Diagnosis and classification of portosystemic shunts: a machine learning retrospective case-control study

Makan Farhoodimoghadam, Krystle L. Reagan^,, Allison L. Zwingenberger

Background

This study aims to enhance the diagnosis of portosystemic shunts (PSS) in dogs using machine learning models (MLMs). PSS are vascular anomalies that allow blood to bypass the liver, leading to significant clinical complications. Current diagnostic methods have limitations in sensitivity and specificity, prompting the exploration of MLMs that utilize demographic data and clinicopathologic features.

Methods

A retrospective case-control study was conducted using records from the University of California-Davis Veterinary Medical Teaching Hospital. Dogs diagnosed with PSS or tested but ruled out (non-PSS) from 2000 to 2020 were included if they had a complete blood count and serum chemistry panel. Two MLMs were trained: one for detecting PSS presence (PSS MLM) and one for determining PSS subcategories (PSS SubCat MLM). Data splitting involved 70% for training and 30% for testing.

Results

The PSS MLM achieved a sensitivity of 94.3% and specificity of 90.5%. It showed high accuracy in diagnosing PSS, particularly in differentiating PSS subtypes with varying sensitivity and specificity. The PSS SubCat MLM had a general accuracy of 85.7% in predicting subtypes, with significant variability based on the specific PSS subtype.

Limitations

The models were trained on a dataset where all dogs had a high suspicion of PSS, limiting generalizability to other populations. Additionally, there was a lack of a definitive diagnosis in many non-PSS dogs, introducing potential biases in control data. The study was based on records from a single institution, which may limit the applicability of the findings to other settings.

Conclusions

The study demonstrated that MLMs could effectively predict the presence and subtype of PSS in dogs using routinely collected data. These models provide a significant improvement over traditional diagnostic methods, offering a reliable, non-invasive diagnostic tool that could enhance clinical decision-making and treatment planning for dogs suspected of having PSS.

Disclaimer: The summary generated in this email was created by an AI large language model. Therefore errors may occur. Reading the article is the best way to understand the scholarly work. The figure presented here remains the property of the publisher or author and subject to the applicable copyright agreement. It is reproduced here as an educational work. If you have any questions or concerns about the work presented here, reply to this email.