Hello. I am Ruben

Manuscript withheld due to ongoing provisional patent application

Unlocking ██████████████ through ██████████████ in Critically Ill Patients.

██████████████ diagnostics in intensive care units (ICUs) rely on ██████████████, which are time-consuming or prone to false positives, respectively. These limitations can delay critical treatment and contribute to ██████████████. This study evaluated ██████████████ combined with ██████████████ to predict ██████████████, based on ██████████████ of ICU patients. Three patient groups were examined: those with ██████████████, those without, and a mixed group. Models accurately distinguished ██████████████ ██████████████, discriminated ██████████████ ██████████████, identified ██████████████ ██████████████, and distinguished ██████████████ ██████████████. Discrimination of ██████████████ due to limited samples, showing varied sensitivity ████ and specificity ████. All these results point to a high potential for rapid, precise diagnostics of ██████████████ among critically ill patients, reducing reliance ██████████████ and enhancing more precise therapy, thereby improving patient outcomes.

INFECTION, BIOMARKERS, MACHINE LEARNING, FTIR SPECTROSCOPY, SERUM

Manuscript published on Proteomes, August 2025

Multiplex Targeted Proteomic Analysis of Cytokine Ratios for ICU Mortality in Severe COVID-19.

Background: Accurate and timely prediction of mortality in intensive care unit (ICU) patients, particularly those with COVID-19, remains clinically challenging due to complex immune responses. Proteomic cytokine profiling holds promise for refining mortality risk assessment. Methods: Serum samples from 89 ICU patients (55 discharged, 34 deceased) were analyzed using a multiplex 21-cytokine panel. Samples were stratified into three groups based on time from collection to outcome: ≤48 h (Group 1: Early), >48 h to ≤7 days (Group 2: Intermediate), and >7 days to ≤14 days (Group 3: Late). Cytokine levels, simple cytokine ratios, and previously unexplored complex ratios between pro- and anti-inflammatory cytokines were evaluated. Machine learning-based feature selection identified the most predictive ratios, with performance evaluated by area under the curve (AUC), sensitivity, and specificity. Results: Complex cytokine ratios demonstrated superior predictive accuracy compared to traditional severity markers (APACHE II, SAPS II, SOFA), individual cytokines, and simple ratios, effectively distinguishing discharged from deceased patients across all groups (AUC: 0.918–1.000; sensitivity: 0.826–1.000; specificity: 0.775–0.900). Conclusions: Multiplex cytokine profiling enhanced by computationally derived complex ratios may offer robust predictive capabilities for ICU mortality risk stratification, serving as a valuable tool for personalized prognosis in critical care.

CYTOKINE PROFILING, ICU MORTALITY, PROTEOMICS, MULTIPLEX CYTOKINE ANALYSIS, COMPLEX CYTOKINE RATIOS, MACHINE LEARNING, CRITICAL CARE

Manuscript published on Metabolites, March 2025

Cytokine-Based Insights into Bloodstream Infections and Bacterial Gram Typing in ICU COVID-19 Patients.

Background: Timely and accurate identification of bloodstream infections (BSIs) in intensive care unit (ICU) patients remains a key challenge, particularly in COVID-19 settings, where immune dysregulation can obscure early clinical signs. Methods: Cytokine profiling was evaluated to discriminate between ICU patients with and without BSIs, and, among those with confirmed BSIs, to further stratify bacterial infections by Gram type. Serum samples from 45 ICU COVID-19 patients were analyzed using a 21-cytokine panel, with feature selection applied to identify candidate markers. Results: A machine learning workflow identified key features, achieving robust performance metrics with AUC values up to 0.97 for BSI classification and 0.98 for Gram typing. Conclusions: In contrast to traditional approaches that focus on individual cytokines or simple ratios, the present analysis employed programmatically generated ratios between pro-inflammatory and anti-inflammatory cytokines, refined through feature selection. Although further validation in larger and more diverse cohorts is warranted, these findings underscore the potential of advanced cytokine-based diagnostics to enhance precision medicine in infection management.

CYTOKINE PROFILING, BLOODSTREAM INFECTIONS, GRAM TYPING, ICU DIAGNOSTICS, COVID-19, MACHINE LEARNING

Manuscript published on the International Journal of Molecular Sciences, December 2024

Early Mortality Prediction in Intensive Care Unit Patients Based on Serum Metabolomic Fingerprint.

Predicting mortality in intensive care units (ICUs) is essential for timely interventions and efficient resource use, especially during pandemics like COVID-19, where high mortality persisted even after the state of emergency ended. Current mortality prediction methods remain limited, especially for critically ill ICU patients, due to their dynamic metabolic changes and heterogeneous pathophysiological processes. This study evaluated how the serum metabolomic fingerprint, acquired through Fourier-Transform Infrared (FTIR) spectroscopy, could support mortality prediction models in COVID-19 ICU patients. A preliminary univariate analysis of serum FTIR spectra revealed significant spectral differences between 21 discharged and 23 deceased patients; however, the most significant spectral bands did not yield high-performing predictive models. By applying a Fast-Correlation-Based Filter (FCBF) for feature selection of the spectra, a set of spectral bands spanning a broader range of molecular functional groups was identified, which enabled Naïve Bayes models with AUCs of 0.79, 0.97, and 0.98 for the first 48 h of ICU admission, seven days prior, and the day of the outcome, respectively, which are, in turn, defined as either death or discharge from the ICU. These findings suggest FTIR spectroscopy as a rapid, economical, and minimally invasive diagnostic tool, but further validation is needed in larger, more diverse cohorts.

ICU MORTALITY PREDICTION, SERUM BIOMARKERS, FTIR SPECTROSCOPY, OMICS

Manuscript published on Metabolites, May 2024

Discovery of Delirium Biomarkers through Minimally Invasive Serum Molecular Fingerprinting.

Delirium presents a significant clinical challenge, primarily due to its profound impact on patient outcomes and the limitations of the current diagnostic methods, which are largely subjective. During the COVID-19 pandemic, this challenge was intensified as the frequency of delirium assessments decreased in Intensive Care Units (ICUs), even as the prevalence of delirium among critically ill patients increased. The present study evaluated how the serum molecular fingerprint, as acquired by Fourier-Transform InfraRed (FTIR) spectroscopy, can enable the development of predictive models for delirium. A preliminary univariate analysis of serum FTIR spectra indicated significantly different bands between 26 ICU patients with delirium and 26 patients without, all of whom were admitted with COVID-19. However, these bands resulted in a poorly performing Naïve-Bayes predictive model. Considering the use of a Fast-Correlation-Based Filter for feature selection, it was possible to define a new set of spectral bands with a wider coverage of molecular functional groups. These bands ensured an excellent Naïve-Bayes predictive model, with an AUC, a sensitivity, and a specificity all exceeding 0.92. These spectral bands, acquired through a minimally invasive analysis and obtained rapidly, economically, and in a high-throughput mode, therefore offer significant potential for managing delirium in critically ill patients.

