Skip to content
Publicly Available Published by De Gruyter April 25, 2022

Artificial intelligence at the time of COVID-19: who does the lion’s share?

  • Davide Negrini ORCID logo , Elisa Danese , Brandon M. Henry , Giuseppe Lippi ORCID logo and Martina Montagnana EMAIL logo

Abstract

Objectives

The development and use of artificial intelligence (AI) methodologies, especially machine learning (ML) and deep learning (DL), have been considerably fostered during the ongoing coronavirus disease 2019 (COVID-19) pandemic. Several models and algorithms have been developed and applied for both identifying COVID-19 cases and for assessing and predicting the risk of developing unfavourable outcomes. Our aim was to summarize how AI is being currently applied to COVID-19.

Methods

We conducted a PubMed search using as query MeSH major terms “Artificial Intelligence” AND “COVID-19”, searching for articles published until December 31, 2021, which explored the possible role of AI in COVID-19. The dataset origin (internal dataset or public datasets available online) and data used for training and testing the proposed ML/DL model(s) were retrieved.

Results

Our analysis finally identified 292 articles in PubMed. These studies displayed large heterogeneity in terms of imaging test, laboratory parameters and clinical-demographic data included. Most models were based on imaging data, in particular CT scans or chest X-rays images. C-Reactive protein, leukocyte count, creatinine, lactate dehydrogenase, lymphocytes and platelets counts were found to be the laboratory biomarkers most frequently included in COVID-19 related AI models.

Conclusions

The lion’s share of AI applied to COVID-19 seems to be played by diagnostic imaging. However, AI in laboratory medicine is also gaining momentum, especially with digital tools characterized by low cost and widespread applicability.

Introduction

The rapid and almost unremitting worldwide spread of the coronavirus disease (COVID-19) pandemic has not only dramatically impacted daily life and habits, but has also been a catalyst for many research fields, thus resulting in development, validation and rapid implementation of multiple novel diagnostic and therapeutic approaches. One of the most relevant examples was the acceleration of all aspects of vaccine development, including vaccine formulation, preclinical testing, clinical trials, manufacturing, and approval according to current regulatory guidelines and legal requirements. Accordingly, for the first time in the history of medicine, it took only 11 months between identification of the viral genetic sequence and the first doses of vaccine being administered to the population [1].

The term Artificial Intelligence (AI), defined as the “theory and development of computer systems which perform tasks that normally require human intelligence” [2], was originally used by John McCarthy in 1956 during a conference on this topic [3]. Although diagnostics skills are still considered a consequence of patient-physician direct relation, with foundation on experience and training [4], a constant increase in the use of Clinical Decision Support Systems (CDSS) has been observed in the past few years, with multiplication of integrations in electronic health records [5], leveraging both on literature information but also on AI, Machine Learning (ML), Deep Learning (DL) or other more traditional statistical pattern-recognition methods. This computer-based technology has evolved and is being implemented in many areas of medicine.

ML models can be categorized in supervised and unsupervised [6]. In supervised models, labelled learning datasets to train the model are used, with labels assigned with reference or consensus methods. Therefore the model can figure out which label to assign to new unlabelled data based on previously labelled data. In unsupervised learning, datasets are not labelled, so the model itself tries to classify groups of data without supervision. The most used approach in medical research is the supervised one [7] and the models for classification are usually Convolutional Neural Networks (CNN), Random Forest (RF), Gradient Boosting (XGBoost), Support Vector Machine (SVM), Logistic Regression (LR) and k-Nearest Neighbours (KNN).

AI and CDSS were not an exception to the rapid scientific advancement during the COVID-19 pandemic, with an increase in their use for many tasks, both medical and industrial. Several medical specialties have hence developed and applied AI innovative approaches to try achieving some particular common goals such as rapid diagnosis, decision making and effective patient monitoring during treatment for COVID-19 [8], [9], [10], [11]. The use of AI was made possible by collaborations between different countries and thanks to availability of many data (i.e., clinical-demographical, imaging and laboratory information) from different COVID-19 patient populations. By analysing all these data collected in internal or public datasets, several models and algorithms to be used for decision making have been developed.

Therefore, with the aim of investigating how AI has been applied and has impacted clinical decision making during the COVID-19 pandemic, we carried out an electronic search for retrieving PubMed-indexed articles containing information on how, and with which methods, worldwide scientists tried to implement new AI/ML technologies in the COVID-19 arena.

