Diagnosis of DMVD at Stage B2

Abstract

The present invention relates to methods for diagnosing stage B2 degenerative mitral valve disease (DMVD) in dogs, methods for determining the probability that a dog has stage B2 DMVD, methods for training models for predicting stage B2 DMVD in dogs, and related computer programs and systems.

Description

[Technical Field]

[0001] The present invention relates to a method for diagnosing stage B2 degenerative mitral valve disease (DMVD) in dogs, a method for determining the probability that a dog has stage B2 DMVD, a method for training a model for predicting stage B2 DMVD in dogs, and related computer programs and systems. [Background technology]

[0002] Degenerative mitral valve disease (DMVD) is an acquired condition characterized by progressive myxomatous degeneration of the mitral valve. 1 DMVD is the most common heart disease in adult dogs. 2 It is estimated that 3.5% of dogs seen in primary care settings are affected. 3 Dogs with DMVD experience a long, asymptomatic period during which they may develop abnormal hypertrophy of the left ventricle to compensate for chronic dendritic blood volume. These structural changes are used in a staging scheme developed by the American College of Veterinary Internal Medicine (ACVIM) to identify dogs with more advanced pre-symptomatic disease. 4 Dogs are classified as stage B2 if echocardiographic measurements of the left atrium and left ventricle exceed an established threshold indicating the presence of cardiac hypertrophy. Accurately identifying dogs in stage B2 is clinically important because the EPIC trial (the effect of pimobendan in dogs with pre-symptomatic myxomatous mitral valve disease and cardiac hypertrophy) demonstrated a clear benefit for the medical management of these cases. 5 In the EPIC trial, treatment with the drug pimobendan reduced the risk of reaching the trial's primary composite endpoint of congestive heart failure (CHF), cardiac-related death, or euthanasia by approximately one-third. Considering the average life expectancy of dogs... 6 An extended pre-symptomatic stage in DMVDs represents a significant extension of a good quality of life. 7 .

[0003] It is difficult to recognize whether an affected animal is in stage B2 using only the parameters obtained from external tests. Furthermore, factors related to the affected animal, the owner, and primary care veterinary practice can also influence access to echocardiography. To enable widespread implementation of the procedures according to the EPIC trial, it is necessary to identify dogs with DMVD at stage B2 without using echocardiography. SUMMARY OF THE INVENTION

[0004] The present invention relates to the identification of dogs with stage B2 DMVD without using echocardiography. The inventors have surprisingly identified that the probability that a dog has stage B2 DMVD is related to parameters that can be easily and routinely evaluated in primary care veterinary practice.

[0005] Using a model based on those parameters, an output value associated with the probability that a dog has stage B2 DMVD can be generated. In this way, dogs can be screened for stage B2 DMVD to identify individuals who are likely to benefit from echocardiography. This can assist in the allocation of owners and diagnostic resources. Instead of traditional echocardiography measurements, the output value can also be used to diagnose the presence or absence of B2 DMVD. This enables disease staging in dogs without access to echocardiography.

[0006] By reducing reliance on echocardiography for the identification of stage B2 DMVD, the findings of the EPIC trial can be more widely implemented. That is, the present invention facilitates the identification of stage B2 in primary care veterinary practice and allows treatment to be advantageously initiated at this pre-onset stage of the disease.

[0007] Therefore, the present invention provides a method for diagnosing degenerative mitral valve disease (DMVD) at stage B2 in dogs, comprising: (a) receiving characteristic data associated with a dog, wherein the characteristic data includes two or more of appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, gender, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; (b) processing the characteristic data using a model, wherein the output of the model is an output value associated with the probability that the dog has DMVD at stage B2; and (c) diagnosing the presence or absence of DMVD at stage B2 based on a comparison of the output value with a predetermined value.

[0008] The present invention also provides - A method for screening for DMVD at stage B2 in dogs, comprising: (a) receiving characteristic data associated with a dog, wherein the characteristic data includes two or more of appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, gender, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; and (b) processing the characteristic data using a model, wherein the output of the model is an output value associated with the probability that the dog has DMVD at stage B2. - A method for training a model to predict stage B2 DMVD in dogs, comprising: (i) processing trait data associated with dogs using a model to produce output values, wherein the trait data includes two or more of the following: appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; (ii) comparing the output values ​​to a diagnosis of the presence or absence of stage B2 DMVD in dogs; and (iii) adjusting the parameters of the model based on the results of the comparison; - A computer program, when executed by a computer system, including coding means for instructing a computer system to carry out the method described in any one of the claims; and - A system for diagnosing stage B2 DMVD in dogs, comprising: an input device configured to receive characteristic data associated with a dog, wherein the characteristic data includes two or more of the following: appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; a model configured to receive the characteristic data and generate an output value associated with the probability that the dog has stage B2 DMVD; and an output device configured to output the output value. We also offer it. [Brief explanation of the drawing]

[0009] [Figure 1] This flowchart shows how to distribute the data before including it in the analysis. [Figure 2] This is the receiver-operated characteristic curve for explanatory multivariate logistic regression analysis of risk factors associated with having stage B2 disease. The 95% confidence interval is represented by the blue area around the receiver-operated curve. Area under the curve: 0.84 (0.82~0.87). [Figure 3] Figures 3 and 4 provide a comparison of receiver operational characteristic curves. These figures show the discriminative performance of the explanatory multivariate logistic regression model when applied to a clean population, a complete population, and an excluded population. The area under the receiver operational characteristic curve was as follows: 0.84 (0.82~0.87) for the clean population; 0.81 (0.79~0.83) for the complete population; and 0.76 (0.72~0.80) for the excluded population. [Figure 4]This figure shows the discriminative performance of explanatory multivariate logistic regression models compared to using NT-proBNP alone or the vertebral heart score. The area under the receiver operational characteristic curve was as follows: multivariate logistic regression model 0.84 (0.82~0.87); univariate NT-proBNP model 0.77 (0.74~0.80); univariate vertebral heart score model 0.76 (0.69~0.83). NT-proBNP: N-terminal propeptide of type B natriuretic peptide; VHS: Vertebral heart score. [Figure 5] This figure shows variable importance plots indicating the five most important variables for the following classifiers: a) Ridge Regression, b) Support Vector Machines with Linear Kernel, c) Random Forest, and d) GBM. Variable importance is presented for the most important predictors in each model. The scores for Ridge Regression and Support Vector Machines with Linear Kernel are coefficients for each variable. The scores for Random Forest and GBM are the mean Shapley values ​​for each variable. BCS (Body Condition Score); BUN (Blood Urea Nitrogen); GBM (Gradient Boosting Machine); NT-proBNP (N-terminal propeptide of type B natriuretic peptide). [Figure 6] This figure shows the gradient related to the interaction between ALT and NT-proBNP. NT-proBNP is plotted against the log odds of having stage B2 disease at different levels of ALT. [Figure 7] This is a diagram illustrating a system for diagnosing stage B2 DMVD in dogs. [Modes for carrying out the invention]

[0010] Screening method The present invention relates to a method for screening dogs for stage B2 DMVD, comprising: (a) receiving characteristic data associated with a dog, wherein the characteristic data includes two or more of the following: appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; and (b) processing the characteristic data using a model, wherein the output of the model is an output value associated with the probability that the dog has stage B2 DMVD. This provides a method that includes [something].

[0011] Screening methods have the advantage of allowing identification of the probability that a dog has stage B2 DMVD without using echocardiography. Traditionally, a dog is classified as stage B2 if echocardiographic measurements of the left atrium and left ventricle exceed an established threshold indicating the presence of cardiac hypertrophy. However, echocardiography requires specialized equipment. Many primary care veterinary practices do not have the necessary equipment to perform echocardiography. Furthermore, performing echocardiography and analyzing the results requires specialized skills that veterinarians used in primary care practice do not typically possess. Thus, identifying stage B2 DMVD often requires referring the animal to a specialized facility. This can be time-consuming and expensive, and often requires travel for the animal and its owner. In some cases, these factors prevent dogs from accessing echocardiography. If stage B2 DMVD cannot be identified, it is difficult to know when to take action.

[0012] Screening methods can help overcome these problems. The method can identify the probability that a dog has stage B2 DMVD without requiring echocardiography. The method generates an output value associated with the probability that a dog has stage B2 DMVD. This output value can be used to inform clinical decisions. For example, the output value can be used to identify dogs for which echocardiography is particularly indicated. That is, dogs with an output value associated with a high probability of having stage B2 DMVD can be prioritized for echocardiography to confirm the presence (or absence) of B2 DMVD. Treatment can be performed depending on the outcome of the echocardiography. Dogs with a low probability of stage B2 DMVD are less likely to show abnormalities during echocardiography. Therefore, an output value associated with a low probability of stage B2 DMVD may indicate that echocardiography may offer little benefit to certain dogs. Furthermore, it may prevent the use of medications that are not indicated due to potential adverse effects. Prioritizing cases for echocardiography can conserve pet owners and veterinary resources. Individual dogs can be subjected to fewer clinical interventions.

[0013] dog This method is used to screen dogs for stage B2 DMVD. The dogs may be domestic dogs (Canis familiaris) or any other member of the genus Canis.

[0014] A dog may have or be suspected of having DMVD. The dog may have been diagnosed (or suspected) with DMVD prior to step (a) of this method. The diagnosis (or suspected diagnosis) may be based, for example, on the dog's signalment. For example, certain breeds and / or ages of dogs may be suspected of having DMVD. The diagnosis (or suspected diagnosis) may be based, for example, on the presence of a left apical systolic murmur. Preferably, the diagnosis (or suspected diagnosis) is based on the presence of a left apical systolic murmur in dogs of certain ages and / or breeds. Breeds and ages at risk are well known in the art.

[0015] DMVD is preferably preclinical; that is, the dog is preferably not congestive heart failure, which can be indicated by radiographic, histological, and / or physical examination findings. Preferably, the dog has not been treated with loop diuretics prior to the procedure. In asymptomatic or preclinical DMVD, physiological changes compensate for mitral valve dysfunction.

[0016] The dog may be of any age. For example, the dog may be at least 1 year old, at least 2 years old, at least 3 years old, at least 4 years old, at least 5 years old, at least 6 years old, at least 7 years old, at least 8 years old, at least 9 years old, at least 10 years old, at least 11 years old, or at least 12 years old. Preferably, the dog is at least 6 years old. DMVD is an acquired disease, and affected animals typically develop it in adulthood as they get older.

[0017] The dog may be of any breed. For example, the dog may be a Cavalier King Charles Spaniel, Jack Russell Terrier, Chihuahua, Cocker Spaniel, or Shih Tzu.

[0018] Dogs may be of any sex. Dogs may be unneutered males. Dogs may be unneutered females. Dogs may be neutered males. Dogs may be neutered females.

[0019] The dog may weigh any weight. For example, the dog may weigh at least 1 kg, at least 2 kg, at least 5 kg, at least 10 kg, at least 15 kg, at least 20 kg, at least 25 kg, or at least 30 kg. The dog may weigh between 1 kg and 70 kg, such as 2 kg to 65 kg, 5 kg to 60 kg, 10 kg to 55 kg, 15 kg to 50 kg, 20 kg to 45 kg, 30 kg to 40 kg, or 30 kg to 35 kg. Preferably, the dog weighs between 2 kg and 25 kg. Typically, DMVD is a condition that affects small breed dogs.

[0020] Preferably, the dog has not been treated with any cardiac medication selected prior to step (a) of the method. These include pimobendan, loop diuretics (e.g., furosemide), and antiarrhythmic drugs. Any cardiac medication prescribed to the dog will be evident from its medical history.