DELIRIUM, BIOMARKERS, FTIR SPECTROSCOPY, SERUM, OMICS

Manuscript published on Methods and Protocols, April 2024

Simplifying Data Analysis in Biomedical Research: An Automated, User-Friendly Tool.

Robust data normalization and analysis are pivotal in biomedical research to ensure that observed differences in populations are directly attributable to the target variable, rather than disparities between control and study groups. ArsHive addresses this challenge using advanced algorithms to normalize populations (e.g., control and study groups) and perform statistical evaluations between demographic, clinical, and other variables within biomedical datasets, resulting in more balanced and unbiased analyses. The tool’s functionality extends to comprehensive data reporting, which elucidates the effects of data processing, while maintaining dataset integrity. Additionally, ArsHive is complemented by A.D.A. (Autonomous Digital Assistant), which employs OpenAI’s GPT-4 model to assist researchers with inquiries, enhancing the decision-making process. In this proof-of-concept study, we tested ArsHive on three different datasets derived from proprietary data, demonstrating its effectiveness in managing complex clinical and therapeutic information and highlighting its versatility for diverse research fields.

BIOMEDICAL RESEARCH, MACHINE LEARNING, LLM MODELS, HIGH DIMENSIONAL DATA ANALYSIS

Manuscript published on Biotech, December 2022

Plasma versus Serum Analysis by FTIR Spectroscopy to Capture the Human Physiological State.

Fourier Transform InfraRed spectroscopy of serum and plasma has been highly explored for medical diagnosis, due to its general simplicity, and high sensitivity and specificity. To evaluate the plasma and serum molecular fingerprint, as obtained by FTIR spectroscopy, to acquire the system metabolic state, serum and plasma spectra were compared to characterize the metabolic state of 30 human volunteers, between 90 days consumption of green tea extract rich in Epigallocatechin-3-gallate (EGCG). Both plasma and serum spectra enabled the high impact of EGCG consumption on the biofluid spectra to be observed, as analyzed by the spectra principal component analysis, hierarchical-cluster analysis, and univariate data analysis. Plasma spectra resulted in the prediction of EGCG consumption with a slightly higher specificity, accuracy, and precision, also pointing to a higher number of significant spectral bands that were different between the 90 days period. Despite this, the lipid regions of the serum spectra were more affected by EGCG consumption than the corresponding plasma spectra. Therefore, in general, if no specific compound analysis is highlighted, plasma is in general the advised biofluid to capture by FTIR spectroscopy the general metabolic state. If the lipid content of the biofluid is relevant, serum spectra could present some advantages over plasma spectra.

EPIGALLOCATECHIN-3-GALLATE, FTIR SPECTROSCOPY, PLASMA, SERUM

Manuscript published on Metabolites, January 2022

Infection Biomarkers Based on Metabolomics.

Current infection biomarkers are highly limited since they have low capability to predict infection in the presence of confounding processes such as in non-infectious inflammatory processes, low capability to predict disease outcomes and have limited applications to guide and evaluate therapeutic regimes. Therefore, it is critical to discover and develop new and effective clinical infection biomarkers, especially applicable in patients at risk of developing severe illness and critically ill patients. Ideal biomarkers would effectively help physicians with better patient management, leading to a decrease of severe outcomes, personalize therapies, minimize antibiotics overuse and hospitalization time, and significantly improve patient survival. Metabolomics, by providing a direct insight into the functional metabolic outcome of an organism, presents a highly appealing strategy to discover these biomarkers. The present work reviews the desired main characteristics of infection biomarkers, the main metabolomics strategies to discover these biomarkers and the next steps for developing the area towards effective clinical biomarkers.

INFECTION, METABOLOMICS, BIOMARKERS, DIAGNOSIS, PROGNOSIS

Manuscript published on Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, March 2021

A New Method to Predict Genotoxic Effects Based on Serum Molecular Profile.

It is critical to develop new methods to assess genotoxic effects in human biomonitoring since the conventional methods are usually laborious, time-consuming, and expensive. It is aimed to evaluate if the analysis of a drop of serum by Fourier Transform Infrared spectroscopy, allow to assess genotoxic effects in occupational exposure to cytostatic drugs in hospital professionals, as obtained by the lymphocyte cytokinesis-block micronucleus assay. It was considered peripheral blood from hospital professionals exposed to cytostatic drugs (n = 22) and from a non-exposed group (n = 36). It was observed that workers occupationally exposed presented a higher number of micronuclei (p < 0.05) in lymphocytes, in relation to the non-exposed group. The serum Fourier Transform Infrared spectra from exposed workers presented diverse different peaks (p < 0.01) in relation to the non-exposed group. The hierarchical cluster analysis of serum spectra separated serum samples of the exposed group from the non-exposed group with 61% sensitivity and 88% specificity. A support vector machine model of serum spectra enables to predict exposure with high accuracy (0.91), precision (0.89), sensitivity (0.86), F1 score (0.87) and AUC (0.96). Therefore, Fourier Transform Infrared spectroscopic analysis of a drop of serum enabled to predict in a rapid and simple mode the genotoxic effects of cytostatic drugs. The method presents therefore potential for high-dimension screening of exposure of genotoxic substances, due to its simplicity and rapid setup mode.

Manuscript published on High-throughput (now BioTech), April 2020

A Simple, Label-Free, and High-Throughput Method to Evaluate the Epigallocatechin-3-Gallate Impact in Plasma Molecular Profile.