Methods

We carried out a PubMed search using as query MeSH major terms “Artificial Intelligence” [2] AND “COVID-19” [12], for detecting articles published until December 31, 2021. We decided to use the term “Artificial Intelligence” because in the MeSH database it includes “Expert Systems”, “Machine Learning” and “Deep Learning”. We analysed all articles that used a AI/ML/DL approach for identification, diagnosis, prognosis and/or risk stratification (e.g., risk of mechanical ventilation, of hospitalisation, of death) of SARS-CoV-2 infection. We excluded all articles dealing with: (a) epidemiology or public health forecasting; (b) facial detection algorithms; (c) psychology or patient communication or sentiment analysis or social medias implications or usage; (d) genetics; (e) ontology or taxonomy; (f) laboratory techniques; (g) radiology techniques; (h) robotics or tele-monitoring instruments or applications; (i) pharmacology studies; (j) corrections to previously published articles; (k) opinion papers, editorials, commentaries, single case reports; (l) reviews; and (m) other (non-SARS-CoV-2-related, non-AI/ML/DL, others not classifiable articles).

In all included articles, COVID-19 positivity or negativity was based on result of molecular tests for detecting severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) performed by means of nucleic acid amplification test (NAAT) or reverse transcription polymerase chain reaction (RT/PCR) on nasopharyngeal swabs. For each article, the following aspects were retrieved: dataset origin (internal dataset or public datasets available online), data used for training and testing the proposed ML model/models, in particular (i) inclusion of clinical-demographic data (and which); (ii) inclusion of laboratory test results (and which); (iii) inclusion of imaging test results (computed tomography [CT], chest X-ray [CXR], ultrasonography [US], combination of CT and CXR); as well as (iv) inclusion of electrocardiography (ECG) data. When reported in the article, the ML/DL model (or models) and the resulting performance (area under ROC curve [AUC] and/or the F1-score) of the proposed models were also retrieved.

Results

Our PubMed search generated 678 document with an assigned and valid PubMed ID. Of them, 386 were excluded and 292 were included in the analysis (Figure 1). The use of different information categories or test-types and dataset origin are reported in Table 1. In 67.8% (198/292) of articles, only imaging information was included in the models. Of those studies using imaging data, the type of imaging included in ML/DL models is shown in Table 2.

Figure 1: 
Search and review procedure.
Figure 1:

Search and review procedure.

Table 1:

Types of tests used in the included articles, divided by dataset origin.

Internal dataset (or not declared) n=198 Public dataset n=94
Only imaging (n=198) 113 85
Laboratory tests + clinical-demographic features (n=37) 33 4
Only clinical-demographic features (n=18) 13 5
Only laboratory tests (n=15) 15 0
Laboratory tests + imaging + clinical-demographic features (n=14) 14 0
Laboratory tests + imaging (n=5) 5 0
Only ECG (n=1) 1 0
Imaging + clinical-demographic features (n=1) 1 0
Clinical-demographic features + ECG (n=1) 1 0
Laboratory tests + clinical-demographic features + ECG (n=1) 1 0
Laboratory tests + imaging + clinical-demographic features + ECG (n=1) 1 0
Table 2:

Types of imaging tests, divided by dataset origin, used in the models.

Internal dataset (or not declared) n=85 Public datasets n=134
CT (n=101) 29 72
CXR (n=99) 48 51
CXR + CT (n=14) 7 7
US (n=5) 1 4

Concerning laboratory tests, the more common tests included in the different models are summarized in Table 3, showing high prevalence of C-reactive protein, leukocyte count, creatinine, lactate dehydrogenase, lymphocytes and platelets counts. The most included demographic features are instead summarized in Table 4, whilst the top-10 most included clinical parameters are listed in Table 5. Finally, the most used models/algorithms are reviewed in Table 6.

Table 3:

Top-20 laboratory tests used in the models.

Laboratory test
C-reactive protein (CRP) of high-sensitivity-CRP (hs-CRP) 47
White blood cells count 39
Creatinine 31
Lactate dehydrogenase 31
Lymphocytes count 30
Platelets count 30
Haemoglobin 26
D-dimer 25
Blood urea nitrogen (BUN) 25
Neutrophils count 23
Bilirubin total 21
Ferritin 21
Alanine aminotransferase (ALT) 20
Aspartate aminotransferase (AST) 20
Sodium 20
Albumin 16
Glucose 16
Lymphocytes % 16
Calcium 15
Bicarbonate 15
Table 4:

Most included demographic features in the models.

