Related Experiment Video
Updated: Jun 16, 2026
04:17
DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning
Published on: May 10, 2024
839
A multiple instance learning approach for detecting COVID-19 in peripheral blood smears.
Colin L Cooke1, Kanghyun Kim2, Shiqi Xu2
1Electrical and Computer Engineering Department, Duke University, United States of America.
PLOS Digital Health
|February 22, 2023
Summary
This study uses machine learning to analyze blood cell morphology in peripheral blood smears, linking cell changes to COVID-19 infection status. The approach achieved 79% accuracy in diagnosing disease from blood cell images.
Area of Science:
- Hematology
- Computational Biology
- Medical Diagnostics
Background:
- Peripheral blood smear analysis is crucial for diagnosing various diseases.
- The impact of COVID-19 on blood cell morphology remains incompletely understood.
- Automated analysis of blood cell morphology can aid disease diagnosis.
Purpose of the Study:
- To investigate the link between blood cell morphology and COVID-19 status.
- To develop and validate a machine learning approach for disease diagnosis using blood smears.
- To enhance understanding of COVID-19's hematological effects.
Main Methods:
- A multiple instance learning framework was employed.
- High-resolution morphological data from blood cells were aggregated.
- Image and diagnostic data from 236 patients were integrated.
Main Results:
- A significant association between blood morphology and COVID-19 infection status was identified.
- The machine learning model demonstrated high diagnostic efficacy.
- Achieved 79% accuracy and a ROC-AUC of 0.90 for disease diagnosis.
Conclusions:
- Novel machine learning methods offer a scalable solution for analyzing peripheral blood smears.
- The study supports and expands existing hematological findings on COVID-19.
- Automated morphological analysis of blood cells shows promise for disease detection.