Epigallocatechin-3-gallate (EGCG), the major catechin present in green tea, presents diverse appealing biological activities, such as antioxidative, anti-inflammatory, antimicrobial, and antiviral activities, among others. The present work evaluated the impact in the molecular profile of human plasma from daily consumption of 225 mg of EGCG for 90 days. Plasma from peripheral blood was collected from 30 healthy human volunteers and analyzed by high-throughput Fourier transform infrared spectroscopy. To capture the biochemical information while minimizing the interference of physical phenomena, several combinations of spectra pre-processing methods were evaluated by principal component analysis. The pre-processing method that led to the best class separation, that is, between the plasma spectral data collected at the beginning and after the 90 days, was a combination of atmospheric correction with a second derivative spectra. A hierarchical cluster analysis of second derivative spectra also highlighted the fact that plasma acquired before EGCG consumption presented a distinct molecular profile after the 90 days of EGCG consumption. It was also possible by partial least squares regression discriminant analysis to correctly predict all unlabeled plasma samples (not used for model construction) at both timeframes. We observed that the similarity in composition among the plasma samples was higher in samples collected after EGCG consumption when compared with the samples taken prior to EGCG consumption. Diverse negative peaks of the normalized second derivative spectra, associated with lipid and protein regions, were significantly affected (p < 0.001) by EGCG consumption, according to the impact of EGCG consumption on the patients’ blood, low density and high density lipoproteins ratio. In conclusion, a single bolus dose of 225 mg of EGCG, ingested throughout a period of 90 days, drastically affected plasma molecular composition in all participants, which raises awareness regarding prolonged human exposure to EGCG. Because the analysis was conducted in a high-throughput, label-free, and economic analysis, it could be applied to high-dimension molecular epidemiological studies to further promote the understanding of the effect of bio-compound consumption mode and frequency.

HIGH-THROUGHPUT, FTIR SPECTROSCOPY, EGCG, PLASMA

Manuscript published on AI in Medicine, November 2025

Mapping Anti-HLA Class I Cross-Reactivity for Transplantation Using Interpretable Embedding and Clustering of SAB MFI.

Background/Objectives: Rapid, objective phenotyping of donor kidneys is needed to support peri-implant decisions. Label-free Fourier-transform infrared (FTIR) spectroscopy of static cold-storage Celsior® perfusion fluid can discriminate kidneys recovered from donation after circulatory death (DCD) versus donation after brain death (DBD). Methods: Preservation solution from isolated kidney allografts (n = 10; 5 DCD/5 DBD) matched on demographics was analyzed in the Amide I and fingerprint regions. Several spectral preprocessing steps were applied, and feature extraction was based on the Fast Correlation-Based Filter. Support vector machines and Naïve Bayes were evaluated. Unsupervised structure was assessed based on cosine distance, multidimensional scaling, and hierarchical clustering. Two-dimensional correlation spectroscopy (2D-COS) was used to examine band co-variation. Results: Donor cohorts were well balanced, except for higher terminal serum creatinine in DCD. Quality metrics were comparable, indicating no systematic technical bias. In Amide I, derivatives improved classification, but performance remained modest (e.g., second derivative with feature selection yielded an area under the curve (AUC) of 0.88 and an accuracy of 0.90 for support vector machines; Naïve Bayes reached an AUC of 0.92 with an accuracy of 0.70). The fingerprint window was most informative. Naïve Bayes with second derivative plus feature selection identified bands at ~1202, ~1203, ~1342, and ~1413 cm⁻¹ and achieved an AUC of 1.00 and an accuracy of 1.00. Unsupervised analyses showed coherent grouping in the fingerprint region, and 2D correlation maps indicated coordinated multi-band changes. Conclusions: Performance in this 10-sample pilot should be interpreted cautiously, as perfect leave-one-out cross-validation (LOOCV) estimates are vulnerable to overfitting. The findings are preliminary and hypothesis-generating, and they require confirmation in larger, multicenter cohorts with a pre-registered analysis pipeline and external validation.

HLA CLASS I, SINGLE-ANTIGEN BEAD, MEAN FLUORESCENCE INTENSITY, MULTIDIMENSIONAL SCALING, HIERARCHICAL CLUSTERING, COMPUTATIONAL IMMUNOLOGY, INTERPRETABLE VISUALIZATION, CROSS-REACTIVITY MAPPING

Manuscript published on Metabolites, October 2025

Rapid FTIR Spectral Fingerprinting of Kidney Allograft Perfusion Fluids Distinguishes DCD from DBD Donors: A Pilot Machine Learning Study.

Background/Objectives: Rapid, objective phenotyping of donor kidneys is needed to support peri-implant decisions. Label-free Fourier-transform infrared (FTIR) spectroscopy of static cold-storage Celsior® perfusion fluid can discriminate kidneys recovered from donation after circulatory death (DCD) versus donation after brain death (DBD). Methods: Preservation solution from isolated kidney allografts (n = 10; 5 DCD/5 DBD) matched on demographics was analyzed in the Amide I and fingerprint regions. Several spectral preprocessing steps were applied, and feature extraction was based on the Fast Correlation-Based Filter. Support vector machines and Naïve Bayes were evaluated. Unsupervised structure was assessed based on cosine distance, multidimensional scaling, and hierarchical clustering. Two-dimensional correlation spectroscopy (2D-COS) was used to examine band co-variation. Results: Donor cohorts were well balanced, except for higher terminal serum creatinine in DCD. Quality metrics were comparable, indicating no systematic technical bias. In Amide I, derivatives improved classification, but performance remained modest (e.g., second derivative with feature selection yielded an area under the curve (AUC) of 0.88 and an accuracy of 0.90 for support vector machines; Naïve Bayes reached an AUC of 0.92 with an accuracy of 0.70). The fingerprint window was most informative. Naïve Bayes with second derivative plus feature selection identified bands at ~1202, ~1203, ~1342, and ~1413 cm⁻¹ and achieved an AUC of 1.00 and an accuracy of 1.00. Unsupervised analyses showed coherent grouping in the fingerprint region, and 2D correlation maps indicated coordinated multi-band changes. Conclusions: Performance in this 10-sample pilot should be interpreted cautiously, as perfect leave-one-out cross-validation (LOOCV) estimates are vulnerable to overfitting. The findings are preliminary and hypothesis-generating, and they require confirmation in larger, multicenter cohorts with a pre-registered analysis pipeline and external validation.

FTIR SPECTROSCOPY, KIDNEY TRANSPLANTATION, PERFUSION FLUID, DCD vs. DBD, MACHINE LEARNING

Manuscript published on GeroScience, July 2025

Predicting Delirium in Critically Ill COVID-19 Patients Using EEG-Derived Data: a Machine Learning Approach.