Demographic features
Age 57
Gender 47
Ethnicity 11
Geographical location 4
Table 5:

Top-10 most included clinical parameters in the models.

Clinical parameter
Medical history (various parameters) 30
Body temperature/Fever 24
Respiratory rate 19
Symptoms (various parameters) 19
Heart rate 15
Diastolic blood pressure 13
Systolic blood pressure 13
Body mass index (BMI) 12
Height 6
Smoking history 6
Table 6:

Most (top-25) used models/algorithms in the included articles.

Model/algorithm
Convoluted neural network (unspecified) 36
Random forest 29
ResNet-50 25
XGBoost 24
Support-vector machine 21
Logistic regression 18
VGG-16 17
VGG-19 17
Inception-v3 16
k-nearest neighbour 9
U-Net 9
DenseNet-201 7
DenseNet-121 6
EfficientNet 6
ResNet-101 6
ResNet-18 6
Xception 6
ADAboost 5
AlexNet 5
Inception-ResNet-v2 5
Long short-term memory 5
MobileNet 5
Multi-layer perceptron 5
Naive Bayes 5
ResNet-50v2 5

As concerns diagnostic performance, the area under the curve (AUC) of the different models identified by our PubMed search ranged between 0.470 and 0.999 (mean 0.859, median 0.878), with F1-scores ranging between 0.420 and 1.000 (mean 0.891, median 0.939).

Discussion

After more than two years of the COVID-19 pandemic, the importance of optimizing the use of health resources has been dramatically emphasized. The diagnosis of COVID-19 is still profoundly based on viral RNA detection (by means of molecular testing on nasopharyngeal swabs), although preanalytical and analytical issues such as correct patient and sample identification, the use of appropriate procedures for collection, handling, transport and storage of the specimens, presence of interfering materials, clerical errors and RNA contamination may impair the diagnostic accuracy of this technique [13]. Moreover, additional issues such as organization, the need to maintain a rapid turn-around time (TAT) and the high costs have made it difficult to use this diagnostic tool in all healthcare settings [14].

In order to contain the spread of SARS-CoV-2 and improve diagnostic capabilities, many researchers were engaged to develop predictive models for rapid and accurate identification of patients with SARS-CoV-2 infection and/or higher risk of unfavourable disease progression. To this end, the first important aspect that emerged from our electronic PubMed search is that the number of articles in the field of AI applied to COVID-19 has grown exponentially, with more than 600 articles published in only two years (i.e., up to the end of 2021). We also found that another boost during the pandemic was the ability to exchange data between researchers from different groups and countries. In facts, about a third of articles included in our analysis used public databases for AI model development and validation. This is a very important aspect since both quality and size of the datasets used are crucial elements to enable the generation and validation of algorithms [15].

We observed that the number of recruited subjects is highly variable, ranging from a few tens to thousands of subjects included in different studies. The size of the sample influences the choice of algorithm, accuracy of model and validation procedure. The description of the population selected for construction and validation of the model is also very important: some studies were in fact conducted only on Emergency Department patients, others on hospitalized patients and/or outpatients. In some cases the origin of the patients was not even specified. Finally, another noteworthy element concerns the proportion of COVID-19-positive individuals in the population studied.

Most of such models proposed for COVID-19 are based on imaging data, in particular CT scans or chest X-rays images. This is in keeping with the fact that, until recently, the field of predominant interest and greatest application of AI was precisely “computer vision” applications, especially to assist radiologists in classifying lesions or identifying patterns [16]. These models use DL and convoluted neural networks (CNN) which codes and usage can be easily implemented in AI-pipelines [17]. Accordingly, as showed in Table 6, in our analysis the most represented models used CNN.

AI-based imaging models have been developed both to early diagnose the disease [1821], to predict the danger of deterioration in COVID-19 patients [22] and finally to predict outcomes [23]. Notably, this seems at odds with current indications that CT should not be used as part of the initial routine clinical assessment of COVID-19, as endorsed by the American College of Emergency Physicians (ACEP) [24]. As concerns the diagnostic performances of the different models, a huge heterogeneity has been observed among different imaging-based models and algorithms, but also when the same model has been used in different setting [25].

