Eline Meijer

Eline Meijer

Auto-generated profile, based on publicly available information

Eline Meijer

Institute for Medical Microbiology, University of Zurich, Zurich, Switzerland

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 October 2023 in Journal of Medical Internet Research
Verified
Journal Article
Open access
MHMaarten Homburg
Maarten Homburg
EMEline Meijer
Eline Meijer
Matthijs S. BerendsMatthijs S. Berends
Matthijs S. Berends
TKThijmen Kupers
Thijmen Kupers
THTim Olde Hartman
Tim Olde Hartman
LPLilian Peters
Lilian Peters
Background: Natural language processing (NLP) models such as bidirectional encoder representations from transformers (BERT) hold promise in revolutionizing disease identification from electronic health records (EHRs) by potentially enhancing efficiency and accuracy. However, their practical application in practice settings demands a comprehensive and multidisciplinary approach to development and validation. The COVID-19 pandemic highlighted challenges in disease identification due to limited testing availability and challenges in handling unstructured data. In the Netherlands, where general practitioners (GPs) serve as the first point of contact for health care, EHRs generated by these primary care providers contain a wealth of potentially valuable information. Nonetheless, the unstructured nature of free-text entries in EHRs poses challenges in identifying trends, detecting disease outbreaks, or accurately pinpointing COVID-19 cases. Objective: This study aims to develop and validate a BERT model for detecting COVID-19 consultations in general practice EHRs in the Netherlands. Methods: The BERT model was initially pretrained on Dutch language data and fine-tuned using a comprehensive EHR data set comprising confirmed COVID-19 GP consultations and non–COVID-19–related consultations. The data set was partitioned into a training and development set, and the model’s performance was evaluated on an independent test set that served as the primary measure of its effectiveness in COVID-19 detection. To validate the final model, its performance was assessed through 3 approaches. First, external validation was applied on an EHR data set from a different geographic region in the Netherlands. Second, validation was conducted using results of polymerase chain reaction (PCR) test data obtained from municipal health services. Lastly, correlation between predicted outcomes and COVID-19–related hospitalizations in the Netherlands was assessed, encompassing the period around the outbreak of the pandemic in the Netherlands, that is, the period before widespread testing. Results: The model development used 300,359 GP consultations. We developed a highly accurate model for COVID-19 consultations (accuracy 0.97, F1-score 0.90, precision 0.85, recall 0.85, specificity 0.99). External validations showed comparable high performance. Validation on PCR test data showed high recall but low precision and specificity. Validation using hospital data showed significant correlation between COVID-19 predictions of the model and COVID-19–related hospitalizations (F1-score 96.8; P<.001; R2=0.69). Most importantly, the model was able to predict COVID-19 cases weeks before the first confirmed case in the Netherlands. Conclusions: The developed BERT model was able to accurately identify COVID-19 cases among GP consultations even preceding confirmed cases. The validated efficacy of our BERT model highlights the potential of NLP models to identify disease outbreaks early, exemplifying the power of multidisciplinary efforts in harnessing technology for disease identification. Moreover, the implications of this study extend beyond COVID-19 and offer a blueprint for the early recognition of various illnesses, revealing that such models could revolutionize disease surveillance.

