Biology subjects

Guerrero-Lopez, A.

Publications and source records attributed to Guerrero-Lopez, A..

9 recordsLinked to original sources

bioRxiv2026

A synthetic MALDI-TOF MS atlas for microbial classification across hospitals

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.

microbiology↗

bioRxiv2025

MARISMa: a routine MALDI-TOF MS database from 2018 to 2024

Clinical microbiology laboratories play a crucial role in identifying pathogens, guiding antibiotic treatment, and managing antimicrobial resistance (AMR). Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS) has become essential for rapid, accurate, and cost-effective microbial identification. Recent advances in integrating MALDI-TOF MS with Artificial Intelligence (AI) show promise in improving microbial detection and prediction of AMR. However, progress is limited by the lack of comprehensive and openly accessible datasets that restrict the validation, reproducibility, and applicability of the model. To address this gap, we introduce a publicly available MALDI-TOF MS dataset comprising 202,700 unique spectra from isolates collected between 2018 and 2024 at the Hospital General Universitario Gregorio Maranon, Spain. This dataset includes 186,213 bacteria, 16,163 fungal, and 371 mycobacterial samples, of which 29,679 contain AMR annotations. This resource is openly and freely shared, rigorously curated, and designed to support a wide range of machine learning. By ensuring unrestricted access to high-quality, standardized data, this dataset aims to promote transparency, reproducibility, comparative benchmarking, and collaborative progress in AI-driven clinical microbiology.

microbiology↗

bioRxiv2025

Bayesian automatic screening of pneumoniaand lung lesions localization from CT scans. Acombined method toward a more user-centredand explainable approach

While semantic segmentation allows precise localization of potential lesions, a segmentation based on object detection using bounding boxes is considered more effective for indicating the location of the target without replacing clinical expertise, reducing potential attentional and automation biases. In this context, this work lays a foundation for better and more explainable localization of lesions in the lungs due to pneumonia by applying a combined approach first using a Bayesian uncertainty measure to evaluate the likelihood of a specific slice being positive in a particular disease and, second, directing the localization of the lesions toward those positive images following object detection methods. This approach provides confidence measures for the models predictions and the regions of interest (i.e., bounding boxes) pointing to the lesions, improves the overall models explainability, and leaves the expert at the centre of the decision. However, despite its advantages, identifying the existing lesions using bounding boxes also comes with challenges, especially in effectively managing overlapping boxes and developing a fusion strategy to combine smaller boxes into larger ones. In this respect, this paper also proposes a methodology to deal with such a problem. The methods proposed are tested with a large corpus of approximately 90,000 Computed Tomography (CT) images extracted from public datasets of COVID-19, bacterial, fungal and other viral types of pneumonia, as well as control individuals, paying special attention to the first of these diseases due to the increasing interest raised by the recent pandemic. For this aim, this work uses a Bayesian neural network based on DenseNet121 for the screening module; and Cascade R-CNN, RetinaNet, and YOLO architectures for identifying the lesions. The method shows screening accuracies for COVID-19 up to 94.61%, which are complemented with robust uncertainty measures. The localization of the associated lesions is carried out with an mAP50 of 0.5559. The results and the methodology can easily be extrapolated to other pneumonias.

bioengineering↗

bioRxiv2025

Applied Machine Learning for human bacteriaMALDI-TOF Mass Spectrometry: a systematicreview