Delirium is a severe and common complication among critically ill patients, particularly those with SARS-CoV-2 infection, contributing to increased morbidity and mortality. Early identification of at-risk patients is crucial for timely intervention and improved outcomes. This prospective observational cohort study explores the potential of electroencephalography (EEG) combined with machine learning (ML) models for predicting delirium in critically ill patients with SARS-CoV-2 infection. A stepwise modeling approach was applied, starting with the independent analysis of specific EEG variables to assess their predictive value. Subsequently, three ML models were developed using data from 70 patients (31 with delirium, 39 without): two relied solely on EEG data, while the third integrated demographic, clinical, laboratory, and EEG data. An additional model analyzed EEG data before and after delirium diagnosis in 11 patients. Several EEG features were identified as predictors of delirium, with increased theta activity emerging as the most consistent. The best EEG-only model achieved an area under the curve (AUC) of 0.733 (sensitivity = 0.645, specificity = 0.692), indicating moderate predictive performance. Including demographic, clinical, and laboratory variables improved performance (AUC = 0.825, sensitivity = 0.613, specificity = 0.795). The model analyzing EEG features before and after delirium diagnosis achieved the highest accuracy (AUC = 0.950, sensitivity and specificity = 0.818), reinforcing the value of EEG-based monitoring. EEG-based ML models show promise for predicting delirium in critically ill patients, with increased theta activity identified as a key predictor. However, their moderate AUC, sensitivity, and specificity highlight the need for further refinement.

DELIRIUM, EEG, COVID-19, SARS-CoV-2 INFECTION, ICU, MACHINE LEARNING

Manuscript published on Life, June 2025

Predicting ICU Delirium in Critically Ill COVID-19 Patients Using Demographic, Clinical, and Laboratory Admission Data: A Machine Learning Approach.

Delirium is a common and underrecognized complication among critically ill patients, associated with prolonged ICU stays, cognitive dysfunction, and increased mortality. Its multifactorial causes and fluctuating course hinder early prediction, limiting timely management. Predictive models based on data available at ICU admission may help to identify high-risk patients and guide early interventions. This study evaluated machine learning models used to predict delirium in critically ill patients with SARS-CoV-2 infections using a prospective cohort of 426 patients. The dataset included demographic characteristics, clinical data (e.g., comorbidities, medication, reason for ICU admission, interventions), and routine lab test results. Five models—Logistic Regression, Support Vector Machine, Decision Tree, Random Forest, and Naïve Bayes—were developed using 112 features. Feature selection relied on Information Gain, and model performance was assessed via 10-fold cross-validation. The Naïve Bayes model showed moderate predictive performance and high interpretability, achieving an AUC of 0.717, accuracy of 65.3%, sensitivity of 62.4%, specificity of 68.1%, and precision of 66.2%. Key predictors included invasive mechanical ventilation, deep sedation with benzodiazepines, SARS-CoV-2 as the reason for ICU admission, ECMO use, constipation, and male sex. These findings support the use of interpretable models for early delirium risk stratification using routinely available ICU data.

DELIRIUM, PREDICTIVE MODELLING, COVID-19, SARS-CoV-2 INFECTION, ICU, MACHINE LEARNING

Manuscript published on Applied Sciences, April 2025

Predictive Models of Patient Severity in Intensive Care Units Based on Serum Cytokine Profiles: Advancing Rapid Analysis.

Predicting disease states and outcomes—and anticipating the need for specific procedures—enhances the efficiency of patient management, particularly in the dynamic and heterogenous environments of intensive care units (ICUs). This study aimed to develop robust predictive models using small sets of blood analytes to predict disease severity and mortality in ICUs, as fewer analytes are advantageous for future rapid analyses using biosensors, enabling fast clinical decision-making. Given the substantial impact of inflammatory processes, this research examined the serum profiles of 25 cytokines, either in association with or independent of nine routine blood analyses. Serum samples from 24 male COVID-19 patients admitted to an ICU were divided into three groups: Group A, including less severe patients, and Groups B and C, that needed invasive mechanical ventilation (IMV). Patients from Group C died within seven days after the current analysis. Naïve Bayes models were developed using the full dataset or with feature subsets selected either through an information gain algorithm or univariate data analysis. Strong predictive models were achieved for IMV (AUC = 0.891) and mortality within homogeneous (AUC = 0.774) or more heterogeneous (AUC = 0.887) populations utilizing two to nine features. Despite the small sample, these findings underscore the potential for effective prediction models based on a limited number of analytes.

CYTOKINE PROFILING, INFLAMMATORY BIOMARKERS, INTENSIVE CARE UNIT, INVASIVE MECHANICAL VENTILATION, MORTALITY, MACHINE LEARNING, PROMPT ANALYSES

Manuscript published on the Journal of Clinical Medicine, January 2025

Integration of FTIR Spectroscopy and Machine Learning for Kidney Allograft Rejection: A Complementary Diagnostic Tool.

Background: Kidney transplantation is a life-saving treatment for end-stage kidney disease, but allograft rejection remains a critical challenge, requiring accurate and timely diagnosis. The study aims to evaluate the integration of Fourier Transform Infrared (FTIR) spectroscopy and machine learning algorithms as a minimally invasive method to detect kidney allograft rejection and differentiate between T Cell-Mediated Rejection (TCMR) and Antibody-Mediated Rejection (AMR). Additionally, the goal is to discriminate these rejection types aiming to develop a reliable decision-making support tool. Methods: This retrospective study included 41 kidney transplant recipients and analyzed 81 serum samples matched to corresponding allograft biopsies. FTIR spectroscopy was applied to pre-biopsy serum samples, and Naïve Bayes classification models were developed to distinguish rejection from non-rejection and classify rejection types. Data preprocessing involved, e.g., atmospheric compensation, second derivative, and feature selection using Fast Correlation-Based Filter for spectral regions 600–1900 cm−1 and 2800–3400 cm−1. Model performance was assessed via area under the receiver operating characteristic curve (AUC-ROC), sensitivity, specificity, and accuracy. Results: The Naïve Bayes model achieved an AUC-ROC of 0.945 in classifying rejection versus non-rejection and AUC-ROC of 0.989 in distinguishing TCMR from AMR. Feature selection significantly improved model performance, identifying key spectral wavenumbers associated with rejection mechanisms. This approach demonstrated high sensitivity and specificity for both classification tasks. Conclusions: The integration of FTIR spectroscopy with machine learning may provide a promising, minimally invasive method for early detection and precise classification of kidney allograft rejection. Further validation in larger, more diverse populations is needed to confirm these findings’ reliability.

KIDNEY ALLOGRAFT, REJECTION, BIOMARKERS, MACHINE LEARNING, FTIR SPECTROSCOPY

Manuscript published on Computers in Biology and Medicine, January 2025

Comparison of two metabolomics-platforms to discover biomarkers in critically ill patients from serum analysis.