[0021] Characteristic data The screening method includes the step of receiving characteristic data associated with dogs being screened for stage B2 DMVD. The characteristic data includes any combination of two or more of the following: appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration. For example, characteristic data include appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, and globulin concentration. It may contain any combination of three or more, four or more, five or more, six or more, seven or more, eight or more, nine or more, ten or more, eleven or more, twelve or more, thirteen or more, four or more, six or more, seven or more, eight or more, nine or more, ten or more, eleven or more, twelve or more, thirteen or more, four or more, fifteen or more, fifteen or more, sixteen or more, seventeen or more, eighteen or more, nine or more, twenty or more, twenty-one

[0022] Characteristic data may include, for example, two or more of the following: (i) appetite, (ii) body condition score (BCS), (iii) creatinine concentration, (iv) heart murmur intensity, and (v) NT-ProBNP concentration. For example, characteristic data is (i) and (ii); (i) and (iii); (i) and (iv); (i) and (v); (ii) and (iii); (ii) and (iv); (ii) and (v); (iii) and (iv); (iv) and (v); (i), (ii), and (iii); (i), (ii) and (iv); (i), (ii) and (v); (i), (iii) and (iv); (i), (iii) and (v); (i), (iv) and (v); (ii), (iii) and (v); (ii), (iv) and (v); (iii), (iv) and (v); (i), (ii) and (iii), (iv); (i), (ii), (iii) and (v); (iii), (iv) and (v); (i), (iii), (iv) and (v); (ii), (iii), (iv), and (v); or may include (i), (ii), (iii), (iv), and (v).

[0023] Preferably, the characteristic data includes NT-ProBNP concentration. The characteristic data may also include, for example, NT-proBNP concentration, appetite, creatinine concentration, and heart murmur intensity. For example, the characteristic data may include NT-proBNP concentration, appetite, creatinine concentration, heart murmur intensity, and BCS.

[0024] If the model used to process the characteristic data is derived using a regression process (such as multivariate logistic regression or regularized regression), the characteristic data may preferably include NT-proBNP concentration, appetite, creatinine concentration, and heart murmur intensity; or it may include NT-proBNP concentration, appetite, creatinine concentration, heart murmur intensity, and BCS.

[0025] Characteristic data may include all of the following: appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration.

[0026] If the model is derived using a machine learning process (such as a support vector machine (SVM) process, a random forest process, or a gradient boosting process) or regularized regression, the characteristic data may preferably include all of the following: appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration.

[0027] The characteristic data for each type can be easily evaluated in primary care practice using routine methods in this technology.

[0028] For example, a primary care physician or dog owner can assess appetite by visually monitoring changes in appetite over a period of time. The period may be, for example, one week, two weeks, three weeks, four weeks, one month, two months, three months, four months, five months, or six months. The period may also be from one week to one year, for example, two weeks to eleven months, three weeks to ten months, four weeks to nine months, one month to eight months, two months to seven months, three months to six months, or four months to five months. The period may also be six months. The period may also be from one month to six months, for example, one month to three months. Appetite can be scored, for example, as "decreased" or "normal."

[0029] The Body Condition Score (BCS) can be scored using the American Animal Hospital Association's 9-point scale (American Animal Hospital Association. Canine Body Condition Score for 1-9 and 1-5 Scales. Veterinary Forensics: Animal Cruelty Investigations. 2013). This scale uses visual indicators and palpation results to characterize an individual's body condition (e.g., muscle mass, fat accumulation). Body condition can indicate whether an individual dog is healthy, overweight, or underweight. For example, a BCS of 3 or less may indicate that the dog is underweight. A BCS of 6 or more may indicate that the dog is overweight. A BCS of 4 or 5 may indicate that the dog is healthy.

[0030] DMVD is associated with a characteristic heart murmur. Dogs with DMVD typically have a left apical systolic murmur that is strongest at the mitral valve. Heart murmur intensity can be assessed using the Levine scale (Levine SA. Notes on the Gradation of the Intensity of Cardiac Murmurs. JAMA J Am Med Assoc. 1961 Jul 29;177(4):261), which assigns grades I to VI based on their audibility during auscultation. The grades assigned by the Levine scale can be reclassified, for example, to reduce the complexity of this method. For example, grades I and II murmurs can be classified as "soft". Grade III murmurs can be classified as "moderate". Grade IV murmurs can be classified as "loud". Grades V and VI murmurs can be classified as "thrilling". Heart murmur intensity can be assessed using a simplified scale that classifies murmurs as soft, moderate, loud, or tremor based on their audibility during heart sound auscultation.

[0031] A dog's age is typically measured in years. To reduce the complexity of this method, a dog's age can be classified into specific age ranges (e.g., under 8 years, 8-10 years, 10-12 years, or over 12 years).

[0032] Breeds can be evaluated by visual inspection or by referring to the dog's pedigree. Dog breeds are well known in the art, and those skilled in the art can easily determine the breed. DMVD is most common in small breeds, and some breeds (such as the Cavalier King Charles Spaniel (CKCS)) are highly susceptible.

[0033] Examples of cardiac medications include pimobendan, angiotensin-converting enzyme (ACE) inhibitors, and diuretics. Examples of ACE inhibitors include benazepril and enalapril. Examples of diuretics include loop diuretics (e.g., furosemide) and potassium-sparing diuretics (e.g., spironolactone). The cardiac medication prescribed to the dog should be evident from its medical history.

[0034] A dog's sex will be determined from a physical examination and / or its medical history. Sex can be classified as unneutered male, unneutered female, neutered male, or neutered female.

[0035] For example, a primary care physician or dog owner can assess coughing by visually monitoring for its presence or absence. Coughing can be classified, for example, as "present" or "absent."

[0036] Exercise tolerance is the ability to perform physical activity that is considered to be at a level or duration normally expected for an individual. For example, a primary care physician or dog owner can assess exercise tolerance by visually monitoring changes in exercise tolerance over a period of time. The period may be, for example, one week, two weeks, three weeks, four weeks, one month, two months, three months, four months, five months, or six months. The period may also be from one week to one year, for example, two weeks to eleven months, three weeks to ten months, four weeks to nine months, one month to eight months, two months to seven months, three months to six months, or four months to five months. The period may also be six months. The period may also be from one month to six months, for example, one month to three months. Exercise tolerance can be scored, for example, as "decreased" or "normal".

[0037] Heart rate can be measured as beats per minute. Heart rate is typically measured, for example, by auscultation of heart sounds. It can also be measured by manually taking the dog's pulse, for example, using pulse oximetry or an ECG.

[0038] Heart rhythm is typically assessed by auscultation of heart sounds. ECG can also be used to assess heart rhythm. Heart rhythm can be classified based on the dominant rhythm throughout the observation period. For example, heart rhythm can be classified as "sinus rhythm," "sinus arrhythmia," or "other."

[0039] Respiratory rate can be measured as breaths per minute. Respiratory rate is typically measured, for example, by observation or chest auscultation in dogs.

[0040] Creatinine concentration, alanine aminotransferase (ALT) activity, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, sodium concentration, and cTnI concentration can be measured in dog samples using routine assays in the art. A lower limit can be assigned to results below the detection limit of the assay. An upper limit can be assigned to results above the detection limit of the assay. The sample may be a blood sample, such as a venous blood sample. The sample may be a serum sample. A serum sample can be obtained by processing a blood sample, such as a venous blood sample.

[0041] NT-proBNP concentration can be measured in a sample obtained from a dog using a routine assay in the art. A lower limit can be assigned to results below the detection limit of the assay. An upper limit can be assigned to results above the detection limit of the assay. The sample may be a blood sample, such as a venous blood sample. The sample may also be a plasma sample. A plasma sample can be obtained by processing a blood sample, such as a venous blood sample.

[0042] Before step (a), any of the characteristic data can be transformed. For example, characteristic data associated with continuous variables (such as heart rate, respiratory rate, creatinine concentration, alanine aminotransferase (ALT) activity, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, sodium concentration, cTnI concentration, and NT-proBNP concentration) can be transformed (e.g., logarithmically transformed), scaled, or classified (e.g., to quartiles). Specific data associated with categorical variables (such as appetite, BCS, heart murmur intensity, age, breed, sex, cough, and exercise tolerance) can be reclassified to a broader level.

[0043] Model The screening method includes a step of processing characteristic data using a model. The model may be an algorithm that performs several processing steps on the characteristic data and produces an output. One or more processing steps may produce sub-outputs that are used by the model in subsequent processing steps. One or more processing steps can be performed simultaneously.

[0044] The processing of characteristic data may include a step of performing mathematical calculations on the characteristic data. For example, the model may use one or more characteristic data as inputs to a linear function for calculating the output.

[0045] The processing of characteristic data may include a step of performing a classification step using the characteristic data. The model may use one or more types of characteristic data to perform the classification step. The classification step may be a binary classification step that produces an output associated with one of two categories based on one or more types of characteristic data. The classification step may produce an output associated with one of more than two categories.

[0046] The model can be derived using a regression process. In the regression process, trait data linked to multiple dogs that have previously been diagnosed with stage B2 DMVD or not can be analyzed to derive a relationship between the set of trait data and the presence of stage B2 DMVD.

[0047] The model can be derived using multivariate logistic regression. Logistic regression is a regression process that uses a logistic function to associate one or more input variables with predictions of the likelihood of an output variable. One or more types of characteristic data can be used as input variables, and the output variable may be a value associated with the likelihood that a dog associated with a set of characteristic data has a DMVD of stage B2. The logistic function may be based on the coefficients in Table 6: The following characteristics: decreased appetite, yes; physical condition score, 5; creatinine, C; heart murmur, moderate; log 10 (NT-proBNP), for dogs with NT,

[0048]

number

[0049] The model can be derived using regularized regression. Regularized regression is a regression process that adds a penalty function to a least-squares fitting process used to derive a linear function that associates one or more types of characteristic data with values ​​associated with the likelihood that a dog associated with a set of characteristic data has a DMVD of stage B2. The regularized regression process may also be a ridge regression process. An exemplary linear relationship is described in Example 2.

[0050] The model can be derived using a machine learning process. In this process, trait data associated with dogs having or not having Stage B2 DMVD can be processed by an untrained model using a set of starting states. The starting states can be determined randomly, or one or more starting states may be predetermined. The output of the untrained model can be compared to the state of the dogs associated with the trait data, and the processing by the untrained model can be adjusted based on this comparison. This process can be repeated until the model's output corresponds to an accurate prediction of whether a dog associated with a particular set of trait data has Stage B2 DMVD. This process can be called the training process. Furthermore, the regression process described above can be achieved by using the training process.

[0051] The model can be derived using a Support Vector Machine (SVM) process. In the SVM process, trait data associated with dogs can be represented as vectors defined by each variable that makes up the trait data. The model derived using the SVM process compares the positions of the vectors identified by the trait data to a hyperplane. Depending on the position of the vectors associated with the hyperplane, the model can classify the trait data as being associated with dogs having a Stage B2 DMVD or with dogs not having a Stage B2 DMVD. During the training process of the SVM model, the position of the hyperplane is modified to maximize the distance between the hyperplane and the nearest vector associated with dogs having a Stage B2 DMVD, and the distance between the hyperplane and the nearest vector associated with dogs not having a Stage B2 DMVD.

[0052] The model can be derived using a random forest process. The random forest process generates multiple decision trees that classify the trait data as either associated with dogs having stage B2 DMVD or with dogs not having stage B2 DMVD. Each decision tree may have several steps in which subcategories are created based on variables in the trait data. For example, a subcategory may be created if a variable is greater than or less than a certain value. Based on combinations of subcategories, each decision tree arrives at a classification of the trait data. The classifications reached by the model derived using the random forest process may be based on combinations of classifications created by multiple decision trees. For example, modal classification or mean class probabilities can be used. In the training process of a model derived using a random forest process, the set of training data and the types of trait data used by each decision tree can be randomly selected.