19 April 2025 in International Journal of Infectious Diseases
Verified
Journal Article
Open access
Matthijs S. BerendsMatthijs S. Berends
Matthijs S. Berends
MHMaarten Homburg
Maarten Homburg
TKThijmen Kupers
Thijmen Kupers
EMEline N. Meijer
Eline N. Meijer
IBIsabelle Bos
Isabelle Bos
LPLilian L. Peters
Lilian L. Peters
Post-acute sequelae of COVID-19 (PASC), or Long COVID, involves persistent symptoms following acute infection, posing a global health challenge. While a growing number of studies have investigated potential predictors and risk factors, uncertainties remain regarding their consistency and clinical applicability. This study investigates PASC prevalence, comorbidities, demographics and viral variants using Dutch primary care electronic healthcare records (EHR). A retrospective cohort study used EHR data from 59 general practices in the Northern Netherlands, including 19,638 SARS-CoV-2 PCR-positive patients from January 1, 2020, to December 31, 2021. PASC was identified via World Health Organization and CDC guidelines, a Dutch Word2Vec model, and clinical assessments. Relative risk (RR) calculations analysed comorbidities, demographics and viral variants. PASC prevalence was 5.8% (95% CI: 5.4-6.1%). Comorbidities significantly increasing PASC risk included lung disease (RR: 1.95), cardiovascular disease (RR: 1.73), diabetes (RR: 1.82), kidney disease (RR: 1.98) and mental illness (RR: 1.29). Females and individuals aged ≥45 had increased risk. Multivariate regression revealed higher odds of prolonged PASC for ages 45-59 (adjusted odds ratios [AOR]: 3.02), 60-74 (AOR: 3.25) and 75+ (AOR: 2.44). Combined mental illness and lung disease further increased risk (AOR: 2.55). Chronic conditions, multimorbidity and demographics significantly influence PASC onset and duration. Targeted interventions may mitigate its long-term impact.

21 December 2023 in BJGP Open
Verified
Journal Article
Open access
PDF file
MHMaarten Homburg
Maarten Homburg
MBMarjolein Berger
Marjolein Berger
Matthijs S. BerendsMatthijs S. Berends
Matthijs S. Berends
EMEline Meijer
Eline Meijer
TKThijmen Kupers
Thijmen Kupers
LPLilian Peters
Lilian Peters
Background: Many countries observed a sharp decline in the use of general practice services after the outbreak of the COVID-19 pandemic. However, research has not yet considered how changes in healthcare consumption varied among regions with the same restrictive measures but different COVID-19 prevalence. Aim: To investigate how the COVID-19 pandemic affected healthcare consumption in Dutch general practice during 2020 and 2021, among regions with known heterogeneity in COVID-19 prevalence, from a pre-pandemic baseline in 2019. Design & setting: Population-based cohort study using electronic health records. The study was undertaken in Dutch general practices involved in regional research networks. Method: An interrupted time-series analysis of changes in healthcare consumption from before to during the pandemic was performed. Descriptive statistics were used on the number of potential COVID-19-related contacts, reason for contact, and type of contact. Results: The study covered 3 595 802 contacts (425 639 patients), 3 506 637 contacts (433 340 patients), and 4 105 413 contacts (434 872 patients) in 2019, 2020, and 2021, respectively. Time-series analysis revealed a significant decrease in healthcare consumption after the outbreak of the pandemic. Despite interregional heterogeneity in COVID-19 prevalence, healthcare consumption decreased comparably over time in the three regions, before rebounding to a level significantly higher than baseline in 2021. Physical consultations transitioned to phone or digital over time. Conclusion: Healthcare consumption decreased irrespective of the regional prevalence of COVID-19 from the start of the pandemic, with the Delta variant triggering a further decrease. Overall, changes in care consumption appeared to reflect contextual factors and societal restrictions rather than infection rates.

8 April 2023 in npj Primary Care Respiratory Medicine
Verified
Journal Article
Open access
PDF file
CRCorinne Rijpkema
Corinne Rijpkema
LRLotte Ramerman
Lotte Ramerman
MHMaarten Homburg
Maarten Homburg
EMEline Meijer
Eline Meijer
JMJean W.M. Muris
Jean W.M. Muris
RVRobert Verheij
Robert Verheij
The impact of the COVID-19 pandemic on general practitioners’ (GP) care for patients with asthma and/or COPD is largely unknown. To describe the impact of the pandemic on asthma or COPD-related GP care, we analysed routinely recorded electronic health records data from Dutch general practices and out-of-hours (OOH) services. During the COVID-19 pandemic (2020), the contact rates for asthma and/or COPD were significantly lower in GP practices and OOH services compared with the pre-pandemic period (2019) (respectively, 15% lower and 28% lower). The proportion of telephone contacts increased significantly with 13%-point in GP practices and 12%-point at OOH services, while the proportion of face-to-face contacts decreased. Furthermore, the proportion of high urgent contacts with OOH services decreased by 8.5%-point. To conclude, the overall contact rates in GP practices and OOH services decreased, while more contacts were remote. Lower contact rates have, after a short follow-up, not resulted in more patients with exacerbations in OOH care. However, this might still be expected after a longer follow-up.

