Related Experiment Video
Updated: Nov 4, 2025
11:03
Use of Capillary Electrophoresis Immunoassay to Search for Potential Biomarkers of Amyotrophic Lateral Sclerosis in Human Platelets
Published on: February 10, 2020
7.4K
Deciphered coagulation profile to diagnose the antiphospholipid syndrome using artificial intelligence
Romy M W de Laat-Kremers1, Denis Wahl2, Stéphane Zuily2
1Department of Data Analysis and Artificial Intelligence, Synapse Research Institute, Maastricht, the Netherlands; Department of Functional Coagulation, Synapse Research Institute, Maastricht, the Netherlands.
Thrombosis Research
|May 22, 2021
Summary
A new neural network accurately diagnoses antiphospholipid syndrome (APS) using thrombin generation data. This AI tool shows high sensitivity and predictive values, offering a potential new method for APS diagnosis.
Area of Science:
- Hematology
- Artificial Intelligence
- Medical Diagnostics
Background:
- Antiphospholipid syndrome (APS) diagnosis relies on specific antibodies and clinical events like thrombosis or pregnancy morbidity.
- Coagulation profiles in APS patients differ significantly from healthy individuals due to accelerated thrombin production and impaired activated protein C pathway.
- Current diagnostic methods for APS are not solely reliant on single test results.
Purpose of the Study:
- To develop and clinically validate a neural network for diagnosing antiphospholipid syndrome (APS).
- To assess the diagnostic accuracy of the neural network using thrombin generation data in a diverse patient cohort.
Main Methods:
- Development of a neural network model trained on patient data to identify APS.
- Clinical validation of the neural network in a separate cohort including APS patients, normal controls, hospital controls, and disease-specific control groups (thrombosis and autoimmune diseases).
Main Results:
- The neural network demonstrated high diagnostic accuracy across various control groups.
- Positive predictive values ranged from 62% to 91%, and negative predictive values ranged from 86% to 95%.
- The neural network achieved a sensitivity exceeding 90% in all tested control groups.
Conclusions:
- A neural network utilizing thrombin generation data can accurately diagnose APS.
- Further clinical validation in newly diagnosed patients is recommended before potential clinical implementation.
- This AI-driven approach may offer a novel diagnostic tool for antiphospholipid syndrome.