[0053] Gradient boosting processes can be used when deriving models. For example, in a random forest process, the accuracy of each generated decision tree can be used as input when determining the parameters of the next decision tree. The XGBoost algorithm is an example of a machine learning process that can be used to derive models using gradient boosting processes.

[0054] Output value The model's output is an output value associated with the probability that a dog has a Stage B2 DMVD. The output value may be a discrete or continuous variable. For example, the output value may be a classification of whether the characteristic data used as input for the model is associated with dogs having a Stage B2 DMVD or dogs not having a Stage B2 DMVD. The output value may also be a variable associated with the probability that the characteristic data used as input for the model is associated with dogs having a Stage B2 DMVD. The output value may be equal to the probability that a dog has a Stage B2 DMVD. The output value does not have to be equal to the probability that a dog has a Stage B2 DMVD; the output value can be obtained by processing it. For example, the output value may be a function of the probability that a dog has a Stage B2 DMVD, e.g., its inverse.

[0055] Diagnostic methods The present invention relates to a method for diagnosing stage B2 degenerative mitral valve disease (DMVD) in dogs, wherein the characteristic data includes appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase ( The method provides a step of: (a) receiving characteristic data associated with a dog, including two or more of the following: GGT concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; (b) processing the characteristic data using a model, the output of which is an output value associated with the probability that the dog has stage B2 DMVD; and (c) diagnosing the presence or absence of stage B2 DMVD based on a comparison of the output value with a predetermined value.

[0056] This diagnostic method advantageously allows for the diagnosis of stage B2 DMVD without the need for echocardiography. As described above, dogs are traditionally classified as stage B2 if echocardiographic measurements of the left atrial and left ventricle exceed an established threshold indicating the presence of cardiac hypertrophy. However, access to echocardiography may be limited by factors related to the animal, the owner, and primary care practice.

[0057] The present invention overcomes this problem by enabling the diagnosis of stage B2 DMVD in the absence of echocardiography. The method generates an output value associated with the probability that a dog has stage B2 DMVD. The presence or absence of stage B2 DMVD can be diagnosed based on the output value. An output value below a certain threshold may indicate the absence of stage B2 DMVD. An output value above a certain threshold may indicate the presence of stage B2 DMVD. If the output value indicates the presence of stage B2 DMVD, treatment can be performed. Thus, the output value generated by the method can be used in place of traditional echocardiographic measurements to diagnose the presence or absence of B2 DMVD. Therefore, the method makes DMVD staging more accessible. As a result, it becomes easier to perform treatment (e.g., using pimobendan) at stage B2 before the onset of the disease, in accordance with the recommendations described in the EPIC trial.

[0058] Dogs, characteristic data, models, and output values The dogs, characteristic data, models, and output values ​​are described in detail above in relation to a method for screening dogs for stage B2 DMVD. Any aspect described in relation to a method for screening dogs for stage B2 DMVD can also be applied to a method for diagnosing stage B2 DMVD in dogs.

[0059] Diagnosis of DMVD at Stage B2 The Stage B2 DMVD method includes a step of diagnosing the presence or absence of a Stage B2 DMVD based on an output value and a comparison with a predetermined value.

[0060] The predetermined value (a pre-determined value) may be a “threshold” output value that can be used to include and / or exclude Stage B2 DMVDs. For example, the presence of Stage B2 DMVDs can be indicated by an output value associated with the probability that a dog has Stage B2 DMVDs, which is higher than or equal to the probability associated with the predetermined value. For example, the absence of Stage B2 DMVDs can be indicated by an output value associated with the probability that a dog has Stage B2 DMVDs, which is lower than the probability associated with the predetermined value. The predetermined value for indicating the presence of Stage B2 DMVDs may be the same as the predetermined value for indicating the absence of Stage B2 DMVDs. The predetermined value for indicating the presence of Stage B2 DMVDs may be different from the predetermined value for indicating the absence of Stage B2 DMVDs.

[0061] The presence of Stage B2 DMVD can be indicated by an output value associated with the probability that the dog has Stage B2 DMVD, which is greater than or equal to the probability that the positive predicted value is at least 75% (e.g., at least 80%, at least 85%, at least 90%, at least 95%, at least 96%, at least 97%, at least 98%, or at least 99%). Preferably, the presence of Stage B2 DMVD can be indicated by an output value associated with the probability that the dog has Stage B2 DMVD, which is greater than or equal to the probability that the positive predicted value is at least 95%. The probability that the positive predicted value is at least 95% may be 0.872.

[0062] The absence of a Stage B2 DMVD can be indicated by an output value associated with the probability that the dog has a Stage B2 DMVD, which is less than or equal to the probability that the negative prediction is at least 75% (e.g., at least 80%, at least 85%, at least 90%, at least 95%, at least 96%, at least 97%, at least 98%, or at least 99%). Preferably, the absence of a Stage B2 DMVD can be indicated by an output value associated with the probability that the dog has a Stage B2 DMVD, which is less than or equal to the probability that the negative prediction is at least 95%. The probability that the negative prediction is at least 95% may be 0.106.

[0063] Preferably, the presence of a Stage B2 DMVD is indicated by an output value associated with the probability that the dog has a Stage B2 DMVD, which is greater than or equal to the probability that a positive predicted value is at least 75% (e.g., at least 80%, at least 85%, at least 90%, at least 95%, at least 96%, at least 97%, at least 98%, or at least 99%), and the absence of a Stage B2 DMVD is indicated by an output value associated with the probability that the dog has a Stage B2 DMVD, which is less than or equal to the probability that a negative predicted value is at least 75% (e.g., at least 80%, at least 98%, or at least 99%). The positive predicted value may be at least 75%, and the negative predicted value may be at least 75%. The positive predicted value may be at least 80%, and the negative predicted value may be at least 80%. Positive predictions may be at least 85%, and negative predictions may be at least 85%. Positive predictions may be at least 90%, and negative predictions may be at least 90%. Positive predictions may be at least 95%, and negative predictions may be at least 95%. Positive predictions may be at least 96%, and negative predictions may be at least 96%. Positive predictions may be at least 97%, and negative predictions may be at least 97%. Positive predictions may be at least 98%, and negative predictions may be at least 98%. Positive predictions may be at least 99%, and negative predictions may be at least 99%. Preferably, positive predictions are at least 95%, and negative predictions are at least 95%.

[0064] The presence of a Stage B2 DMVD can be indicated by an output value greater than or equal to 0.6, associated with the probability that a dog has a Stage B2 DMVD. The presence of a Stage B2 DMVD can also be indicated by an output value between 0.7 and 1.0, associated with the probability that a dog has a Stage B2 DMVD, such as 0.72-0.98, 0.74-0.96, 0.76-0.94, 0.78-0.92, 0.8-0.9, 0.82-0.88, or 0.84-0.86. The presence of DMVD in Stage B2 is determined by, for example, values ​​greater than or equal to 0.65, greater than or equal to 0.7, greater than or equal to 0.75, greater than or equal to 0.8, greater than or equal to 0.825, greater than or equal to 0.85, greater than or equal to 0.870, greater than or equal to 0.872, greater than or equal to 0.875, greater than or equal to 0.9, and 0.92. The presence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, such as a value greater than or equal to 5, greater than or equal to 0.95, greater than or equal to 0.96, greater than or equal to 0.97, greater than or equal to 0.975, greater than or equal to 0.98, greater than or equal to 0.985, greater than or equal to 0.99, or greater than or equal to 0.995. The presence of a Stage B2 DMVD can be indicated, for example, by an output value associated with the probability that a dog has a Stage B2 DMVD, such as a value greater than or equal to 0.7. The presence of a Stage B2 DMVD can be indicated, for example, by an output value associated with the probability that a dog has a Stage B2 DMVD, such as a value greater than or equal to 0.75. The presence of a Stage B2 DMVD can be indicated, for example, by an output value associated with the probability that a dog has a Stage B2 DMVD, which is greater than or equal to 0.8.Preferably, the presence of a stage B2 DMVD can be indicated by an output value greater than or equal to 0.872, which is associated with the probability that the dog has a stage B2 DMVD.

[0065] The absence of a Stage B2 DMVD can be indicated by an output value less than or equal to 0.995, associated with the probability that a dog has a Stage B2 DMVD. The absence of a Stage B2 DMVD can also be indicated by an output value between 0 and 0.8, associated with the probability that a dog has a Stage B2 DMVD, such as 0.05-0.75, 0.1-0.7, 0.15-0.65, 0.2-0.6, 0.25-0.55, 0.3-0.5, 0.35-0.45, etc. The absence of Stage B2 DMVD can be indicated by output values ​​associated with the probability that a dog has Stage B2 DMVD, for example, less than 0.99, less than 0.985, less than 0.98, less than 0.975, less than 0.97, less than 0.96, less than 0.95, less than 0.925, less than 0.9, less than 0.875, less than 0.872, less than 0.87, less than 0.85, less than 0.825, less than 0.8, less than 0.75, less than 0.7, less than 0.65, less than 0.6, less than 0.55, less than 0.45, less than 0.4, less than 0.35, less than 0.3, less than 0.25, less than 0.2, less than 0.15, less than 0.125, less than 0.11, less than 0.106, or less than 0.1. The absence of a Stage B2 DMVD can be indicated, for example, by an output value associated with the probability that a dog has a Stage B2 DMVD, such as less than 0.8. The absence of a Stage B2 DMVD can be indicated, for example, by an output value associated with the probability that a dog has a Stage B2 DMVD, such as less than 0.75. The absence of a Stage B2 DMVD can be indicated, for example, by an output value associated with the probability that a dog has a Stage B2 DMVD, such as less than 0.7. Preferably, the absence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, such as less than 0.106.

[0066] For example, the presence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, greater than or equal to 0.6, 0.7, 0.75, 0.8, 0.85, 0.872, 0.9, 0.95, 0.96, 0.97, 0.98, or 0.99, and the absence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, less than 0.106, 0.6, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95, 0.96, 0.97, 0.98, or 0.99. The presence of a Stage B2 DMVD can be indicated by output values ​​associated with the probability that a dog has a Stage B2 DMVD, which are greater than or equal to 0.6, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95, 0.96, 0.97, 0.98, 0.99, or 0.872, respectively, and the absence of a Stage B2 DMVD can be indicated by output values ​​associated with the probability that a dog has a Stage B2 DMVD, which are greater than or equal to 0.6, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95, 0.96, 0.97, 0.98, 0.99, or less than 0.106. The presence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD that is greater than or equal to 0.6, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95, 0.96, 0.97, 0.98, or 0.99, and the absence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD that is less than 0.6, 0.55, 0.5, 0.45, 0.4, 0.35, 0.3, 0.25, 0.2, 0.15, or 0.125.The presence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, greater than or equal to 0.6, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95, 0.96, 0.97, 0.98, or 0.99, respectively. The absence of a Stage B2 DMVD can be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, greater than or equal to 0.6, 0.55, 0.5, 0.45, 0.4, 0.35, 0.3, 0.25, 0.2, 0.15, or less than 0.125. The presence of a Stage B2 DMVD can also be indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, greater than or equal to 0.872. The absence of a Stage B2 DMVD can be indicated by an output value less than 0.106, associated with the probability that a dog has a Stage B2 DMVD. Preferably, the presence of a Stage B2 DMVD is indicated by an output value greater than or equal to 0.872, associated with the probability that a dog has a Stage B2 DMVD, and the absence of a Stage B2 DMVD is indicated by an output value less than 0.106, associated with the probability that a dog has a Stage B2 DMVD.