5 September 2025 in Microbes and Infection
Verified
Journal Article
Open access
MGMariella Greutmann
Mariella Greutmann
KBKarsten Borgwardt
Karsten Borgwardt
SBSarah C. Brüningk
Sarah C. Brüningk
FFFabian Franzeck
Fabian Franzeck
CGChristian G. Giske
Christian G. Giske
AEAdrian Egli
Adrian Egli
Rapid advancements in artificial intelligence (AI) and machine learning (ML) offer significant potential to transform medical microbiology diagnostics, improving pathogen identification, antimicrobial susceptibility prediction and outbreak detection. To address these opportunities and challenges, the ESCMID workshop, "Artificial Intelligence and Machine Learning in Medical Microbiology Diagnostics", was held in Zurich, Switzerland, from June 2-5, 2025. The course featured expert lectures, practical sessions and panel discussions covering foundational ML concepts and deep learning architectures, data interoperability, quality control processes, model development and validation strategies. Key applications discussed included whole-genome sequencing for antimicrobial resistance detection, AI-enhanced digital microscopy automation and MALDI-TOF mass spectrometry-based diagnostics. Participants gained hands-on experience with essential AI tools and platforms. Special emphasis was placed on standardised laboratory protocols, regulatory compliance and ethical considerations, including data governance and patient privacy. Panel sessions further highlighted critical issues of equity, global disparities in AI access, sustainability and environmental impacts related to AI infrastructure. The workshop concluded by underscoring a necessity for ongoing interdisciplinary collaboration, continued education, and substantial investment in equitable AI infrastructure to realise the full potential of AI in clinical diagnostics.

19 October 2026 in bioRxiv
Verified
Preprint
Open access
PDF file
AGAlejandro Guerrero-López
Alejandro Guerrero-López
EMEline Meijer
Eline Meijer
YGYukino Gütlin
Yukino Gütlin
KKKristoffer Koch
Kristoffer Koch
NRNighat Rabbani
Nighat Rabbani
AEAdrian Egli
Adrian Egli
Matrix-Assisted Laser Desorption/Ionization Time-of-Flight (MALDI–TOF) mass spectrometry (MS) is central to routine clinical microbiology for species identification. Its combination with machine learning is constrained by fragmented public datasets, species imbalance and acquisition-related variability. Synthetic datasets could provide more training data, but they must preserve the spectral information needed to make predictions on real samples. Here, we present an open, synthetic MALDI–TOF MS atlas comprising 4.3 million spectra across 43 clinically relevant bacterial species. To construct the atlas, we harmonised public repositories from three countries and trained a species- and institution-conditioned concept-bottleneck generative variational autoencoder (CBGM-VAE) on 179,086 measured spectra. We assessed the atlas as both a stand-alone training resource and a complement to real data, using bacterial species classification as a functional benchmark. Evaluation comprised 7,269 real spectra from more than 7 countries, 24 institutions and multiple acquisition platforms, covering: temporal, technical, biological and combined shifts. A neural network trained only on synthetic spectra achieved F1 scores ranging from 0.788 to 1.000 across external cohorts. When synthetic spectra were combined with real training data, classification performance improved or matched using real data only. These findings demonstrate that the atlas retains transferable information and can complement measured data. The openly released atlas provides a reusable resource for supervised learning and benchmarking. It can also be explored for pretraining representation-learning and foundation models in MALDI–TOF MS research.