Serum metabolome analysis is essential for identifying disease biomarkers and predicting patient outcomes in precision medicine. Thus, this study aims to compare Ultra-High Performance Liquid Chromatography-High-Resolution Mass Spectrometry (UHPLC-HRMS) with Fourier Transform Infrared (FTIR) spectroscopy in acquiring the serum metabolome of critically ill patients, associated with invasive mechanical ventilation (IMV), and predicting death. Three groups of 8 patients were considered. Group A did not require IMV and survived hospitalization, while Groups B and C required IMV. Group C patients died a median of 5 days after sample harvest. Good prediction models were achieved when comparing groups A to B and B to C using both platforms’ data, with UHPLC-HRMS showing 8–17 % higher accuracies (≥83 %). However, developing predictive models using metabolite sets was not feasible when comparing unbalanced populations, i.e., Groups A and B combined to Group C. Alternatively, FTIR-spectroscopy enabled the development of a model with 83 % accuracy. Overall, UHPLC-HRMS data yields more robust prediction models when comparing homogenous populations, potentially enhancing understanding of metabolic mechanisms and improving patient therapy adjustments. FTIR-spectroscopy is more suitable for unbalanced populations. Its simplicity, speed, cost-effectiveness, and high-throughput operation make it ideal for large-scale studies and clinical translation in complex populations.

Manuscript published on GeroScience, November 2024

Analysis of six consecutive waves of ICU-admitted COVID-19 patients: key findings and insights from a Portuguese population.

Identifying high-risk patients, particularly in intensive care units (ICUs), enhances treatment and reduces severe outcomes. Since the pandemic, numerous studies have examined COVID-19 patient profiles and factors linked to increased mortality. Despite six pandemic waves, to the best of our knowledge, there is no extensive comparative analysis of patients’ characteristics across these waves in Portugal. Thus, we aimed to analyze the demographic and clinical features of 1041 COVID-19 patients admitted to an ICU and their relationship with the different SARS-Cov-2 variants in Portugal. Additionally, we conducted an in-depth examination of factors contributing to early and late mortality by analyzing clinical data and laboratory results from the first 72 h of ICU admission. Our findings revealed a notable decline in ICU admissions due to COVID-19, with the highest mortality rates observed during the second and third waves. Furthermore, immunization could have significantly contributed to the reduction in the median age of ICU-admitted patients and the severity of their conditions. The factors contributing to early and late mortality differed. Age, wave number, D-dimers, and procalcitonin were independently associated with the risk of early death. As a measure of discriminative power for the derived multivariable model, an AUC of 0.825 (p < 0.001; 95% CI, 0.719–0.931) was obtained. For late mortality, a model incorporating age, wave number, hematologic cancer, C-reactive protein, lactate dehydrogenase, and platelet counts resulted in an AUC of 0.795 (p < 0.001; 95% CI, 0.759–0.831). These findings underscore the importance of conducting comprehensive analyses across pandemic waves to better understand the dynamics of COVID-19.

COVID-19 WAVES, ICU, COVID-19, VACCINATION, EARLY MORTALITY, LATE MORTALITY

Manuscript published on Biotechnology and Bioengineering, May 2024

A new methodology for a rapid and high-throughput comparison of molecular profiles and biological activity of phytoextracts.

To robustly discover and explore phytocompounds, it is necessary to evaluate the interrelationships between the plant species, plant tissue, and the extraction process on the extract composition and to predict its cytotoxicity. The present work evaluated how Fourier Transform InfraRed spectroscopy can acquire the molecular profile of aqueous and ethanol-based extracts obtained from leaves, seeds, and flowers of Cynara Cardunculus, and ethanol-based extracts from Matricaria chamomilla flowers, as well the impact of these extracts on the viability of mammalian cells. The extract molecular profile enabled to predict the extraction yield, and how the plant species, plant tissue, and extraction process affected the extract's relative composition. The molecular profile obtained from the culture media of cells exposed to extracts enabled to capture its impact on cells metabolism, at a higher sensitivity than the conventional assay used to determine the cell viability. Furthermore, it was possible to detect specific impacts on the cell's metabolism according to plant species, plant tissue, and extraction process. Since spectra were acquired on small volumes of samples (25 µL), after a simple dehydration step, and based on a plate with 96 wells, the method can be applied in a rapid, simple, high-throughput, and economic mode, consequently promoting the discovery of phytocompounds.

CYNARA CARDUNCULUS, FTIR-SPECTROSCOPY, MATRICARIA CHAMOMILLA, PHYTOCOMPOUNDS, POLYPHENOLS

Manuscript published on the International Journal of Molecular Sciences, March 2024

Predicting Cellular Rejection of Renal Allograft Based on the Serum Proteomic Fingerprint.

Kidney transplantation is an essential medical procedure that significantly enhances the survival rates and quality of life for patients with end-stage kidney disease. However, despite advancements in immunosuppressive therapies, allograft rejection remains a leading cause of organ loss. Notably, predictions of cellular rejection processes primarily rely on biopsy analysis, which is not routinely performed due to its invasive nature. The present work evaluates if the serum proteomic fingerprint, as acquired by Fourier Transform Infrared (FTIR) spectroscopy, can predict cellular rejection processes. We analyzed 28 serum samples, corresponding to 17 without cellular rejection processes and 11 associated with cellular rejection processes, as based on biopsy analyses. The leave-one-out-cross validation procedure of a Naïve Bayes model enabled the prediction of cellular rejection processes with high sensitivity and specificity (AUC > 0.984). The serum proteomic profile was obtained in a high-throughput mode and based on a simple, rapid, and economical procedure, making it suitable for routine analyses and large-scale studies. Consequently, the current method presents a high potential to predict cellular rejection processes translatable to clinical scenarios, and that should continue to be explored.

KIDNEY ALLOGRAFT, CELLULAR REJECTION, PROTEOMIC FINGERPRINT, FTIR SPECTROSCOPY

Manuscript published on BioMedInformatics, March 2024

Revolutionizing Kidney Transplantation: Connecting Machine Learning and Artificial Intelligence with Next-Generation Healthcare—From Algorithms to Allografts.

This review explores the integration of artificial intelligence (AI) and machine learning (ML) into kidney transplantation (KT), set against the backdrop of a significant donor organ shortage and the evolution of ‘Next-Generation Healthcare’. Its purpose is to evaluate how AI and ML can enhance the transplantation process, from donor selection to postoperative patient care. Our methodology involved a comprehensive review of current research, focusing on the application of AI and ML in various stages of KT. This included an analysis of donor–recipient matching, predictive modeling, and the improvement in postoperative care. The results indicated that AI and ML significantly improve the efficiency and success rates of KT. They aid in better donor–recipient matching, reduce organ rejection, and enhance postoperative monitoring and patient care. Predictive modeling, based on extensive data analysis, has been particularly effective in identifying suitable organ matches and anticipating postoperative complications. In conclusion, this review discusses the transformative impact of AI and ML in KT, offering more precise, personalized, and effective healthcare solutions. Their integration into this field addresses critical issues like organ shortages and post-transplant complications. However, the successful application of these technologies requires careful consideration of their ethical, privacy, and training aspects in healthcare settings.