[0067] The presence or absence of a diagnosed stage B2 DMVD can be used to inform clinical decisions, such as the decision to initiate treatment with cardiac medication. For example, if the presence of stage B2 DMVD is diagnosed, treatment with cardiac medication can be initiated. The cardiac medication may be, for example, pimobendan.

[0068] training method The present invention provides a method for training a model to predict stage B2 DMVD in dogs, comprising the steps of (i) processing dog-associated characteristic data, which includes two or more of the following: appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration, using a model to output output values; (ii) comparing the output values ​​with a diagnosis of the presence or absence of stage B2 DMVD in dogs; and (iii) adjusting the parameters of the model based on the results of the comparison. The training method may further include a step in which steps (i) to (iii) are repeated one or more times, wherein the characteristic data pertains to a different dog each time steps (i) to (iii) are performed.

[0069] Essentially, the model is provided with trait data for dogs already diagnosed with the presence or absence of stage B2 DMVD. The output values ​​generated by this model are compared to known diagnoses to check for consistency. The model parameters are adjusted based on the results of the comparison. For example, if an output value is associated with a low probability of stage B2 DMVD, but the dog is actually diagnosed with stage B2 DMVD, the model parameters are adjusted so that reprocessing of the trait data produces an output value that better reflects a positive diagnosis (e.g., an output value associated with a higher probability of stage B2 DMVD). In this way, the accuracy and reliability of the model can be improved. Iteration step (iv) results in an iterative improvement of the model using data derived from a population of dogs already diagnosed with the presence or absence of stage B2 DMVD. Comparison of model outputs and adjustment of model parameters can be performed using trait data associated with multiple dogs previously diagnosed with the presence or absence of stage B2 DMVD. Comparing model outputs and tuning model parameters can be done by minimizing the loss function used to compare model outputs using a gradient descent process. Model outputs can also be compared using a cross-validation process, where characteristic data associated with multiple dogs is divided into multiple subsets, and each subset is compared separately with the model output.

[0070] When a model is derived using a machine learning or regression process, the training methods discussed above can be applied. For example, if a model is derived using a regularized regression process, the above training method can be used, where the penalty function of the regularized regression process is a model parameter that can be adjusted based on the results of the comparison. When using an SVM process, the hyperplane position is a model parameter that can be adjusted based on the results of the comparison. When using a random forest process, the parameters that can be adjusted based on the results of the comparison are one or more types of characteristic data used to make decisions at each node in the tree, the threshold of the type of characteristic data used to make decisions at each node, and the number of nodes in the tree. The hyperparameters of the process used to derive the model may also be parameters that can be adjusted based on the results of the comparison.

[0071] dog A method for training a model to predict stage B2 DMVD in dogs includes the step of comparing the output value of step (i) with a diagnosis of the presence or absence of stage B2 DMVD in a dog to which the characteristic data relate. In other words, it is already known whether the dog to which the characteristic data relates has stage B2 DMVD. The diagnosis of the presence or absence of stage B2 DMVD may be based on echocardiography. For example, echocardiography can be used to determine the ratio of the left atrium to the aortic root (LA:Ao) and / or the left ventricular end-diastolic diameter (LVIDDN) normalized to body weight (kg). If LA:Ao is greater than or equal to 1.6 and LVIDDN is greater than or equal to 1.7, stage B2 DMVD can be identified as present. If LA:Ao is less than 1.6 and / or LVIDDN is less than 1.7, stage B2 DMVD can be identified as not present.

[0072] The dogs associated with the trait data may be at least 1 year old (at least 2 years, at least 3 years, at least 4 years, at least 5 years, at least 6 years, at least 7 years, at least 8 years, at least 9 years, or at least 10 years old, etc.). Preferably, the dogs associated with the trait data may be at least 6 years old.

[0073] The weight of the dogs associated with the characteristic data may be at least 1 kg (at least 2 kg, at least 5 kg, at least 10 kg, at least 15 kg, at least 20 kg, at least 25 kg, or at least 30 kg, etc.). The weight of the dogs associated with the characteristic data may be 1 kg to 70 kg, such as 2 kg to 65 kg, 5 kg to 60 kg, 10 kg to 55 kg, 15 kg to 50 kg, 20 kg to 45 kg, 30 kg to 40 kg, or 30 kg to 35 kg. Preferably, the weight of the dogs associated with the characteristic data is 2 kg to 25 kg.

[0074] Preferably, dogs associated with the characteristic data have a left apical systolic murmur that shows its strongest point on the mitral valve. Preferably, dogs associated with the characteristic data do not have radiographic, histological, or physical examination findings consistent with congestive heart failure. Preferably, dogs associated with the characteristic data had not been treated with loop diuretics at the time of data collection. Preferably, dogs associated with the characteristic data had not been treated with pimobendan at the time of data collection. Preferably, dogs associated with the characteristic data do not have comorbidities that are expected to interfere with echocardiographic measurements or biomarker concentrations.

[0075] Dogs are further described in detail above in relation to methods for screening dogs for stage B2 DMVD. Any aspect described in relation to methods for screening dogs for stage B2 DMVD can also be applied to methods for training models to predict stage B2 DMVD in dogs.

[0076] Characteristic data, model, and output values Characteristic data, models, and output values ​​are described in detail above in relation to a method for screening dogs for stage B2 DMVD. Any aspect described in relation to a method for screening dogs for stage B2 DMVD can also be applied to a method for training a model to predict stage B2 DMVD in dogs.

[0077] Parameter adjustment The parameters of a model can be adjusted by adjusting the weights assigned to one or more characteristic data. For example, in a model derived using a regression process, the weights associated with one or more characteristic data in the derived function can be changed during adjustment. In a model derived using an SVM process, the position of the hyperplane used to classify the input vectors associated with the characteristic data can be modified. In a model derived using a random forest process, the type of characteristic data, the number of decision trees, the number of nodes in one or more decision trees, and the threshold applied by any one node in the decision trees can be adjusted.

[0078] Computer programs and systems The methods and processes described herein can be embodied as code (e.g., software code) and / or data. Such code and data can be stored on one or more computer-readable media, which may include any device or medium capable of storing code and / or data for use by a computer system. When a computer system reads and executes code and / or data stored on a computer-readable media, the computer system implements the methods and processes embodied as data structures and code stored in the computer-readable storage medium. In certain embodiments, one or more steps of the methods and processes described herein can be implemented by a processor (e.g., a processor in a computer system or data storage system). Those skilled in the art should understand that computer-readable media include removable and non-removable structures / devices that can be used for storing information, such as computer-readable instructions, data structures, program modules, and other data used by a computer system / environment. Computer-readable media include, but are not limited to, volatile memory such as random-access memory (RAM, DRAM, SRAM); non-volatile memory such as flash memory, various read-only memories (ROM, PROM, EPROM, EEPROM), magnetic and ferromagnetic / ferroelectric memories (MRAM, FeRAM); magnetic and optical storage devices (hard drives, magnetic tapes, CDs, DVDs); network devices; or other media currently known or to be developed that can store computer-readable information / data. Computer-readable media should not be understood or interpreted as containing propagating signals.

[0079] The system may include an input device 10 configured to receive characteristic data associated with dogs, a model 20 configured to receive characteristic data to generate output values, and an output device 30 configured to output the output values ​​generated by the model. A diagrammatic example of the system is shown in Figure 7. Each component of the system may be in a single location, or at least one component may be in a different location from the others. Data may be transmitted and received between each component. For example, the input device may be located in a first location, such as a veterinary clinic. The model may be located in a second location. Upon receiving characteristic data, the input device may transmit the characteristic data to the model located in the second location. The input device may perform transformations of the characteristic data before transmitting it, such as a compression process. The model may receive the characteristic data, generate output values, and transmit the output values ​​to an output device which may be located in the first location.

[0080] It should be understood that different applications of the disclosed products and methods can be adapted to meet specific requirements in the art. The terms used herein are for the purpose of describing only specific embodiments of the invention and are not intended to limit them.

[0081] In addition, as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include multiple references unless the context otherwise clearly indicates otherwise. Thus, for example, a reference to “a peptide” includes “peptides,” and a reference to “a nanoparticle” includes two or more such nanoparticles, and so on.

[0082] All publications, patents, and patent applications cited herein, whether listed above or below, are incorporated herein by reference in their entirety.

[0083] The following examples illustrate the present invention. [Examples]

[0084] method The trial was predictive and cross-sectional in its design. Recruitment was international, and data were captured at 17 sites in Germany, 25 in the United Kingdom, and 16 in the United States (USA). Cases were recruited for the trial between January 2018 and January 2019. Examinations of the affected animals were performed by veterinary cardiac specialists who possessed at least one of the following requirements: a license in the subspecialty of cardiology from the European College of Veterinary Internal Medicine (ECVIM-CA) or the American College of Veterinary Internal Medicine (ACVIM); a license in cardiology from the Royal College of Veterinary Surgeons (RCVS); a cardiology license from the RCVS; membership in the Collegium Cardiologicum (CC); or membership in the Cardiology Working Group of the Deutsche Gesellschaft fur Kleintiermedizin - Deutsche Veterinarmedizinische Gesellschaft (DGK-DVG). Participation in training was permitted for trainees if they were under the direct supervision of a qualified cardiologist. The collection and storage of patient data was carried out with the consent of the owners and with the authorization of the Ethics and Welfare Committee of the Royal Veterinary College (URN: 2017 1749-3).

[0085] Case Selection The test population consisted of dogs owned by the requester that had already received a diagnostic evaluation for heart disease. A dog was considered eligible for the study if it had received a diagnosis of DMVD by a board-certified veterinary cardiologist based on echocardiography. This was defined as visible prolapse or thickening of the mitral valve and associated structures, along with mitral regurgitation on color Doppler examination. Dogs were required to be at least 6 years of age, weigh between 2 and 25 kg, and have a left apical systolic murmur with its loudest point over the mitral valve. Dogs with radiographic, histologic, or physical examination findings consistent with CHF, or those that had already received treatment with loop diuretics, were excluded from the test population. Concurrent diseases that were expected to interfere with echocardiographic measurements or biomarker concentrations were considered additional reasons for exclusion.

[0086] Evaluated population A "clean" population was created from the overall "complete" population to remove the influence of potential confounding factors derived from the analysis. Sick animals that violated the criteria for test subjects, such as those showing hypernitremia, hypercalcemia, endocrine disorders, or moderate to marked elevation of alanine aminotransferase (ALT), as well as sick animals whose samples took more than 72 hours to arrive at the reference laboratory after collection, were excluded from this refined population. Dogs receiving treatment with pimobendan were also excluded to eliminate the reported effect of the drug on echocardiographic dimensions. Data from the sick animals excluded at this stage were retained and used to form an "excluded" population for use in sub-analysis. 26 Sick animals that violated the criteria for test subjects, such as those showing hypernitremia, hypercalcemia, endocrine disorders, or moderate to marked elevation of alanine aminotransferase (ALT), as well as sick animals whose samples took more than 72 hours to arrive at the reference laboratory after collection, were excluded from this refined population. Reported effects of drugs on echocardiographic dimensions 7 To exclude, dogs receiving treatment with pimobendan were also excluded. Data from the sick animals excluded at this stage were retained and used to form an "excluded" population for use in sub-analysis.