Unlike what might have been expected, the use of AI applied to laboratory medicine has not grown exponentially, as has been the case for other disciplines, such as radiology [26, 27]. As reported in the articles of Cabitza et al. [26] and Ronzio et al. [27], 37 studies of ML in laboratory medicine were published between 2007 and 2017, and other 44 studies between 2017 and 2020. Represented by continuous variables (i.e., numbers), results of laboratory tests may appear the easiest and fastest tools to be included within AI models. However, both preanalytical factors (i.e., biological variation) and use of different analytic methods and/or instrument, may produce rapid changes in laboratory parameters, thus making more difficult to apply models based only on laboratory tests. To the best of our knowledge, there are no studies that have evaluated the impact and importance of pre-analytical variables on the AI models accuracy, with the only exception of that conducted by Jones et al. in the histopathology field [28].

Importantly, we found that laboratory data were combined with clinical/demographic or imaging data to improve AI models accuracy in most of the articles considered in our analysis. However, a very important aspect that should be taken into consideration is that many factors can influence the representativeness of datasets for assessing the robustness of a model, especially when a mix of data (imaging, laboratory data, clinical data) are used. Accordingly, differences in testing equipment and analytic procedures, lack of specific reference ranges applied in ethnically heterogeneous populations and large variability in disease manifestations make the reference population extremely vast and variegated. Consequently, all these aspects may impair the reproducibility of results obtained in the model validation phase.

Among most included laboratory parameters (Table 3), we predictably found markers of inflammation (in particular C-reactive protein and white blood cells count), along with other parameters recommended by the Task Force on COVID-19 of the International Federation of Clinical Chemistry (IFCC) and Laboratory Medicine [29]. The advantage is that all these laboratory parameters are available as emergency-urgency tests in the vast majority of hospitals, such that these algorithms would be easily applicable almost everywhere.

The role of laboratory professional should be central in the selection of laboratory tests to be included in ML models. There are, in fact, some critical elements that should be considered, as well as relation between variables/laboratory parameters and COVID-19 pathology, the characteristics of the diagnostic test, the biological variability of the laboratory parameters, the relative importance of single test in a series of assays. The laboratory expert, who knows all these aspects, can contribute both in the choice of the laboratory parameters to be included in the model and in the correct interpretation of the results in the clinical context.

We certainly acknowledge that our electronic search cannot be considered universally comprehensive of all articles containing information on AI in COVID-19, since we limited our search to PubMed, up to the end of the year 2021. Moreover, in our analysis the studies were not classified according to their purpose (i.e., retrospective or prospective/prognostic), since we aimed to explore the contribution of different medical disciplines in the context of artificial intelligence within the COVID-19 emergency. Nonetheless, we still believe that the synthesis that we have provided in this work may provide a rather reliable reflection of the potentiality of AI on COVID-19, as well as on the main characteristics of the different models that have been generated and used in clinical practice.

In conclusion, the potential of AI in the global health emergency has become clearly evident during the past two years of the COVID-19 pandemic [8], [9], [10], [11]. Cumulatively, the lion’s share (i.e., the major contribution) of AI applied to COVID-19 seems to be played by diagnostic imaging, although laboratory medicine is also gaining momentum, especially with digital tools characterized by low cost and widespread applicability to all healthcare contexts.


Corresponding author: Prof. Martina Montagnana, Section of Clinical Biochemistry and School of Medicine, University Hospital of Verona, Piazzale L.A. Scuro, 10, 37134 Verona, Italy, Phone: 0039 045 8122970, Fax: 0039 045 8124308, E-mail:

  1. Research funding: None declared.

  2. Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.

  3. Competing interests: Authors state no conflict of interest.

  4. Informed consent: Not applicable.

  5. Ethical approval: Not applicable.

References

1. Kuter, BJ, Offit, PA, Poland, GA. The development of COVID-19 vaccines in the United States: why and how so fast? Vaccine 2021;39:2491–5. https://doi.org/10.1016/j.vaccine.2021.03.077.Search in Google Scholar PubMed PubMed Central

2. PubMed artificial intelligence MeSH term definition. Available from: https://www.ncbi.nlm.nih.gov/mesh/68001185.Search in Google Scholar