Bacterial identification, antimicrobial resistance prediction, and strain typification are critical tasks in clinical microbiology, essential for guiding patient treatment and controlling the spread of infectious diseases. While Machine Learning (ML) has shown immense promise in enhancing Matrix-Assisted Laser Desorption/Ionization-Time of Flight Mass Spectrometry (MALDI-TOF MS) applications for these tasks, there is currently no comprehensive review that fully addresses this from a technical ML perspective. To address this gap, we systematically reviewed 115 studies published between 2004 and 2025, focusing on key ML aspects such as data size and balance, pre-processing pipelines, model selection and evaluation, open-source data, and code availability. Our analysis highlights the predominant use of classical ML models like Random Forest and Support Vector Machines, alongside emerging interest in Deep Learning approaches for handling complex, high-dimensional data. Despite significant progress, challenges such as inconsistent pre-processing workflows, reliance on black-box models, limited external validation, and insufficient open-source resources persist, hindering transparency, reproducibility, and broader adoption. This review offers actionable insights to enhance ML-driven bacterial diagnostics, advocating for standardized methodologies, greater transparency, and improved data accessibility. In addition, we provide guidelines on how to approach ML for MALDI-TOF MS analysis, helping researchers navigate key decisions in model development and evaluation.

bioengineering↗

bioRxiv2024

Automatic surveillance of Escherichia coli bacteriological strains within clinical settings

Healthcare-associated infections (HAIs) are a significant concern within hospital environments, with the World Health Organization (WHO) identifying them as a major source of bacteriological infections. HAIs affect millions of patients annually, leading to substantial morbidity and mortality. However, a significant proportion of HAIs are preventable through early detection and appropriate intervention and isolation. Traditional methods for identifying bacterial species and strains, such as antigen tests, are often time-consuming and hamper the realtime tracking of outbreaks. The matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) technique has emerged as a more rapid and precise alternative, though it remains limited by manual identification processes and database constraints. Recent advancements have demonstrated the potential of combining MALDI-TOF MS data with machine learning (ML) to enhance in silico identification speed. In this paper we propose the first method for unsupervised Escherichia coli novel strain detection through the application of a efficient PIKE tailored to strain identification, spectral clustering techniques and a one-class support vector machine (OCSVM) novelty detection model, presenting promising results.

microbiology↗

bioRxiv2024

Overcoming Challenges of Reproducibility and Variability for the Clostridioides difficile typification

Machine learning (ML) approaches applied to Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry (MALDI-TOF MS) spectra have shown promise for the typing of Clostridioides difficile, yet their deployment in routine clinical settings remains challenging due to strong sensitivity to acquisition variability. Differences in culture media, incubation time, protein extraction protocols, and instrumentation across hospitals often lead to substantial performance degradation when models are evaluated under heterogeneous or previously unseen conditions. In this work, we systematically analyze the impact of methodological and technical variability on ML-based C. difficile typing and investigate whether data augmentation (DA) strategies can mitigate these effects. Using a dedicated dataset of 60 isolates acquired under diverse conditions, we show that DA substantially improves robustness to variability when training on spectra from selective C. difficile agar media. Importantly, models trained with DA achieve performance levels approaching those obtained using enriched Schaedler agar media, while relying exclusively on standard 24-hour incubation. Evaluation on an independent cohort of 28 newly acquired isolates confirms that DA significantly reduces performance degradation under real-world domain shift. To facilitate adoption and reproducibility, we release MAL-DIDA, an open-source Python library for DA of MALDI-TOF MS spectra.

microbiology↗

bioRxiv2024

Automatic semantic segmentation of the osseous structures of the paranasal sinuses

Endoscopic sinus and skull base surgeries require the use of precise neuronavigation techniques, which may take advantage of accurate delimitation of surrounding structures. This delimitation is critical for robotic-assisted surgery procedures to limit volumes of no resection. In this respect, accurate segmentation of the Osseous Structures surrounding the Paranasal Sinuses (OSPS) is a relevant issue to protect critical anatomic structures during these surgeries. Currently, manual segmentation of these structures is a labour-intensive task and requires expertise, often leading to inconsistencies. This is due to the lack of publicly available automatic models specifically tailored for the automatic delineation of the complex OSPS. To address this gap, we introduce an open-source data/model for the segmentation of these complex structures. The initial model was trained on nine complete ex vivo CT scans of the paranasal region and then improved with semi-supervised learning techniques. When tested on an external data set recorded under different conditions and with various scanners, it achieved a DICE score of 94.82 {+/-} 0.9. These results underscore the effectiveness of the model and its potential for broader research applications. By providing both the dataset and the model publicly available, this work aims to catalyse further research that could improve the precision of clinical interventions of endoscopic sinus and skull-based surgeries.