[0087] Clinical evaluation At the time of examination, data were captured by a board-certified veterinary cardiologist. The presence of cough for more than 6 months, as well as changes in appetite and exercise tolerance, were recorded. Heart rate and respiratory rate were measured, and the main heart rhythm through auscultation was classified as sinus rhythm, atrial arrhythmia, or "other". Heart murmur intensity was on the Levine scale 27We initially assigned values ​​I-VI using [a specific method / tool]. For analysis, we reclassified these to reduce complexity. Grade I and II murmurs were labeled as soft, grade III as moderate, grade IV as loud, and grades V and VI as tremor. 14 Body Condition Score (BCS) is assessed using the American Animal Hospital Association's 9-point scale. 28 The scores were calculated using the following method. Echocardiography was performed on all dogs to obtain a standard right parasternal view. The ratio of the left atrium to the aortic root was recorded from a short-axis 2D image in early ventricular diastole. 29 Left ventricular end-diastolic diameter (LVIDD) was recorded at the chordae tendineae level in short axis, M-mode. 30 The formula for LVIDD is: LVIDDN = LVIDD(cm) / body weight 0.294 (LVIDDN) normalized to body weight using (kg) 31 In accordance with the guidelines developed by ACVIM, if LA:Ao ≥ 1.6 and LVIDDN ≥ 1.7, the affected animal was considered to be in stage B2. 4 Dogs that did not meet both of these criteria were classified as Stage B1. Where available, vertebral heart scores were recorded for use in sub-analyses. 32、33 .

[0088] Venous blood samples were collected from all dogs, and serum biochemical analysis and cardiac biomarker concentrations were obtained. Processed aliquots on ice were sent to research laboratories in Germany or the USA, depending on the sample origin (IDEXX BioResearch, Ludwigsburg, Germany; IDEXX BioAnalytics, West Sacramento, California, USA). Plasma NT-proBNP concentrations (second-generation ELISA: Canine Cardiopet® proBNP) and serum biochemical profiles were prepared upon receipt. Serum samples for cTnI measurement (two-site immunoenzyme sandwich assay: Beckman Access 2 troponin assay) were stored at -80°C, and batches were processed after recruitment was complete. Lower limits were assigned to results below the detection limit of the assay. 9、23 .

[0089] Analysis method The analysis was performed using commercially available software and open-source freeware (Python Software Foundation. Python Language Reference, version 3.7; R 1.2.5019, R Foundation for Statistical Computing, Vienna, Austria; SPSS version 26.0 for Macintosh, released 2018, SPSS Inc. San Diego, USA). Statistical significance was set at P<0.05. Continuous data were reported as medians (25th and 75th percentiles), and categorical variables were presented as proportions (frequency). The normality of continuous variables was evaluated by visually inspecting the histograms. If the transformation did not result in a Gaussian distribution, variables showing a prominent right tail were logarithmically transformed or classified into quartiles. Collinearity between continuous variables was observed if Spearman's rho was greater than 0.7. Categorical variables with small group sizes were reclassified to broader levels before inclusion in the analysis.

[0090] Identification of factors associated with the presence of DMVD in stage B2 Within the "clean" population, binary logistic regression was used to identify risk factors associated with having stage B2 disease. Cases were bifurcated according to whether they were stage B2 and whether clinical data and blood test concentrations were enrolled as explanatory variables. 4 The laboratory location was tested as a potential confounding factor. A univariate-restricted cubic spline model was used to evaluate the assumption of linearity with the logit. 34 If this rule was violated, the continuous variable was classified into quartiles for all subsequent analyses. 35 Variables that showed an association with the outcome at the univariate level (P<0.2) were included in the explanatory multivariate analysis, and a preliminary main effects model was selected using likelihood ratio testing with backward stepwise elimination. 36 If variables excluded by the univariate trial are individually registered in the main effects model and the substantial change (>20%) of the coefficients showing confounding effects is derived, then the retained variables are... 36 Two-way interaction terms were tested for reasonable combinations of variables, and if they showed a significant association with disease stage, they were included in the multivariate model. 36 The posterior estimated marginal means were calculated for all classification variables retained in the final model. The results are reported as coefficients (β) and odds ratios (OR), along with their 95% confidence intervals (CI).

[0091] Comparison of discriminative ability in alternating settings Model performance was evaluated by plotting receiver operating characteristic (ROC) curves using predicted probabilities and calculating the area under the curve (AUC) along with 95% confidence intervals. To assess the extent to which comorbidities, sample handling, or pimobendan administration affected discriminative ability, coefficients for the explanatory multivariate model were applied to data from the "clean" and "excluded" populations. The AUCs for the clean, complete, and excluded populations were compared using the DeLong test. 37The discriminative ability of the explanatory multivariate model was further compared with other methods that can be used to identify stage B2 DMVD. In separate univariate logistic regression models for which AUC was calculated, disease stage was regressiond against NT-proBMP only and against vertebral heart score only.

[0092] Evaluation of the predictive performance of classifiers trained to identify stage B2 diseases. We developed a series of diagnostic classifiers to evaluate how easily they can predict pre-symptomatic disease states. The models tested included logistic regression and ridge regression. 38 Support Vector Machines (SVMs) 39 Random Forest 40 and Gradient Boosting Machine (GBM) XGBoost 41 The clean data was distributed, with 80% used for training and the remaining 20% ​​kept separately as a holdout test population. Rows containing missing data were not included in this split. A transformation function for data preprocessing was developed on the training set and applied to the test data at prediction points. For all models except logistic regression and decision tree algorithms, continuous variables were expressed using the formula:

[0093]

number

[0094] Scaling was performed using [a specific method]. Categorical variables were dummy encoded (kl) for logistic regression and one-hot encoded (k) for other algorithmic types. Where applicable, the optimal combination of hyperparameters was selected using grid search in the hyperparameter space, and each combination was tested in an internal 5x cross-validation loop. For the development of the predictive logistic model, features were selected by backward stepwise elimination using likelihood ratio tests. A significance level of 0.05 was used to select the most frugal model. The tuned model was then fitted to 500 bootstraps randomly sampled from the training set using substitution. 42 By calculating the standard deviation of the results 43 We estimated the variability of the predictions. We fitted the model to a test set to generate predictions of the outcome state. We evaluated the overall accuracy using Brier scores, which represent the mean squared error between the predicted probability and the outcome. 44 The ability to distinguish between stages B1 and B2 was evaluated using AUC (95% CI), and the agreement between the predicted probability and the number of affected animals in the study population was evaluated using the intercept (calibration in the large) and slope of the calibration curve. 44~46 We created variable importance plots to evaluate the agreement between models and examined the capabilities of black-box algorithms such as SVM, Random Forest, and GBM.

[0095] result The complete test population consisted of 1887 dogs with pre-symptomatic DMVD. 642 dogs were excluded from the complete population based on age (n=56, 2.75%), body weight (n=10, 0.53%), comorbidities (n=126, 6.68%), pimobendan administration (n=361, 19.13%), or due to errors in sample handling (n=162, 8.59%). Of the 361 dogs treated with pimobendan, 56.51% (n=204) met the criteria for stage B2 disease. After excluding these dogs, a clean population of 1245 dogs was created (Figure 1).

[0096] Of the clean population, 27.1% (n=337) of dogs were classified as having stage B2 disease. The most common breeds evaluated were Cavalier King Charles Spaniels (n=292, 27.07%), followed by Chihuahuas (n=84, 6.74%), Jack Russell Terriers (n=56, 4.50%), Shih Tzus (n=43, 3.45%), and Cocker Spaniels (n=43, 3.45%). The median age was 10.00 years (LQ, 8.08; UQ, 11.63), and the group consisted of more males (n=718, 57.67%) than females (n=527, 42.33%). 31% of dogs (n=387) reported clinical signs, of which coughing was the most common symptom (n=299, 24.02%). Only 14.1% (n=175) of dogs that underwent chest imaging reported a VHS score, with a median score of 11.00 (LQ, 10.50; UQ, 11.50). Further descriptive statistics are reported in Tables 1 and 2 and Supplementary Tables 1 and 2.

[0097] Identification of factors associated with the presence of DMVD in stage B2 In univariate testing, 18 variables showed association with disease stage. In multivariate analysis, the following variables were identified as independent risk factors: age, ALT activity, appetite, BCS, creatinine concentration, murmur intensity, and NT-proBNP concentration (Table 3). Reduced appetite and lower physical condition scores were associated with higher odds of being in stage B2, and post-hoc testing of BCS showed this to be true when underweighted scores (BCS ≤ 3) were compared to almost all other values ​​(Table 4b). Estimated marginal mean of murmur intensity showed that the likelihood of being in stage B2 was higher when more murmurs were heard, and the comparison between loud murmurs and tremor murmurs was the only pairwise combination that did not differ statistically (Table 4c). Age of the affected animal was also associated with the outcome, with dogs aged 8-10 years having the highest risk. In dogs older than 10 years, the likelihood of being in stage B2 was significantly lower (Table 4a). Elevated serum creatinine levels were associated with a slight decrease in the odds of being stage B2 (β, -0.02; OR, 0.98, CI, 0.97~0.99; P<0.001). In contrast, when these variables were modeled as main effects, log10(NT-proBNP) and log 10 The higher the value of (ALT), the higher the likelihood. ALT and NT-proBNP interact negatively, which is log 10 The relationship between (NT-proBNP) and the outcome is log 10 A higher (ALT) value indicates that the intensity was not very strong (Figure 6).

[0098] Comparison of discriminative ability in alternating settings The final explanatory multivariate model, including all significant predictors, demonstrated good differentiation between pre-symptomatic disease stages (AUC, 0.84; 95% CI, 0.82–0.87; Nagelkerke's R). 2(AUC, 0.42) (Figure 2). When applied to the complete population, discriminative performance decreased slightly (AUC, 0.81; 95% CI, 0.79~0.83; P=0.048) (Figure 3). In the excluded population, which included data not used for model derivation, performance was fair but significantly lower than when tested under more optimal conditions (AUC, 0.76; 95% CI, 0.72~0.80; P<0.001). 36 Univariate models constructed for NT-proBNP and VHS found that these variables were positively associated with the odds of being in stage B2 (NT-proBNP: β, 3.65; OR, 38.45; 95% CI, 23.14~65.42; P<0.001; Nagelkerke's R). 2 , 0.26, VHS:β, 1.28;OR, 3.81, 95%CI, 2.33~5.96, P<0.001, Nagelkerke's R 2 , 0.29). The AUC values ​​were lower and had wider confidence intervals than the multivariate explanatory model, indicating that the model performance was poor when using these alternative diagnostic methods (NT-proBNP: AUC, 0.77; 95% CI, 0.74~0.80, VHS: AUC, 0.76; 95% CI, 0.69~0.83) (Figure 4). This difference in AUC indicated that both models based on a single parameter performed significantly worse than the explanatory multivariate model (NT-proBNP: P<0.001, VHS: P=0.032).

[0099] Evaluation of the predictive accuracy of a classifier trained to identify stage B2 disease. When evaluating AUC for prediction on the test data, model performance was relatively consistent across classifiers, with a mean of 0.87, indicating that all models generalized well to new data. Global calibration was positive for all models, indicating a small tendency for some models to overestimate prediction probabilities overall. 44Further performance metrics are summarized in Table 5. NT-proBNP and murmur grade were consistently found among the most important predictors, with NT-proBNP ranking first in all models tested (Figure 5). These variables, along with appetite, creatinine, and BCS, were included in a predictive logistic regression model (Table 6). This predictive logistic regression model performed similarly to a more complex explanatory model containing all variables, along with five features selected through backward stepwise elimination (AUC test, 0.86; 95% CI, 0.81–0.91), indicating that appetite, BCS, serum creatinine concentration, murmur intensity, and NT-proBNP concentration explained the majority of the variability in the data. The outcome obtained by this predictive model represents the predicted probability that an animal will have stage B2 disease. The presentation of results is influenced by how well the model performs. 47 Positive and negative predicted values ​​are provided to guide the interpretation (Table 7).

