A Machine Learning PROGRAM to Identify COVID-19 and Other Diseases From Hematology Data

Abstract

Aim: We propose a method for screening full blood count metadata for evidence of communicable and noncommunicable diseases using machine learning (ML). Materials & methods: High dimensional hematology metadata was extracted over an 11-month period from Sysmex hematology analyzers from 43,761 patients. Predictive models for age, sex and individuality were developed to demonstrate the personalized nature of hematology data. Both numeric and raw flow cytometry data were used for both supervised and unsupervised ML to predict the presence of pneumonia, urinary tract infection and COVID-19. Heart failure was used as an objective to prove method generalizability. Results: Chronological age was predicted by a deep neural network with R2: 0.59; mean absolute error: 12; sex with AUROC: 0.83, phi: 0.47; individuality with 99.7% accuracy, phi: 0.97; pneumonia with AUROC: 0.74, sensitivity 58%, specificity 79%, 95% CI: 0.73–0.75, p < 0.0001; urinary tract infection AUROC: 0.68, sensitivity 52%, specificity 79%, 95% CI: 0.67–0.68, p < 0.0001; COVID-19 AUROC: 0.8, sensitivity 82%, specificity 75%, 95% CI: 0.79–0.8, p = 0.0006; and heart failure area under the receiver operator curve (AUROC): 0.78, sensitivity 72%, specificity 72%, 95% CI: 0.77–0.78; p < 0.0001. Conclusion: ML applied to hematology data could predict communicable and noncommunicable diseases, both at local and global levels.

Digital assessment of peripheral blood and bone marrow aspirate smears

“…Zhu et al used similar methods to improve upon the performance of the MDS‐CBC score for prediction of MDS 46 . By integrating high‐dimensional hematology data with patient clinical data, Gladding et al developed accurate machine learning models for the prediction of infectious diseases including pneumonia, urinary tract infection, and COVID‐19 47 . Finally, a significant amount of research has focused on the identification of novel PB cell parameters with improved predictive value over parameters currently provided by hematology analyzers 48 .…”

“…46 By integrating highdimensional hematology data with patient clinical data, Gladding et al developed accurate machine learning models for the prediction of infectious diseases including pneumonia, urinary tract infection, and COVID-19. 47 Finally, a significant amount of research has focused on the identification of novel PB cell parameters with improved predictive value over parameters currently provided by hematology analyzers. 48 While many of these parameters are known features of PB cells, unsupervised machine learning-based feature extraction methods may identify additional parameters that can be integrated with hematology analyzer outputs into accurate diagnostic models.…”

Macrophage Boolean networks in the time of SARS-CoV-2

“…Macrophage phenotypes tilted to a more pro-inflammatory phenotype as severity increased ( 51 ). Finally, but not less important, some ML approaches have been used to differentiate flow cytometry profiles of blood samples from positive COVID-19 patients with respect to other diseases (like pneumonia) ( 52 ). On the other hand, the second factor that completes the global outlook is given by advanced techniques to gather biological data, and use them to build high-curated gene regulatory networks (GRN) in macrophages ( 32 , 53 ).…”

Covid19predictor: Kli̇ni̇k Veri̇lere Ve Ruti̇n Testlere Dayali Olarak Covid-19 Teşhi̇si̇ İçi̇n Maki̇ne Öğreni̇mi̇ Modelleri̇ Geli̇şti̇rmeye Yarayan Web Tabanli Arayüz

“…Various papers discussed machine learning to diagnose Covid-19 on different data sources, methods, performance criteria and reporting [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]. Systematic reviews emphasized the increasing use of machine learning for Covid-19 diagnosis [22][23][24].…”

“…The source code also can be used to create pipelines on local servers. Although other machine learning methods exist in literature such as Batista et al, Yang et al, Joshi et al, Kukar et al, Brinati et al, Tordjman et al, Cabitza et al, Gladding et al most of these studies work with a certain dataset and report finding of this particular data[7,[10][11][12]15,[18][19][20]. Thus, user who wants to develop and test their data still needs to create a separate software pipeline where the standard workflow mentioned in these works might not be created easily resulting a reproducibility problem.…”