AUTHOR=Zhang Aixian , Lou Jingjiao , Pan Zijie , Luo Jiaqi , Zhang Xiaomeng , Zhang Han , Li Jianpeng , Wang Lili , Cui Xiang , Ji Bing , Chen Li 

TITLE=Prediction of anemia using facial images and deep learning technology in the emergency department

JOURNAL=Frontiers in Public Health

VOLUME=Volume 10 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2022.964385

DOI=10.3389/fpubh.2022.964385

ISSN=2296-2565

ABSTRACT=Background: Anemia is a highly prevalent disease according to the WHO, especially for patients in the Emergency Department. The pathophysiological mechanism that anemia can affect face characteristics such as membrane pallor has been proven and used to detect anemia with the help of updating deep learning technology. Quick prediction method for the patient in Emergency Department is important to screen anemic state and even further judge the necessity of blood transfusion treatment.
Method: We trained a deep learning system for the prediction from anemia using videos of 316 inclusion patients. All the videos were taken by the same portable pad in the ambient environment of Emergency Department. Video extraction method, face recognition method were used to highlight the facial area for analysis. Accuracy, area under the curve were used to assess the performance of the machine learning system in the image level and the patient level.
Results: Three tasks were applied for performance evaluation. The objective of task 1 was to predict the anemia of patients (Hemoglobin (Hb)<13g/dL in men and Hb<12g/dL in women). The accuracy of the image level was 82.37%, the area under the curve (AUC) of the image level was 0.84 and the accuracy of the patient level was 84.02%, the sensitivity of the patient level was 92.59%, specificity of the patient level was 69.23%. The objective of task 2 was to predict the mild anemia (Hb<9g/dL). The accuracy of the image level was 68.37%, the AUC of the image level was 0.69 and the accuracy of the patient level was 70.58%, sensitivity was 73.52%, and specificity was 67.64%. Aim of task-3 was to predict severe anemia (Hb<7g/dL). The accuracy of the image level was 74.01%, the AUC of the image level was 0.82 and the accuracy of the patient level was 68.42%, sensitivity was 61.53%, specificity was 83.33%.
Conclusion: The machine learning system could fast and accurately predict anemia of patients in the Emergency Department and aid the decision of treatment of emergent blood transfusion. It offers great clinical value and practical significance in improving fast diagnosis, medical resource allocation, and appropriate treatment in the future.