[0100] Consideration Our research found that clinical observations and cardiac biomarker concentrations can be used to predict the risk of dogs with DMVD being at stage B2. Using information obtained from a single examination of 1887 affected animals, a series of classifiers were able to predict the pre-symptomatic disease stage with good discrimination (mean AUC, 0.87) and calibration. 36 .

[0101] Previous research recognized the need to diversify the range of diagnostic options available for DMVD due to differences in the environment of affected animals. 9、18、24、48Information-based decisions made in the pre-symptomatic stage of the disease are particularly important to maximize the number of dogs that are properly managed. The predictive model defined in this study has the ability to act as an early screening test and quantifies the risk of having stage B2 disease. High risk scores can be used to select affected animals that would benefit from further examination, while low risk scores can identify dogs that are more likely to be stage B1. In this study, the model derived using multivariate logistic regression had similar predictive performance to other more complex classifiers tested. This may be advantageous because it requires fewer parameters to make predictions, thus reducing the cost that could be a barrier to adoption. 44 The model was internally validated against a 20% cohort holdout set, and based on this analysis, we were able to infer that it would function well in a typical canine population in primary care practice. The important next step is to evaluate the model's accuracy in the exact set of environments in which it is intended to be used. 44、49 .

[0102] All predictive models ranked NT-proBNP as the most important variable for distinguishing between stages B1 and B2. In explanatory analysis, the likelihood of having stage B2 disease increased with increasing NT-proBNP concentration, supporting the previous association with disease severity. 18、19、50、51 When comparing a multivariate explanatory model with one containing only NT-proBNP, it was clear that including other risk factors along with the biomarker increased discriminative performance and reduced the number of misclassified cases. Similar to capturing further sources of variation in outcome discrimination methods, this approach can have improved performance by controlling for the variability of the biomarker itself. In dogs, NT-proBNP concentrations are affected by comorbidities, and continuous measurements within the same individual exhibit biological variability. 52~54Therefore, the inclusion of markers with a disease severity level greater than 1 can improve predictive quality in cases showing abnormal biomarker concentrations. The findings of this explanatory analysis are similar to previous studies, indicating that NT-proBNP provides more information when interpreted in conjunction with other factors. 9、23、24 This results in improved accuracy when staging diseases before they manifest.

[0103] In addition to NT-proBNP, several other risk factors were identified. Another important predictor variable, murmur intensity, was previously associated with disease severity before the onset of the disease. 11、14 Consistently, explanatory analysis of our studies shows that the likelihood of being stage B2 increases with murmur grade, with dogs having large or trembling murmurs being at highest risk. Compared to other parameters included in this analysis, murmur intensity is one of the more subjective measures. Auscultation of heart sounds is subject to inter- and intra-observer variability, and given the apparent importance of this variable, it is potentially limited. 11、55 Previous studies have shown that using a simpler scheme improves agreement, and therefore, for the purposes of this study, audibility was graded using the four-level system proposed by Ljungvall et al. (2014). 14、56 It is even more important to note that all dogs were examined by a veterinary cardiac specialist using a standardized protocol. Further research is needed to evaluate whether sampling in different settings affects the accuracy of predictions.

[0104] Loss of appetite was found to increase the likelihood of being in stage B2. In DMVD, loss of appetite is considered a negative prognostic indicator, and dogs that subsequently develop CHF may experience weight loss. 5、7、48 Although body weight was not examined in this analysis, poor physical condition was associated with an increased risk of B2. Anorexia-cachexia syndrome is recognized in human patients with functional heart failure. 57、58In dogs, cachexia can occur before the onset of CHF and may result in changes that can be detected as clinical signs. Subsequent muscle mass loss may also affect serum creatinine concentration. This study observed a negative association between creatinine and odds of stage B2, supporting this hypothesis. Creatinine was selectively retained in the multivariate model, but this indicates that it described further variability except for glomerular filtration rate (GFR). Since increased circulating fluid volume has been shown to induce a more rapid creatinine clearance rate, 59~63 Furthermore, GFR itself can be expected to be associated with the severity of the disease before the onset of symptoms. Since GFR is a known confounding factor of biomarker concentrations, adjusting for creatinine in models containing NT-proBNP is potentially advantageous.

[0105] Age and ALT were associated with the likelihood of stage B2, but neither variable was retained in predictive logistic regression models derived from smaller subsets of data. This raises questions about the strength of these associations. In explanatory analysis, the highest risk was observed when dogs were 8–10 years old. After this age, older affected animals were less likely to have stage B2 disease. There is evidence that the tendency to remodel changes in aging hearts, but this has not been studied in DMVD. 64 Pro-fibrotic changes in myocardial composition may influence the development of eccentric hypertrophy. 65 Alternatively, these findings may reflect phenotypic differences within each age group. Early-onset DMVD, as seen in some breeds, may be accompanied by a more rapid rate of disease progression. 3、9 In humans, age is taken into consideration when defining diagnostic thresholds for NT-proBNP, and studies have shown that this adds further value to analyses already involving creatinine. 66 Thus, including age in the model can correct for potential confounding in dogs.

[0106] ALT was positively associated with the likelihood of stage B2 in the explanatory analysis. The hepatic vascular system is sensitive to changes in central venous pressure, and elevated ALT may occur secondary to cardiovascular disease as a result of congestion or decreased perfusion. 67、68 ALT has been shown to modify the strength of the relationship between NT-proBNP and disease stage, but the precise relevance of this finding in DMVD is unclear. At high ALT concentrations, NT-proBNP may be partially elevated as a result of liver disease, leading to a weaker association with the severity of DMVD. 69、70 .

[0107] This study benefited from the large number of patients examined. This facilitated robust analysis, particularly when developing predictive models for clinical use. When training models, algorithms can overfit to meaningless noise in the data, reducing their generalizability. 71 In this study, there were enough affected animals to form separate training and test cohorts. This allocation technique simulates performance under new conditions because the trained models are applied to sets of data they had not previously encountered. A comparison of several algorithms showed good agreement in model fit, internal generalizability, and most important variables. All models identified NT-proBNP as the highest-ranking variable, which supports studies describing its potential association in pre-symptomatic DMVD. 9、18、19 .

[0108] The data is substantial enough to evaluate several machine learning algorithms, providing them in comparison to more conventional regression-based models. Because the algorithms can describe complex nonlinear relationships between variables, machine learning has potential applications in medicine. 72 Using machine learning to distinguish between stages B1 and B2 did not yield a significant performance advantage in this study, which suggests that the linear parameterization method adequately captured the data structure. 73In this example, the logistic regression model can be considered more clinically useful because it provides a more economical and interpretable set of specifications. Machine learning is promising in veterinary medicine. 74、75 The results of this study emphasize that it does not always provide the optimal solution. Model selection depends equally on the intended use of the data and the model.

[0109] Predictive sampling of a large number of dogs captured data from other diagnostic tests performed at the time of examination. This was sufficient to allow a sub-analysis of VHS, an alternative method for identifying cardiac hypertrophy. The multi-parameter approach proved more accurate compared to single tests such as VHS or NT-proBNP. This is consistent with the results of other studies. 9、23、24、48 Although the difference in clinical utility was not precisely measured, integrating a single blood test into routine data allows users to avoid the risk of radiation exposure or chemosuppression. 76 .

[0110] conclusion In conclusion, this study demonstrates that the likelihood of a dog having stage B2 DMVD can be predicted using data obtained from multiple aspects of the examination of affected animals, specifically appetite, BCS, creatinine, heart murmur intensity, and NT-proBNP. This has potential as a screening test and can provide an informed method for allocating referrers and veterinary resources. Ultimately, the accurate application of clinical predictive models can improve outcomes for dogs with pre-symptomatic DMVD. table [Table 1] [Table 2] TIFF0007872230000005.tif114164 [Table 3] [Table 4] [Table 5] [Table 6] [Table 7] [Table 8] [Table 9] TIFF0007872230000013.tif122162 [Examples]

[0111] Example 2 provides further details of the ridge regression analysis described in Example 1. For this analysis, one-hot encoding was used and it ran within the Scikit-Learn library in Python. The lambda function was tuned to minimize the averaged Brier score across five cross-validation loops on the training set. The intercept was -1.226. [Table 10] TIFF0007872230000015.tif94164 For animals exhibiting decreased appetite, cough, decreased exercise tolerance, bilirubin (>3.4), GGT (>6), BCS (3), cTni (>0.08), heart rhythm (sinus rhythm), tremor murmur, respiratory rate (>32), age (>12), and sex (MN),

