a set of targeted treatment scores, each targeted treatment score in the set of targeted treatment scores corresponding to one of the plurality of patients and calculated based on a weight for one or more medical codes associated with the one patient; and Targeted therapy candidate data generated based on the set of targeted therapy scores is provided.

2. The computer system of claim 1, wherein the targeted therapy candidate data identifies one or more candidate patients whose candidates have a targeted therapy score that satisfies a targeted therapy score threshold.

3. The computer system of claim 1, wherein the targeted therapy candidate data comprises an individual report for at least one patient among the plurality of patients.

4. The computer system of claim 1, wherein the targeted therapy candidate data comprises summary data for one or more targeted therapy classifications across one or more groupings of the plurality of patients.

5. The computer system of claim 1, wherein receiving the patient billing data comprises receiving the patient billing data from a medical institution. 6 . The computer system of claim 5 , wherein providing the targeted therapy candidate data comprises providing the set of targeted therapy scores to the medical institution.

7. The computer system of claim 5, wherein the patient billing data is received in a de-identified form.

8. The computer system of claim 7, wherein the targeted therapy candidate data references individual patients from the plurality of patients by pseudonym.

9. The computer system of claim 1, wherein generating the candidate classification model for the targeted therapy comprises: assigning one or more negative weights to one or more medical codes corresponding to one or more exclusion criteria for the targeted therapy; as well as One or more positive weights are assigned to one or more medical codes corresponding to one or more inclusion criteria for the targeted therapy.

10. The computer system of claim 1 , wherein when the instructions are executed by the one or more hardware processors, the instructions cause the one or more hardware processors to: A user interface is provided for adjusting one or more weights assigned to one or more medical codes in the set of medical codes.

11. The computer system of claim 1 , wherein generating the candidate classification model comprises using a machine learning technique to assign one or more weights of the candidate classification model based on a training data set, the training data set comprising: Input data, the input data comprising associations between medical codes and a plurality of sample patients; as well as Output data includes a sample targeted therapy score for each sample patient in the plurality of sample patients.

12. The computer system of claim 1 , wherein when the instructions are executed by the one or more hardware processors, the instructions cause the one or more hardware processors to: receiving care provider review data corresponding to one or more targeted therapy scores in the set of targeted therapy scores; and The candidate classification model is updated by adjusting one or more weights assigned to one or more medical codes in the set of medical codes based on the care provider review data.

13. The computer system of claim 1 , wherein when the instructions are executed by the one or more hardware processors, the instructions cause the one or more hardware processors to: identifying one or more unprocessed medical codes in the patient billing data that are not in the plurality of medical codes; and The candidate classification model is updated by selecting one or more untreated medical codes from the one or more untreated medical codes and assigning a weight to each of the selected untreated medical codes.

14. The computer system of claim 1 , wherein generating the candidate classification model comprises: grouping two or more medical codes from the set of medical codes into a code group; as well as A single weight is assigned to the code group to indirectly assign weights to the two or more medical codes in the code group.

15. The computer system of claim 1, wherein the targeted therapy is left atrial appendage (LAA) occlusion, including LAA occluder implantation.

16. A method comprising: identifying a plurality of medical codes corresponding to patient billing data; generating a candidate classification model for targeted treatment by selecting a set of medical codes from the plurality of medical codes and assigning a weight to each of the selected medical codes; receiving patient billing data for a plurality of patients, the patient billing data comprising an association between one or more of the plurality of patients and one or more of the plurality of medical codes; applying the candidate classification model to the patient billing data to generate a set of targeted treatment scores, each targeted treatment score in the set of targeted treatment scores corresponding to one of the plurality of patients and calculated based on a weight for one or more medical codes associated with the one patient; and Targeted therapy candidate data generated based on the set of targeted therapy scores is provided. The method is implemented by one or more processors.

17. The method according to claim 16, Wherein receiving the patient billing data comprises receiving the patient billing data from a medical institution; wherein providing the targeted therapy candidate dataset comprises providing the set of targeted therapy scores to the medical institution; wherein the patient billing data is received from the medical institution in a de-identified form; and wherein the targeted therapy candidate data refers to individual patients among the plurality of patients by pseudonym.

18. The method of claim 16, wherein generating the candidate classification model comprises using a machine learning technique to assign one or more weights to the candidate classification model based on a training data set, the training data set comprising: Input data, the input data comprising associations between medical codes and a plurality of sample patients; as well as Output data includes a sample targeted therapy score for each sample patient in the plurality of sample patients.

19. The method according to claim 16, further comprising: receiving care provider review data corresponding to one or more targeted therapy scores from the set of targeted therapy scores; as well as The candidate classification model is updated by adjusting one or more weights assigned to one or more medical codes in the set of medical codes based on the care provider review data.

20. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors of a network system, cause the network system to: identifying a plurality of medical codes corresponding to patient billing data; generating a candidate classification model for targeted treatment by selecting a set of medical codes from the plurality of medical codes and assigning a weight to each of the selected medical codes; receiving patient billing data for a plurality of patients, the patient billing data comprising an association between one or more of the plurality of patients and one or more of the plurality of medical codes; applying the candidate classification model to the patient billing data to generate a set of targeted treatment scores, each targeted treatment score in the set of targeted treatment scores corresponding to one of the plurality of patients and calculated based on a weight of one or more medical codes associated with the one patient; and Targeted therapy candidate data generated based on the set of targeted therapy scores is provided.

Images