KIDNEY TRANSPLANTATION, MACHINE LEARNING, ARTIFICIAL INTELLIGENCE, PRECISION MEDICINE

Manuscript published on HLA, February 2024

Evaluation of rapid optimized flow cytometry crossmatch (Halifaster) in living donor kidney transplantation.

Kidney transplantation is often the preferred treatment for end-stage renal disease. However, the presence of preformed donor-specific antibodies (DSA), including those against HLA, can lead to antibody-mediated rejection and significantly impact transplant outcomes. The Flow Cytometry Crossmatch (FCXM) is a crucial tool in kidney transplantation, as it also enables the measurement of low levels of anti-HLA DSA antibodies. However, current methodologies for detecting these antibodies, however, are time-consuming and require extensive reagents. In this study, we analyzed the performance of the Halifaster FCXM protocol in 133 consecutive living kidney donor pairs, correlating these results with single antigen-based anti-HLA DSA results. Anti-HLA DSA was identified in 31 patients (23.3%). Both T and B lymphocyte FCXM assays demonstrated high sensitivity and specificity in detecting anti-HLA DSA. Furthermore, a Tree model to determine the levels of anti-HLA DSA to produce a flow crossmatch positivity, was developed offering an accuracy of 93% and 90% for T and B lymphocytes, respectively. Both approaches point to a thresh old of 1000–2000 MFI for T lymphocytes and 3000 MFI for B lymphocytes. Our findings indicate that the Halifaster protocol facilitates fast and efficient FCXM testing without compromising accuracy, marking a significant advancement in the field of kidney transplantation. The inclusion of HLA-specific antibody analysis underscores the protocol's comprehensive approach to improving transplant outcomes.

DONOR SPECIFIC ANTIBODIES, FLOW CYTOMETRY CROSSMATCH, HALIFASTER PROTOCOL, KIDNEY TRANSPLANTATION, LIVING DONOR

Manuscript published on Medicina, December 2023

The Characteristics and Laboratory Findings of SARS-CoV-2 Infected Patients during the First Three COVID-19 Waves in Portugal—A Retrospective Single-Center Study.

Background and Objectives: Given the wide spectrum of clinical and laboratory manifestations of the coronavirus disease 2019 (COVID-19), it is imperative to identify potential contributing factors to patients’ outcomes. However, a limited number of studies have assessed how the different waves affected the progression of the disease, more so in Portugal. Therefore, our main purpose was to study the clinical and laboratory patterns of COVID-19 in an unvaccinated population admitted to the intensive care unit, identifying characteristics associated with death, in each of the first three waves of the pandemic. Materials and Methods: This study included 337 COVID-19 patients admitted to the intensive care unit of a single-center hospital in Lisbon, Portugal, between March 2020 and March 2021. Comparisons were made between three COVID-19 waves, in the second (n = 325) and seventh (n = 216) days after admission, and between discharged and deceased patients. Results: Deceased patients were considerably older (p = 0.021) and needed greater ventilatory assistance (p = 0.023), especially in the first wave. Differences between discharged and deceased patients’ biomarkers were minimal in the first wave, on both analyzed days. In the second wave significant differences emerged in troponins, lactate dehydrogenase, procalcitonin, C-reactive protein, and white blood cell subpopulations, as well as platelet-to-lymphocyte and neutrophil-to-lymphocyte ratios (all p < 0.05). Furthermore, in the third wave, platelets and D-dimers were also significantly different between patients’ groups (all p < 0.05). From the second to the seventh days, troponins and lactate dehydrogenase showed significant decreases, mainly for discharged patients, while platelet counts increased (all p < 0.01). Lymphocytes significantly increased in discharged patients (all p < 0.05), while white blood cells rose in the second (all p < 0.001) and third (all p < 0.05) waves among deceased patients. Conclusions: This study yields insights into COVID-19 patients’ characteristics and mortality-associated biomarkers during Portugal’s first three COVID-19 waves, highlighting the importance of considering wave variations in future research due to potential significant outcome differences.

CORONAVIRUS DISEASE 2019, INTENSIVE CARE UNIT, MORTALITY, BLOOD BIOMARKERS, CORONAVIRUS DISEASE 2019 WAVES

Manuscript published on Pathology, February 2024

Exosomes and microvesicles in kidney transplantation: the long road from trash to gold.

Kidney transplantation significantly enhances the survival rate and quality of life of patients with end-stage kidney disease. The ability to predict post-transplantation rejection events in their early phases can reduce subsequent allograft loss. Therefore, it is critical to identify biomarkers of rejection processes that can be acquired on routine analysis of samples collected by non-invasive or minimally invasive procedures. It is also important to develop new therapeutic strategies that facilitate optimisation of the dose of immunotherapeutic drugs and the induction of allograft immunotolerance. This review explores the challenges and opportunities offered by extracellular vesicles (EVs) present in biofluids in the discovery of biomarkers of rejection processes, as drug carriers and in the induction of immunotolerance. Since EVs are highly complex structures and their composition is affected by the parent cell's metabolic status, the importance of defining standardised methods for isolating and characterising EVs is also discussed. Understanding the major bottlenecks associated with all these areas will promote the further investigation of EVs and their translation into a clinical setting.

EXTRACELLULAR VESICLES, KIDNEY TRANSPLANT, REJECTION, DIAGNOSIS, THERAPEUTICS

Manuscript published on Proteomes, August 2023

Proteomics-Driven Biomarkers in Pancreatic Cancer.

Pancreatic cancer is a devastating disease that has a grim prognosis, highlighting the need for improved screening, diagnosis, and treatment strategies. Currently, the sole biomarker for pancreatic ductal adenocarcinoma (PDAC) authorized by the U.S. Food and Drug Administration is CA 19-9, which proves to be the most beneficial in tracking treatment response rather than in early detection. In recent years, proteomics has emerged as a powerful tool for advancing our understanding of pancreatic cancer biology and identifying potential biomarkers and therapeutic targets. This review aims to offer a comprehensive survey of proteomics’ current status in pancreatic cancer research, specifically accentuating its applications and its potential to drastically enhance screening, diagnosis, and treatment response. With respect to screening and diagnostic precision, proteomics carries the capacity to augment the sensitivity and specificity of extant screening and diagnostic methodologies. Nonetheless, more research is imperative for validating potential biomarkers and establishing standard procedures for sample preparation and data analysis. Furthermore, proteomics presents opportunities for unveiling new biomarkers and therapeutic targets, as well as fostering the development of personalized treatment strategies based on protein expression patterns associated with treatment response. In conclusion, proteomics holds great promise for advancing our understanding of pancreatic cancer biology and improving patient outcomes. It is essential to maintain momentum in investment and innovation in this arena to unearth more groundbreaking discoveries and transmute them into practical diagnostic and therapeutic strategies in the clinical context.