number

[0112] References 1. Fox PR. Pathology of myxomatous mitral valve disease in the dog. J Vet Cardiol. 2012;14(1):103-26. 2. Egenvall A, Bonnett BN, Hedhammar A, Olson P. Mortality in over 350,000 insured Swedish dogs from 1995-2000: II. Breed-specific age and survival patterns and relative risk for causes of death. Acta Vet Scand. 2005;46(3):121-36. 3. Mattin MJ, Boswood A, Church DB, McGreevy PD, O’Neill DG, Thomson PC, et al. Degenerative mitral valve disease: Survival of dogs attending primary-care practice in England. Prev Vet Med. 2015;122(4):436-42. 4. Keene BW, Atkins CE, Bonagura JD, Fox PR, Haeggstroem J, Fuentes VL, et al. ACVIM consensus guidelines for the diagnosis and treatment of myxomatous mitral valve disease in dogs. J Vet Intern Med. 2019 May 11;33(3):1127-40. 5. Boswood A, Haeggstroem J, Gordon SG, Wess G, Stepien RL, Oyama MA, et al. Effect of Pimobendan in Dogs with Preclinical Myxomatous Mitral Valve Disease and Cardiomegaly: The EPIC Study-A Randomized Clinical Trial. J Vet Intern Med. 2016 Nov;30(6):1765-79. 6. Patronek GJ, Waters DJ, Glickman LT. Comparative longevity of pet dogs and humans: Implications for gerontology research. Journals Gerontol - Ser A Biol Sci Med Sci. 1997 May 1;52(3):171-8. 7. Boswood A, Gordon SG, Haeggstroem J, Wess G, Stepien RL, Oyama MA, et al. Longitudinal Analysis of Quality of Life, Clinical, Radiographic, Echocardiographic, and Laboratory Variables in Dogs with Preclinical Myxomatous Mitral Valve Disease Receiving Pimobendan or Placebo: The EPIC Study. J Vet Intern Med. 2018 Dec 6;32(1):72-85. 8. Borgarelli M, Crosara S, Lamb K, Savarino P, La Rosa G, Tarducci A, et al. Survival Characteristics and Prognostic Variables of Dogs with Preclinical Chronic Degenerative Mitral Valve Disease Attributable to Myxomatous Degeneration. J Vet Intern Med. 2012 Jan;26(1):69-75. 9. Mattin MJ, Brodbelt DC, Church DB, Boswood A. Factors associated with disease progression in dogs with presumed preclinical degenerative mitral valve disease attending primary care veterinary practices in the United Kingdom. J Vet Intern Med. 2019 Mar 1;33(2):445-54. 10. Eriksson AS, Haeggstroem J, Pedersen HD, Hansson K, Jaervinen A-K, Haukka J, et al. Increased NT-proANP predicts risk of congestive heart failure in Cavalier King Charles spaniels with mitral regurgitation caused by myxomatous valve disease. J Vet Cardiol. 2014 Sep 1;16(3):141-54. 11. Pedersen HD, Haeggstroem J, Falk T, Mow T, Olsen LH, Iversen L, et al. Auscultation in Mild Mitral Regurgitation in Dogs: Observer Variation, Effects of Physical Maneuvers, and Agreement with Color Doppler Echocardiography and Phonocardiography. J Vet Intern Med. 1999 Jan 1;13(1):56-64. 12. Haggstrom J, Kvart C, Hansson K. Heart Sounds and Murmurs: Changes Related to Severity of Chronic Valvular Disease in the Cavalier King Charles Spaniel. J Vet Intern Med. 1995;9(2):75-85. 13. Ljungvall I, Ahlstrom C, Hoeglund K, Hult P, Kvart C, Borgarelli M, et al. Use of signal analysis of heart sounds and murmurs to assess severity of mitral valve regurgitation attributable to myxomatous mitral valve disease in dogs. Am J Vet Res. 2009;70(5):604-13. 14. Ljungvall I, Rishniw M, Porciello F, Ferasin L, Ohad DG. Murmur intensity in small-breed dogs with myxomatous mitral valve disease reflects disease severity. J Small Anim Pract. 2014;55(11):545-50. 15. Kvart C, Haeggstroem J, Pedersen HD, Hansson K, Eriksson A, Jaervinen A-K, et al. Efficacy of enalapril for prevention of congestive heart failure in dogs with myxomatous valve disease and asymptomatic mitral regurgitation. J Vet Intern Med. 2002;16(1):80-8. 16. Oyama MA, Rush JE, O’Sullivan ML, Williams RM, Rozanski EA, Petrie JP, et al. Perceptions and priorities of owners of dogs with heart disease regarding quality versus quantity of life for their pets. J Am Vet Med Assoc. 2008 Jul 1;233(1):104-8. 17. Chetboul V, Athanassiadis N, Concordet D, Nicolle A, Tessier D, Castagnet M, et al. Observer-dependent variability of quantitative clinical endpoints: the example of canine echocardiography. J Vet Pharmacol Ther. 2004 Feb;27(1):49-56. 18. Moonarmart W, Boswood A, Fuentes VL, Brodbelt D, Souttar K, Elliott J. N-terminal pro B-type natriuretic peptide and left ventricular diameter independently predict mortality in dogs with mitral valve disease. J Small Anim Pract. 2010 Feb 1;51(2):84-96. 19. Serres F, Pouchelon J-L, Poujol L, Lefebvre HP, Trumel C, Daste T, et al. Plasma N-terminal pro-B-type natriuretic peptide concentration helps to predict survival in dogs with symptomatic degenerative mitral valve disease regardless of and in combination with the initial clinical status at admission. J Vet Cardiol. 2009 Dec;11(2):103-21. 20. Oyama MA, Sisson DD. Cardiac troponin-I concentration in dogs with cardiac disease. J Vet Intern Med. 2004;18(6):831-9. 21. Spratt DP, Mellanby RJ, Drury N, Archer J. Cardiac troponin I: evaluation I of a biomarker for the diagnosis of heart disease in the dog. J Small Anim Pract. 2005 Mar;46(3):139-45. 22. Ljungvall I, Hoeglund K, Tidholm A, Olsen LH, Borgarelli M, Venge P, et al. Cardiac troponin I is associated with severity of myxomatous mitral valve disease, age, and C-reactive protein in dogs. J Vet Intern Med. 2010 Jan 1;24(1):153-9. 23. Hezzell MJ, Boswood A, Chang Y-M, Moonarmart W, Souttar K, Elliott J. The Combined Prognostic Potential of Serum High-Sensitivity Cardiac Troponin I and N-Terminal pro-B-Type Natriuretic Peptide Concentrations in Dogs with Degenerative Mitral Valve Disease. J Vet Intern Med. 2016;26(2):302-11. 24. Mattin MJ, Boswood A, Church DB, Brodbelt DC. Prognostic factors in dogs with presumed degenerative mitral valve disease attending primary-care veterinary practices in the United Kingdom. Journal of Veterinary Internal Medicine. 2019 Dec 18;432-44. 25. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstem AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996 Dec 1;49(12):1373-9. 26. Chapman SE, Hostutler RA. A Laboratory Diagnostic Approach to Hepatobiliary Disease in Small Animals. Vol. 43, Veterinary Clinics of North America - Small Animal Practice. 2013. p. 1209-25. 27. Levine SA. Notes on the Gradation of the Intensity of Cardiac Murmurs. JAMA J Am Med Assoc. 1961 Jul 29;177(4):261. 28. American Animal Hospital Association. Canine Body Condition Score for 1-9 and 1-5 Scales. Veterinary Forensics: Animal Cruelty Investigations. 2013. 29. Hansson K, Haeggstroem J, Kvart C, Lord P. Left atrial to aortic root indices using two-dimensional and M-mode echocardiography in cavalier King Charles Spaniels with and without left atrial enlargement. Vet Radiol Ultrasound. 2002;43(6):568-75. 30. Thomas WP, Gaber CE, Jacobs GJ, Kaplan PM, Lombard CW, Sydney Moise N, et al. Recommendations for Standards in Transthoracic Two‐Dimensional Echocardiography in the Dog and Cat. Vet Radiol Ultrasound. 1994;35(3):173-8. 31. Cornell CC, Kittleson MD, Della Torre P, Haeggstroem J, Lombard CW, Pedersen HD, et al. Allometric Scaling of M-Mode Cardiac Measurements in Normal Adult Dogs. J Vet Intern Med. 2004;18(1):311-21. 32. Buchanan JW, Buecheler J. Vertebral scale system to measure canine heart size in radiographs. J Am Vet Med Assoc. 1995 Jan 15;206(2):194-9. 33. Hansson K, Haeggstroem J, Kvart C, Lord P. Interobserver variability of vertebral heart size measurements in dogs with normal and enlarged hearts. Vet Radiol Ultrasound. 2005;46(2):122-30. 34. Dohoo IR, Martin SW, Stryhn H. Veterinary epidemiologic research. No. V413. Charlottetown, Canada: AVC Inc; 2003. 706 p. 35. Royston P, Sauerbrei W. Multivariable Model-Building: A Pragmatic Approach to Regression Analysis based on Fractional Polynomials for Modelling Continuous Variables. Multivariable Model-Building: A Pragmatic Approach to Regression Analysis based on Fractional Polynomials for Modelling Continuous Variables. Wiley Blackwell; 2008. 1-303 p. 36. Hosmer DW, Lemeshow S, Sturdivant RX. Applied Logistic Regression. Third. John Wiley & Sons, Ltd; 2013. 37. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the Areas under Two or More Correlated Receiver Operating Characteristic Curves: A Nonparametric Approach. Biometrics. 1988 Sep;44(3):837. 38. Hoerl AE, Kennard RW. Ridge Regression: Biased Estimation for Nonorthogonal Problems. Technometrics. 1970;12(1):55-67. 39. Cortes C, Vapnik V. Support-vector networks. Mach Learn. 1995 Sep;20(3):273-97. 40. Breiman L. Random forests. Mach Learn. 2001 Oct;45(1):5-32. 41. Chen T, Guestrin C. XGBoost: A scalable tree boosting system. In: Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York, New York, USA: Association for Computing Machinery; 2016. p. 785-94. 42. Efron B, Tibshirani RJ. An Introduction to the Bootstrap. Chapman & Hall / CRC; 1993. 43. Kulesa A, Krzywinski M, Blainey P, Altman N. Points of Significance: Sampling distributions and the bootstrap. Nat Methods. 2015 May 28;12(6):477-8. 44. Steyerberg EW, Vickers AJ, Cook NR, Gerds T, Gonen M, Obuchowski N, et al. Assessing the performance of prediction models: A framework for traditional and novel measures. Vol. 21, Epidemiology. 2010. p. 128-38. 45. Walsh CG, Sharman K, Hripcsak G. Beyond discrimination: A comparison of calibration methods and clinical usefulness of predictive models of readmission risk. J Biomed Inform. 2017 Dec 1;76:9-18. 46. Steyerberg EW, Vergouwe Y. Towards better clinical prediction models: seven steps for development and an ABCD for validation. Eur Heart J. 2014 Aug 1;35(29):1925-31. 47. Bonnett LJ, Snell KIE, Collins GS, Riley RD. Guide to presenting clinical prediction models for use in clinical settings. BMJ. 2019 Apr 17;365. 48. Lopez-Alvarez J, Elliott J, Pfeiffer D, Chang Y-M, Mattin M, Moonarmart W, et al. Clinical Severity Score System in Dogs with Degenerative Mitral Valve Disease. J Vet Intern Med. 2015;29:575-81. 49. Collins GS, de Groot JA, Dutton S, Omar O, Shanyinde M, Tajar A, et al. External validation of multivariable prediction models: a systematic review of methodological conduct and reporting. BMC Med Res Methodol. 2014 Dec 19;14(1):40. 50. Hezzell MJ, Boswood A, Moonarmart W, Elliott J. Selected echocardiographic variables change more rapidly in dogs that die from myxomatous mitral valve disease. J Vet Cardiol. 2012;14:269-79. 51. Chetboul V, Serres F, Tissier R, Lefebvre HPP, Sampedrano CC, Gouni V, et al. Association of Plasma N-Terminal Pro-B-Type Natriuretic Peptide Concentration with Mitral Regurgitation Severity and Outcome in Dogs with Asymptomatic Degenerative Mitral Valve Disease. J Vet Intern Med. 2009 Sep;23(5):984-94. 52. Oyama MA, Boswood A, Connolly DJ, Ettinger SJ, Fox PR, Gordon SG, et al. Clinical usefulness of an assay for measurement of circulating n-terminal pro-b-type natriuretic peptide concentration in dogs and cats with heart disease. J Am Vet Med Assoc. 2013 Jul 1;243(1):71-82. 53. Kellihan HB, Oyama MA, Reynolds CA, Stepien RL. Weekly variability of plasma and serum NT-proBNP measurements in normal dogs. J Vet Cardiol. 2009 May;11(Supplementary Materials):93-7. 54. Winter RL, Saunders AB, Gordon SG, Buch JS, Miller MW. Biologic variability of N-terminal pro-brain natriuretic peptide in healthy dogs and dogs with myxomatous mitral valve disease. J Vet Cardiol. 2017 Apr;19(2):124-31. 55. Hoeglund K, French A, Dukes-McEwan J, Haeggstroem J, Smith P, Corcoran B, et al. Low intensity heart murmurs in boxer dogs: Inter-observer variation and effects of stress testing. J Small Anim Pract. 2004 Apr;45(4):178-85. 56. Keren R, Tereschuk M, Luan X. Evaluation of a Novel Method for Grading Heart Murmur Intensity. Arch Pediatr Adolesc Med. 2005 Apr 1;159(4):329. 57. Fudim M, Wagman G, Altschul R, Yucel E, Bloom M, Vittorio TJ. Pathophysiology and treatment options for cardiac anorexia. Curr Heart Fail Rep. 2011 Jun;8(2):147-53. 58. Fonfara S, Hetzel U, Tew SR, Dukes-Mcewan J, Cripps P, Clegg PD. Leptin Expression in Dogs with Cardiac Disease and Congestive Heart Failure. J Vet Intern Med. 2011 Sep;25(5):1017-24. 59. Adlbrecht C, Kommata S, Hulsmann M, Szekeres T, Bieglmayer C, Strunk G, et al. Chronic heart failure leads to an expanded plasma volume and pseudoanaemia, but does not lead to a reduction in the body’s red cell volume. Eur Heart J. 2008 Oct 1;29(19):2343-50. 60. Androne A-S, Katz SD, Lund L, LaManca J, Hudaihed A, Hryniewicz K, et al. Hemodilution is common in patients with advanced heart failure. Circulation. 2003 Jan 21;107(2):226-9. 61. Abramov D, Cohen RS, Katz SD, Mancini D, Maurer MS. Comparison of blood volume characteristics in anemic patients with low versus preserved left ventricular ejection fractions. Am J Cardiol. 2008 Oct 15;102(8):1069-72. 62. Duarte K, Monnez JM, Albuisson E, Pitt B, Zannad F, Rossignol P. Prognostic Value of Estimated Plasma Volume in Heart Failure. JACC Heart Fail. 2015 Nov 1;3(11):886-93. 63. Dittrich S, Schuth A, Aurich H, vonLoeper J, Grosse-Siestrup C, Lange PE. Haemodilution improves organ function during normothermic cardiopulmonary bypass: investigations in isolated perfused pig kidneys. Perfusion. 2000 Jun;15(3):225-9. 64. Horn MA, Trafford AW. Aging and the cardiac collagen matrix: Novel mediators of fibrotic remodelling. Vol. 93, Journal of Molecular and Cellular Cardiology. Academic Press; 2016. p. 175-85. 65. Hezzell MJ, Falk T, Olsen LH, Boswood A, Elliott J. Associations between N-terminal procollagen type III, fibrosis and echocardiographic indices in dogs that died due to myxomatous mitral valve disease. J Vet Cardiol. 2014 Dec;16(4):257-64. 66. Bay M, Kirk V, Parner J, Hassager C, Nielsen H, Krogsgaard K, et al. NT-proBNP: A new diagnostic screening tool to differentiate between patients with normal and reduced left ventricular systolic function. Heart. 2003;89(2):150-4. 67. Nikolaou M, Parissis J, Yilmaz MB, Seronde MF, Kivikko M, Laribi S, et al. Liver function abnormalities, clinical profile, and outcome in acute decompensated heart failure. Eur Heart J. 2013 Mar;34(10):742-9. 68. Kubo SH, Walter BA, John DHA, Clark M, Cody RJ. Liver Function Abnormalities in Chronic Heart Failure: Influence of Systemic Hemodynamics. Arch Intern Med. 1987;147(7):1227-30. 69. Henriksen JH, Gotze JP, Fuglsang S, Christensen E, Bendtsen F, Moller S. Increased circulating pro-brain natriuretic peptide (proBNP) and brain natriuretic peptide (BNP) in patients with cirrhosis: Relation to cardiovascular dysfunction and severity of disease. Gut. 2003 Oct 1;52(10):1511-7. 70. Licata A, Corrao S, Petta S, Genco C, Cardillo M, Calvaruso V, et al. NT Pro BNP Plasma Level and Atrial Volume Are Linked to the Severity of Liver Cirrhosis. Vinciguerra M, editor. PLoS One. 2013 Aug 5;8(8):e68364. 71. Cawley GC, Talbot NLC. On over-fitting in model selection and subsequent selection bias in performance evaluation. Vol. 11, Journal of Machine Learning Research. 2010. 72. Rajkomar A, Dean J, Kohane I. Machine Learning in Medicine. N Engl J Med. 2019 Apr 4;380(14):1347-58. 73. Steyerberg EW, van der Ploeg T, Van Calster B. Risk prediction with machine learning and regression methods. Vol. 56, Biometrical Journal. Wiley-VCH Verlag; 2014. p. 601-6. 74. Reagan KL, Reagan BA, Gilor C. Machine learning algorithm as a diagnostic tool for hypoadrenocorticism in dogs. Domest Anim Endocrinol. 2020 Jul 1;72:106396. 75. Bradley R, Tagkopoulos I, Kim M, Kokkinos Y, Panagiotakos T, Kennedy J, et al. Predicting early risk of chronic kidney disease in cats using routine clinical laboratory tests and machine learning. J Vet Intern Med. 2019 Nov 1;33(6):2644-56. 76. Bille C, Auvigne V, Libermann S, Bomassi E, Durieux P, Rattez E. Risk of anaesthetic mortality in dogs and cats: An observational cohort study of 3546 cases. Vet Anaesth Analg. 2012;39(1):59-68. 77. Collins GS, Ogundimu EO, Altman DG. Sample size considerations for the external validation of a multivariable prognostic model: A resampling study. Stat Med. 2016 Jan 30;35(2):214-26. 78. Riley RD, Ensor J, Snell KIE, Debray TPA, Altman DG, Moons KGM, et al. External validation of clinical prediction models using big datasets from e-health records or IPD meta-analysis: Opportunities and challenges. BMJ. 2016;353. This disclosure includes the following embodiments. [1] A method for diagnosing stage B2 degenerative mitral valve disease (DMVD) in dogs, (a) A step of receiving characteristic data associated with a dog, wherein the characteristic data includes two or more of the following: appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; (b) A step of processing characteristic data using a model, wherein the output of the model is an output value associated with the probability that a dog has a DMVD of stage B2; and (c) A step to diagnose the presence or absence of the DMVD in stage B2 based on a comparison of the output value with a predetermined value. Methods that include... [2] The method according to Embodiment 1, wherein the presence of a Stage B2 DMVD is indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, which is greater than or equal to 0.872, and the absence of a Stage B2 DMVD is indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, which is less than 0.106. [3] A method for screening dogs for stage B2 DMVD, (a) A step of receiving characteristic data associated with a dog, wherein the characteristic data includes two or more of the following: appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; and (b) A step of processing characteristic data using a model, where the output of the model is an output value associated with the probability that a dog has a DMVD of stage B2. Methods that include... [4] The method according to any one of embodiments 1 to 3, wherein the dog has been diagnosed with DMVD prior to step (a). [5] A method for training a model to predict stage B2 DMVD in dogs, (i) A step of processing trait data associated with a dog using a model for outputting output values, wherein the trait data includes two or more of the following: appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; (ii) A step of comparing the output value with a diagnosis of the presence or absence of stage B2 DMVD in dogs; and (iii) Step of adjusting the model parameters based on the results of the comparison Methods that include... [6] The method according to Embodiment 5, wherein the diagnosis of the presence or absence of stage B2 DMVD is based on echocardiography. [7] (iv) A step in which steps (i) to (iii) are repeated one or more times, wherein the characteristic data pertains to a different dog each time steps (i) to (iii) are performed. The method according to embodiment 5 or 6, further comprising: [8] The method according to any of embodiments 1 to 7, wherein the model is derived using a regression process. [9] The method according to embodiment 8, wherein the model is derived using multivariate logistic regression or regularized regression.

