Machine-learning based prediction of Cushing's syndrome in dogs attending UK primary-care veterinary practice

External validation of a United Kingdom primary-care Cushing's prediction tool in a population of referred Dutch dogs

BACKGROUND

A prediction tool was developed and internally validated to aid the diagnosis of Cushing's syndrome in dogs attending UK primary-care practices. External validation is an important part of model validation to assess model performance when used in different populations.

OBJECTIVES

To assess the original prediction model's transportability, applicability, and diagnostic performance in a secondary-care practice in the Netherlands.

ANIMALS

Two hundred thirty client-owned dogs.

METHODS

Retrospective observational study. Medical records of dogs under investigation of Cushing's syndrome between 2011 and 2020 were reviewed. Dogs diagnosed with Cushing's syndrome by the attending internists and fulfilling ALIVE criteria were defined as cases, others as non-cases. All dogs were scored using the aforementioned prediction tool. Dog characteristics and predictor-outcome effects in development and validation data sets were compared to assess model transportability. Calibration and discrimination were examined to assess model performance.

RESULTS

Eighty of 230 dogs were defined as cases. Significant differences in dog characteristics were found between UK primary-care and Dutch secondary-care populations. Not all predictors from the original model were confirmed to be significant predictors in the validation sample. The model systematically overestimated the probability of having Cushing's syndrome (a = -1.10, P < .001). Calibration slope was 1.35 and discrimination proved excellent (area under the receiver operating curve = 0.83).

CONCLUSIONS AND CLINICAL IMPORTANCE

The prediction model had moderate transportability, excellent discriminatory ability, and overall overestimated probability of having Cushing's syndrome. This study confirms its utility, though emphasizes that ongoing validation efforts of disease prediction tools are a worthwhile effort.

Machine learning-based risk prediction model for canine myxomatous mitral valve disease using electronic health record data

Introduction

Myxomatous mitral valve disease (MMVD) is the most common cause of heart failure in dogs, and assessing the risk of heart failure in dogs with MMVD is often challenging. Machine learning applied to electronic health records (EHRs) is an effective tool for predicting prognosis in the medical field. This study aimed to develop machine learning-based heart failure risk prediction models for dogs with MMVD using a dataset of EHRs.

Methods

A total of 143 dogs with MMVD between May 2018 and May 2022. Complete medical records were reviewed for all patients. Demographic data, radiographic measurements, echocardiographic values, and laboratory results were obtained from the clinical database. Four machine-learning algorithms (random forest, K-nearest neighbors, naïve Bayes, support vector machine) were used to develop risk prediction models. Model performance was represented by plotting the receiver operating characteristic (ROC) curve and calculating the area under the curve (AUC). The best-performing model was chosen for the feature-ranking process.

Results

The random forest model showed superior performance to the other models (AUC = 0.88), while the performance of the K-nearest neighbors model showed the lowest performance (AUC = 0.69). The top three models showed excellent performance (AUC ≥ 0.8). According to the random forest algorithm's feature ranking, echocardiographic and radiographic variables had the highest predictive values for heart failure, followed by packed cell volume (PCV) and respiratory rates. Among the electrolyte variables, chloride had the highest predictive value for heart failure.

Discussion

These machine-learning models will enable clinicians to support decision-making in estimating the prognosis of patients with MMVD.

Predicting health outcomes in dogs using insurance claims data

In this paper we propose a machine learning-based approach to predict a multitude of insurance claim categories related to canine diseases. We introduce several machine learning approaches that are evaluated on a pet insurance dataset consisting of 785,565 dogs from the US and Canada whose insurance claims have been recorded over 17 years. 270,203 dogs with a long insurance tenure were used to train a model while the inference is applicable to all dogs in the dataset. Through this analysis we demonstrate that with this richness of data, supported by the right feature engineering, and machine learning approaches, 45 disease categories can be predicted with high accuracy.