bioengineering↗

bioRxiv2024

Rapid, automatic typing of Clostridioides difficile Ribotypes Using MALDI-TOF MS

Clostridioides difficile is a major cause of hospital-acquired diarrhea, posing significant clinical challenges due to its high morbi-mortality rates and its involvement in nosocomial outbreaks. Detecting its toxigenic ribotypes rapidly and accurately is crucial for effective outbreak control. This study aimed to create a rapid diagnostic methodology based on MALDI-TOF MS and Machine Learning algorithms to differentiate toxigenic C. difficile RTs. MALDI-TOF spectra were acquired from 379 clinical isolates sourcing from 10 Spanish hospitals and analysed using Clover MSDAS, a specific software for MALDI-TOF spectra analysis, considered as the state-of-the-art tool for this purpose, and AutoCdiff, an ad hoc software developed in this study. Seven biomarker peaks were found to differentiate epidemic RT027 and RT181 strains from other RTs (2463, 3353, 4933, 4993, 6187, 6651 and 6710 m/z). Two peaks (2463 and 4993 m/z) were specifically found in RT027 isolates while combinations of the other 5 peaks allowed the differentiation of RT181 from other ribotypes. Automatic classification tools developed in Clover MSDAS and AutoCdiff using the specific peaks and the entire protein spectra, respectively, showed up to 100% balanced accuracy. Both methods allowed correct ribotype assignment for isolates sourcing from real-time outbreaks. The developed models, available from Clover MSDAS and the AutoCdiff website -https://bacteria.id-offer researchers a valuable tool for quick C. difficile ribotype determination based on MALDI-TOF spectra analysis. Although further validation of the models is still required, they represent rapid and cost-effective methods for standardized C. difficile ribotype assignment.

microbiology↗

bioRxiv2021

Development and Validation of a MALDI-TOF-Based Model to Predict Extended-Spectrum Beta-Lactamase and/or Carbapenemase-Producing in Klebsiella pneumoniae Clinical Isolates

Matrix-Assisted Laser Desorption Ionization Time-Of-Flight (MALDI-TOF) Mass Spectrometry (MS) is a reference method for microbial identification and it can be used to predict Antibiotic Resistance (AR) when combined with artificial intelligence methods. However, current solutions need time-costly preprocessing steps, are difficult to reproduce due to hyperparameter tuning, are hardly interpretable, and do not pay attention to epidemiological differences inherent to data coming from different centres, which can be critical. We propose using a multi-view heterogeneous Bayesian model (KSSHIBA) for the prediction of AR using MALDI-TOF MS data together with their epidemiological differences. KSSHIBA is the first model that removes the ad-hoc preprocessing steps that work with raw MALDI-TOF data. In addition, due to its Bayesian probabilistic nature, it does not require hyperparameter tuning, provides interpretable results, and allows exploiting local epidemiological differences between data sources. To test the proposal, we used data from 402 Klebsiella pneumoniae isolates coming from two different domains and 20 different hospitals located in Spain and Portugal. KSSHIBA outperforms current state-of-the-art approaches in antibiotic susceptibility prediction, obtaining a 0.78 AUC score in Wild Type classification and a 0.90 AUC score in Extended-Spectrum Beta-Lactamases (ESBL)+Carbapenemases (CP)-producers. The proposal consistently removes the need for ad-hoc preprocessing by working with raw MALDI-TOF data, which, in turn, reduces the time needed to obtain the results of the resistance mechanism in microbiological laboratories. The proposed model implementation as well as both data domains are publicly available.

microbiology↗