[10] The method according to any of embodiments 1 to 7, wherein the model is derived using a machine learning process.

[11] The method according to embodiment 10, wherein the model is derived using a support vector machine (SVM) process, a random forest process, or a gradient boosting process.

[12] The method according to any one of Embodiments 1 to 11, wherein the characteristic data includes two or more of the following: appetite, body condition score (BCS), creatinine concentration, heart murmur intensity, and NT-ProBNP concentration.

[13] The method according to any one of Embodiments 1 to 12, wherein the characteristic data includes NT-ProBNP concentration.

[14] The method according to any one of Embodiments 1 to 13, wherein the characteristic data includes NT-proBNP concentration, appetite, creatinine concentration and heart murmur intensity.

[15] The method according to Embodiment 14, wherein the characteristic data includes a BCS.

[16] The method according to Embodiment 15, wherein the characteristic data includes age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration.

[17] A computer program, when executed by a computer system, including coding means for instructing the computer system to carry out the method described in any of embodiments 1 to 16.

[18] A system for diagnosing stage B2 DMVD in dogs, An input device configured to receive characteristic data associated with a dog, wherein the characteristic data includes two or more of the following: appetite, BCS, creatinine concentration, heart murmur intensity, NT-proBNP concentration, age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration; A model configured to receive characteristic data and generate output values ​​associated with the probability that a dog has a stage B2 DMVD; and Output device configured to output an output value A system that includes this.

Claims

1. A method for diagnosing stage B2 degenerative mitral valve disease (DMVD) in dogs, (a) A step of receiving characteristic data associated with a dog, wherein the characteristic data includes NT-proBNP concentration, appetite, creatinine concentration, heart murmur intensity and body condition score (BCS); (b) A step of processing characteristic data using a model, wherein the output of the model is an output value associated with the probability that a dog has a DMVD of stage B2; and (c) A step to diagnose the presence or absence of the DMVD in stage B2 based on a comparison of the output value with a predetermined value. Methods that include...

2. The method according to claim 1, wherein the presence of a Stage B2 DMVD is indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, which is greater than or equal to 0.872, and the absence of a Stage B2 DMVD is indicated by an output value associated with the probability that a dog has a Stage B2 DMVD, which is less than 0.

106.

3. A method for screening dogs for stage B2 DMVD, (a) A step of receiving characteristic data associated with a dog, wherein the characteristic data includes NT-proBNP concentration, appetite, creatinine concentration, heart murmur intensity and body condition score (BCS); and (b) A step of processing characteristic data using a model, where the output of the model is an output value associated with the probability that a dog has a DMVD of stage B2. Methods that include...

4. The method according to any one of claims 1 to 3, wherein the dog has been diagnosed with DMVD before step (a).

5. A method for training a model to predict stage B2 DMVD in dogs, (i) A step of processing characteristic data associated with dogs using a model for outputting output values, wherein the characteristic data includes NT-proBNP concentration, appetite, creatinine concentration, heart murmur intensity and body condition score (BCS); (ii) A step of comparing the output value with a diagnosis of the presence or absence of stage B2 DMVD in dogs; and (iii) Step of adjusting the model parameters based on the results of the comparison Methods that include...

6. The method according to claim 5, wherein the diagnosis of the presence or absence of stage B2 DMVD is based on echocardiography.

7. (iv) A step in which steps (i) to (iii) are repeated one or more times, wherein the characteristic data pertains to a different dog each time steps (i) to (iii) are performed. The method according to claim 5 or 6, further comprising:

8. The method according to any one of claims 1 to 7, wherein the model is derived using a regression process.

9. The method according to claim 8, wherein the model is derived using multivariate logistic regression or regularized regression.

10. The method according to any one of claims 1 to 7, wherein the model is derived using a machine learning process.

11. The method according to claim 10, wherein the model is derived using a support vector machine (SVM) process, a random forest process, or a gradient boosting process.

12. The method according to claim 1, wherein the characteristic data includes age, alanine aminotransferase (ALT) activity, breed, sex, cTnI, cough, exercise tolerance, heart rate, heart rhythm, respiratory rate, albumin concentration, alkaline phosphatase (ALKP) concentration, bilirubin concentration, blood urea nitrogen (BUN) concentration, calcium concentration, cholesterol concentration, gamma-glutamyltransferase (GGT) concentration, globulin concentration, glucose concentration, phosphate concentration, potassium concentration, symmetric dimethylarginine (SDMA) concentration, and sodium concentration.

13. A computer program, when executed by a computer system, includes code means for instructing a computer system to carry out the method described in any one of claims 1 to 12.

14. A system for diagnosing stage B2 DMVD in dogs, An input device configured to receive characteristic data associated with a dog, wherein the characteristic data includes NT-proBNP concentration, appetite, creatinine concentration, heart murmur intensity, and body condition score (BCS); A model configured to receive characteristic data and generate output values ​​associated with the probability that a dog has a stage B2 DMVD; and Output device configured to output an output value A system that includes this.

Images