PANCREATIC CANCER, PROTEOMICS, BIOMARKER, CHEMOTHERAPY EFFECTIVENESS

Manuscript published on Mutation Research/Genetic Toxicology and Environmental Mutagenesis, August 2023

Blood molecular profile to predict genotoxicity from exposure to antineoplastic drugs.

Genotoxicity is an important information that should be included in human biomonitoring programmes. However, the usually applied cytogenetic assays are laborious and time-consuming, reason why it is critical to develop rapid and economic new methods. The aim of this study was to evaluate if the molecular profile of frozen whole blood, acquired by Fourier Transform Infrared (FTIR) spectroscopy, allows to assess genotoxicity in occupational exposure to antineoplastic drugs, as obtained by the cytokinesis-block micronucleus assay. For that purpose, 92 samples of peripheral blood were studied: 46 samples from hospital professionals occupationally exposed to antineoplastic drugs and 46 samples from workers in academia without exposure (controls). It was first evaluated the metabolome from frozen whole blood by methanol precipitation of macromolecules as haemoglobin, followed by centrifugation. The metabolome molecular profile resulted in 3 ratios of spectral bands, significantly different between the exposed and non-exposed group (p < 0.01) and a spectral principal component-linear discriminant analysis (PCA-LDA) model enabling to predict genotoxicity from exposure with 73 % accuracy. After optimization of the dilution degree and solution used, it was possible to obtain a higher number of significant ratios of spectral bands, i.e., 10 ratios significantly different (p < 0.001), highlighting the high sensitivity and specificity of the method. Indeed, the PCA-LDA model, based on the molecular profile of whole blood, enabled to predict genotoxicity from the exposure with an accuracy, sensitivity, and specificity of 92 %, 93 % and 91 %, respectively. All these parameters were achieved based on 1 μL of frozen whole blood, in a high-throughput mode, i.e., based on the simultaneous analysis of 92 samples, in a simple and economic mode. In summary, it can be conclude that this method presents a very promising potential for high-dimension screening of exposure to genotoxic substances.

MOLECULAR PROFILE, FTIR-SPECTROSCOPY, GENOTOXICITY, CYTOKINESIS-BLOCKED MICRONUCLEUS ASSAY, FROZEN BLOOD, ANTINEOPLASTICS

Manuscript published on ACS Omega, June 2023

The Impact of the Serum Extraction Protocol on Metabolomic Profiling Using UPLC-MS/MS and FTIR Spectroscopy.

Biofluid metabolomics is a very appealing tool to increase the knowledge associated with pathophysiological mechanisms leading to better and new therapies and biomarkers for disease diagnosis and prognosis. However, due to the complex process of metabolome analysis, including the metabolome isolation method and the platform used to analyze it, there are diverse factors that affect metabolomics output. In the present work, the impact of two protocols to extract the serum metabolome, one using methanol and another using a mixture of methanol, acetonitrile, and water, was evaluated. The metabolome was analyzed by ultraperformance liquid chromatography associated with tandem mass spectrometry (UPLC-MS/MS), based on reverse-phase and hydrophobic chromatographic separations, and Fourier transform infrared (FTIR) spectroscopy. The two extraction protocols of the metabolome were compared over the analytical platforms (UPLC-MS/MS and FTIR spectroscopy) concerning the number of features, the type of features, common features, and the reproducibility of extraction replicas and analytical replicas. The ability of the extraction protocols to predict the survivability of critically ill patients hospitalized at an intensive care unit was also evaluated. The FTIR spectroscopy platform was compared to the UPLC-MS/MS platform and, despite not identifying metabolites and consequently not contributing as much as UPLC-MS/MS in terms of information concerning metabolic information, it enabled the comparison of the two extraction protocols as well as the development of very good predictive models of patient’s survivability, such as the UPLC-MS/MS platform. Furthermore, FTIR spectroscopy is based on much simpler procedures and is rapid, economic, and applicable in the high-throughput mode, i.e., enabling the simultaneous analysis of hundreds of samples in the microliter range in a couple of hours. Therefore, FTIR spectroscopy represents a very interesting complementary technique not only to optimize processes as the metabolome isolation but also for obtaining biomarkers such as those for disease prognosis.

Manuscript published on Vibrational Spectroscopy, April 2023

Label-free discrimination of T and B lymphocyte activation based on vibrational spectroscopy – A machine learning approach.

B and T-lymphocytes are major players of the specific immune system, responsible by an efficient response to target antigens. Despite the high relevance of these cells’ activation in diverse human pathophysiological processes, its analysis in clinical context presents diverse constraints. In the present work, MIR spectroscopy was used to acquire the cells molecular profile in a label-free, simple, rapid, economic, and high-throughput mode. Recurring to machine learning algorithms MIR data was subsequently evaluated. Models were developed based on specific spectral bands as selected by Gini index and the Fast Correlation Based Filter. To determine if it was, possible to predict from the spectra, if B and T lymphocyte were activated, and what was the molecular fingerprint of T- or B- lymphocyte activation. The molecular composition of activated lymphocytes was so different from naïve cells, that very good prediction models were developed with whole spectra (with AUC=0.98). Activated B lymphocytes also present a very distinct molecular profile in relation to activated T lymphocytes, leading to excellent prediction models, especially if based on target bands (AUC=0.99). The identification of critical target bands, according to the metabolic differences between B and T lymphocytes and in association with the molecular mechanism of the activation process highlighted bands associated to lipids and glycogen levels. The method developed presents therefore, appealing characteristics to promote a new diagnostic tool to analyze and discriminate B from T-lymphocytes.

Manuscript published on Proteomes, July 2022

Proteomics for Biomarker Discovery for Diagnosis and Prognosis of Kidney Transplantation Rejection.