3. McCarthy, J, Minsky, ML, Rochester, N, Shannon, CE. A proposal for the Dartmouth summer research project on artificial intelligence. AI Mag 2006;27:12–4.Search in Google Scholar

4. Lippi, G. Machine learning in laboratory diagnostics: valuable resources or a big hoax? Diagnosis 2019;8:133–5. https://doi.org/10.1515/dx-2019-0060.Search in Google Scholar PubMed

5. Sutton, RT, Pincock, D, Baumgart, DC, Sadowski, DC, Fedorak, RN, Kroeker, KI. An overview of clinical decision support systems: benefits, risks, and strategies for success. NPJ Digit Med 2020;3:17. https://doi.org/10.1038/s41746-020-0221-y.Search in Google Scholar PubMed PubMed Central

6. Maleki, F, Ovens, K, Najafian, K, Forghani, B, Reinhold, C, Forghani, R. Overview of machine learning part 1: fundamentals and classic approaches. Neuroimaging Clin 2020;30:e17–32. https://doi.org/10.1016/j.nic.2020.08.007.Search in Google Scholar PubMed

7. Topol, EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med 2019;25:44–56. https://doi.org/10.1038/s41591-018-0300-7.Search in Google Scholar PubMed

8. Bragazzi, NL, Dai, H, Damiani, G, Behzadifar, M, Martini, M, Wu, J. How big data and artificial intelligence can help better manage the COVID-19 pandemic. Int J Environ Res Publ Health 2020;17:3176. https://doi.org/10.3390/ijerph17093176.Search in Google Scholar PubMed PubMed Central

9. Naseem, M, Akhund, R, Arshad, H, Ibrahim, MT. Exploring the potential of artificial intelligence and machine learning to combat COVID-19 and existing opportunities for LMIC: a scoping review. J Prim Care Community Health 2020;11:2150132720963634. https://doi.org/10.1177/2150132720963634.Search in Google Scholar PubMed PubMed Central

10. Gudigar, A, Raghavendra, U, Nayak, S, Ooi, CP, Chan, WY, Gangavarapu, MR, et al.. Role of artificial intelligence in COVID-19 detection. Sensors 2021;21:8045. https://doi.org/10.3390/s21238045.Search in Google Scholar PubMed PubMed Central

11. Verde, L, De Pietro, G, Ghoneim, A, Alrashoud, M, Al-Mutib, KN, Sannino, G. Exploring the use of artificial intelligence techniques to detect the presence of coronavirus covid-19 through speech and voice analysis. IEEE Access 2021;9:65750–7. https://doi.org/10.1109/access.2021.3075571.Search in Google Scholar

12. PubMed COVID-19 MeSH term definition. Available from: https://www.ncbi.nlm.nih.gov/mesh/2052179.Search in Google Scholar

13. Lippi, G, Simundic, AM, Plebani, M. Potential preanalytical and analytical vulnerabilities in the laboratory diagnosis of coronavirus disease 2019 (COVID-19). Clin Chem Lab Med 2020;58:1070–6. https://doi.org/10.1515/cclm-2020-0285.Search in Google Scholar PubMed

14. Yang, HS, Hou, Y, Vasovic, LV, Steel, PAD, Chadburn, A, Racine-Brzostek, SE, et al.. Routine laboratory blood tests predict SARS-CoV-2 infection using machine learning. Clin Chem 2020;66:1396–404. https://doi.org/10.1093/clinchem/hvaa200.Search in Google Scholar PubMed PubMed Central

15. Chahar, S, Roy, PK. COVID-19: a comprehensive review of learning models. Arch Comput Methods Eng 2021;18:1–26. https://doi.org/10.1007/s11831-021-09641-3 [Epub ahead of print].Search in Google Scholar PubMed PubMed Central

16. Hosny, A, Parmar, C, Quackenbush, J, Schwartz, LH, Aerts, HJWL. Artificial intelligence in radiology. Nat Rev Cancer 2018;18:500–10. https://doi.org/10.1038/s41568-018-0016-5.Search in Google Scholar PubMed PubMed Central

17. Wang, L, Wang, H, Huang, Y, Yan, B, Chang, Z, Liu, Z, et al.. Trends in the application of deep learning networks in medical image analysis: evolution between 2012 and 2020. Eur J Radiol 2022;146:110069. https://doi.org/10.1016/j.ejrad.2021.110069.Search in Google Scholar PubMed