Non-destructive diagnostic testing of cardiac myxoma by serum confocal Raman microspectroscopy combined with multivariate analysis

The symptoms of cardiac myxoma (CM) mainly occur when the tumor is growing, and the diagnosis is determined by clinical presentation. Unfortunately, there is no evidence that specific blood tests are useful in CM diagnosis. Raman spectroscopy (RS) has emerged as a promising auxiliary diagnostic tool because of its ability to simultaneously detect multiple molecular features without labelling. The objective of this study was to identify spectral markers for CM, one of the most common benign cardiac tumors with insidious onset and rapid progression. In this study, a preliminary analysis was conducted based on serum Raman spectra to obtain the spectral differences between CM patients (CM group) and healthy control subjects (normal group). Principal component analysis-linear discriminant analysis (PCA-LDA) was constructed to highlight the differences in the distribution of biochemical components among the groups according to the obtained spectral information. Principal component analysis was combined with a support vector machine model (PCA-SVM) based on three different kernel functions (linear, polynomial, and Gaussian radial basis function (RBF)) to resolve spectral variations between all study groups. The results showed that CM patients had lower serum levels of phenylalanine and carotenoid than those in the normal group, and increased levels of fatty acids. The resulting Raman data was used in a multivariate analysis to determine the Raman range that could be used for CM diagnosis. Also, the chemical interpretation of the spectral results obtained is further presented in the discussion section based on the multivariate curve resolution-alternating least squares (MCR-ALS) method. These results suggest that RS can be used as an adjunct and promising tool for CM diagnosis, and that vibrations in the fingerprint region can be used as spectral markers for the disease under study.

Diagnostic Classification of Cases of Canine Leishmaniasis Using Machine Learning

Proposal techniques that reduce financial costs in the diagnosis and treatment of animal diseases are welcome. This work uses some machine learning techniques to classify whether or not cases of canine visceral leishmaniasis are present by physical examinations. For validation of the method, four machine learning models were chosen: K-nearest neighbor, Naïve Bayes, support vector machine and logistic regression models. The tests were performed on three hundred and forty dogs, using eighteen characteristics of the animal and the ELISA (enzyme-linked immunosorbent assay) serological test as validation. Logistic regression achieved the best metrics: Accuracy of 75%, sensitivity of 84%, specificity of 67%, a positive likelihood ratio of 2.53 and a negative likelihood ratio of 0.23, showing a positive relationship in the evaluation between the true positives and rejecting the cases of false negatives.

Metabolomic Abnormalities in Serum from Untreated and Treated Dogs with Hyper- and Hypoadrenocorticism

The adrenal glands play a major role in metabolic processes, and both excess and insufficient serum cortisol concentrations can lead to serious metabolic consequences. Hyper- and hypoadrenocorticism represent a diagnostic and therapeutic challenge. Serum samples from dogs with untreated hyperadrenocorticism (n = 27), hyperadrenocorticism undergoing treatment (n = 28), as well as with untreated (n = 35) and treated hypoadrenocorticism (n = 23) were analyzed and compared to apparently healthy dogs (n = 40). A validated targeted proton nuclear magnetic resonance (¹H NMR) platform was used to quantify 123 parameters. Principal component analysis separated the untreated endocrinopathies. The serum samples of dogs with untreated endocrinopathies showed various metabolic abnormalities with often contrasting results particularly in serum concentrations of fatty acids, and high- and low-density lipoproteins and their constituents, which were predominantly increased in hyperadrenocorticism and decreased in hypoadrenocorticism, while amino acid concentrations changed in various directions. Many observed serum metabolic abnormalities tended to normalize with medical treatment, but normalization was incomplete when compared to levels in apparently healthy dogs. Application of machine learning models based on the metabolomics data showed good classification, with misclassifications primarily observed in treated groups. Characterization of metabolic changes enhances our understanding of these endocrinopathies. Further assessment of the recognized incomplete reversal of metabolic alterations during medical treatment may improve disease management.