18. Harmon, SA, Sanford, TH, Xu, S, Turkbey, EB, Roth, H, Xu, Z, et al.. Artificial intelligence for the detection of COVID-19 pneumonia on chest CT using multinational datasets. Nat Commun 2020;11:4080–6. https://doi.org/10.1038/s41467-020-17971-2.Search in Google Scholar PubMed PubMed Central

19. Ko, H, Chung, H, Kang, WS, Kim, KW, Shin, Y, Kang, SJ, et al.. COVID-19 pneumonia diagnosis using a simple 2D deep learning framework with a single chest CT image: model development and validation. J Med Internet Res 2020;22:e19569. https://doi.org/10.2196/19569.Search in Google Scholar PubMed PubMed Central

20. Sakagianni, A, Feretzakis, G, Kalles, D, Koufopoulou, C, Kaldis, V. Setting up an easy-to-use machine learning pipeline for medical decision support: a case study for COVID-19 diagnosis based on deep learning with CT scans. Stud Health Technol Inf 2020;272:13–6. https://doi.org/10.3233/SHTI200481.Search in Google Scholar PubMed

21. Neri, E, Miele, V, Coppola, F, Grassi, R. Use of CT and artificial intelligence in suspected or COVID-19 positive patients: statement of the Italian society of medical and interventional radiology. Radiol Med 2020;125:505–8. https://doi.org/10.1007/s11547-020-01197-9.Search in Google Scholar PubMed PubMed Central

22. Shamout, FE, Shen, Y, Wu, N, Kaku, A, Park, J, Makino, T, et al.. An artificial intelligence system for predicting the deterioration of COVID-19 patients in the emergency department. NPJ Digit Med 2021;4:80. https://doi.org/10.1038/s41746-021-00453-0.Search in Google Scholar PubMed PubMed Central

23. Mushtaq, J, Pennella, R, Lavalle, S, Colarieti, A, Steidler, S, Martinenghi, CMA, et al.. Initial chest radiographs and artificial intelligence (AI) predict clinical outcomes in COVID-19 patients: analysis of 697 Italian patients. Eur Radiol 2021;31:1770–9. https://doi.org/10.1007/s00330-020-07269-8.Search in Google Scholar PubMed PubMed Central

24. American College of Emergency Physicians. ACEP COVID-19 field guide. Available from: https://www.acep.org/corona/covid-19-field-guide/assessment/chest-x-ray-and-ct/ [Accessed 18 Mar 2022].Search in Google Scholar

25. Guo, Y, Zhang, Y, Lyu, T, Prosperi, M, Wang, F, Xu, H, et al.. The application of artificial intelligence and data integration in COVID-19 studies: a scoping review. J Am Med Inf Assoc 2021;28:2050–67. https://doi.org/10.1093/jamia/ocab098.Search in Google Scholar PubMed PubMed Central

26. Cabitza, F, Banfi, G. Machine learning in laboratory medicine: waiting for the flood? Clin Chem Lab Med 2018;56:516–24. https://doi.org/10.1515/cclm-2017-0287.Search in Google Scholar PubMed

27. Ronzio, L, Cabitza, F, Barbaro, A, Banfi, G. Has the flood entered the basement? A systematic literature review about machine learning in laboratory medicine. Diagnostics 2021;11:372. https://doi.org/10.3390/diagnostics11020372.Search in Google Scholar PubMed PubMed Central

28. Jones, AD, Graff, JP, Darrow, M, Borowsky, A, Olson, KA, Gandour-Edwards, R, et al.. Impact of pre-analytical variables on deep learning accuracy in histopathology. Histopathology 2019;75:39–53. https://doi.org/10.1111/his.13844.Search in Google Scholar PubMed PubMed Central

29. Thompson, S, Bohn, MK, Mancini, N, Loh, TP, Wang, CB, Grimmler, M, et al.. IFCC Taskforce on COVID-19. IFCC interim guidelines on biochemical/hematological monitoring of COVID-19 patients. Clin Chem Lab Med 2020;58:2009–16. https://doi.org/10.1515/cclm-2020-1414.Search in Google Scholar PubMed

Received: 2022-03-29
Accepted: 2022-04-13
Published Online: 2022-04-25
Published in Print: 2022-11-25

© 2022 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 29.3.2024 from https://www.degruyter.com/document/doi/10.1515/cclm-2022-0306/html
Scroll to top button