Renal transplantation is currently the treatment of choice for end-stage kidney disease, enabling a quality of life superior to dialysis. Despite this, all transplanted patients are at risk of allograft rejection processes. The gold-standard diagnosis of graft rejection, based on histological analysis of kidney biopsy, is prone to sampling errors and carries high costs and risks associated with such invasive procedures. Furthermore, the routine clinical monitoring, based on urine volume, proteinuria, and serum creatinine, usually only detects alterations after graft histologic damage and does not differentiate between the diverse etiologies. Therefore, there is an urgent need for new biomarkers enabling to predict, with high sensitivity and specificity, the rejection processes and the underlying mechanisms obtained from minimally invasive procedures to be implemented in routine clinical surveillance. These new biomarkers should also detect the rejection processes as early as possible, ideally before the 78 clinical outputs, while enabling balanced immunotherapy in order to minimize rejections and reducing the high toxicities associated with these drugs. Proteomics of biofluids, collected through non-invasive or minimally invasive analysis, e.g., blood or urine, present inherent characteristics that may provide biomarker candidates. The current manuscript reviews biofluids proteomics toward biomarkers discovery that specifically identify subclinical, acute, and chronic immune rejection processes while allowing for the discrimination between cell-mediated or antibody-mediated processes. In time, these biomarkers will lead to patient risk stratification, monitoring, and personalized and more efficient immunotherapies toward higher graft survival and patient quality of life.

KIDNEY ALLOGRAFT, PROTEOMICS, BIOMARKER, REJECTION, BIOFLUIDS, EXOSOMES

Manuscript published on the International Journal of Oral and Maxillofacial Surgery, August 2021

Surgical complications related to temporomandibular joint arthroscopy: a prospective analysis of 39 single-portal versus 43 double-portal procedures.

Temporomandibular joint (TMJ) arthroscopy is a minimally invasive surgical procedure proposed for diverse TMJ intra-articular disorders. A prospective study was designed with the aim of investigating intraoperative and postoperative surgical complications for single and double-portal TMJ arthroscopy. All interventions were performed by one surgeon with the same surgical protocol. A total of 55 patients were enrolled, resulting in 82 TMJ arthroscopies (28 unilateral and 27 bilateral). A total of 39 single portal (47.57%) and 43 double-portal (52.43%) arthroscopies were performed. No severe and irreversible complications were observed. Most complications were resolved after 4 weeks. Double-portal was associated with more complications (n = 23) compared with single-portal TMJ arthroscopy (n = 14), with a statistically significant difference found between single and double-portal TMJ arthroscopy in two intraoperative complications: intra-articular bleeding (P = 0.044) and oedema of the preauricular area (P = 0.042). This study confirms the safety of TMJ arthroscopy for single and double-portal procedures, with the authors suggesting a multicentre study, in an effort to minimize any possible bias.

Manuscript published on Vibrational Spectroscopy, October 2020

Discriminating B and T-lymphocyte from its molecular profile acquired in a label-free and high-throughput method.

B and T-lymphocytes are one of the main players of the adaptive immune system and consequently, the ability to discriminate these cells is relevant in diverse pathophysiological cases. In the present work, FTIR spectroscopy was used to acquire the cells molecular profile in a label-free, simple, rapid, economic, and high-throughput mode. It was possible, by hierarchical cluster analysis of the second derivative spectra, between 900 to 1500 cm−1, to correctly classify 98 % of samples as B and T-lymphocyte. It was also possible to develop a very good support vector machine model, which could predict B and T-lymphocyte from the second derivative spectra, with high accuracy (0.95), precision (0.95), recall (0.95), and F1 score (0.95). The method developed presents therefore, appealing characteristics to promote a new diagnostic tool to analyze and discriminate B from T-lymphocytes.

Conference: 28th LISBON International Conference on Science, Engineering, Technology and Healthcare (SETH-24) December 16-18, 2024 Lisbon (Portugal)

Survival Prediction in Critically Ill Patients based on the Serum Molecular Fingerprint.

It is relevant to discover biomarkers enabling to predict critically ill patients’ survival. This study focused on 45 patients, from which 22 deceased and 23 were discharged from an Intensive Care Unit (ICU). It was considered the serum molecular fingerprint, as acquired by Fourier Transform Infra-Red (FTIR) spectroscopy, obtained 3 days before the patients discharged or death at the ICU. It was possible to obtain ratios of bands of the sera spectra, statistically different between the two groups of patients. Furthermore, good Naïve Bayes models were developed based on the second derivative spectra enabling an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.77. These promising outputs suggest further investigation with a larger cohort.

INTENSIVE CARE UNIT, SURVIVAL, BIOMARKERS

Conference: 4th ROME International Conference on Challenges in Engineering, Medical, Economics and Education: Research & Solutions (CEMEERS-24b) December 4-6, 2024 Rome (Italy)

Infection Biomarkers at Intensive Care Units.

It is relevant to discover infection biomarkers, especially for critically ill patients in intensive care units (ICU), as these patients often present non-infectious inflammatory processes that obscure typical infectious markers. This study focused on 20 ICU patients, half of whom had acquired bacterial blood infections (bacteremia). Due to the significance of inflammatory processes in these patients, it was evaluated how 21 serum cytokines could be used to develop predictive models for bacteremia. Feature selection using a Gain Information algorithm allowed for the construction of an excellent Naïve Bayes model, achieving an AUC of 0.950. These promising results strongly support future studies with larger cohorts, to further evaluate these types of platforms for infection diagnosis in such critical populations.

INFECTION BIOMARKERS, INTENSIVE CARE UNIT, CYTOKINES

Conference: 4th ROME International Conference on Challenges in Engineering, Medical, Economics and Education: Research & Solutions (CEMEERS-24b) December 4-6, 2024 Rome (Italy)

Comparison of the Serum Whole Molecular Composition with the Serum Metabolome to Acquire the Pathophysiological State.

Omics Sciences serve as an essential tool to advance precision medicine. Since conventional omics sciences rely on laborious, complex and time-consuming analytical processes, this study evaluated whether the serum molecular fingerprint, captured by FTIR spectroscopy, could predict mortality risk in critically ill patients. Both the whole serum and the serum metabolome (i.e., serum after removal of macromolecules) were analyzed. PCA-LDA models demonstrated strong performance in predicting patients’ pathophysiological state. A significantly more accurate model for predicting the patients’ pathophysiological state was achieved using the serum metabolome (94%) compared to the whole serum (81%). This is consistent with metabolomics, which provides a more direct view of the systems’ functionality. These promising results highlight the importance of FTIR spectroscopy analysis of the serum metabolome, offering a rapid, cost-effective, and high-throughput method for assessing patients' pathophysiological state.

BIOMARKERS, FTIR SPECTROSCOPY, METABOLOME, INTENSIVE CARE UNIT

Conference: 4th ROME International Conference on Challenges in Engineering, Medical, Economics and Education: Research & Solutions (CEMEERS-24b) December 4-6, 2024 Rome (Italy)

Trends in COVID-19 Patient Characteristics and Mortality Throughout the Pandemic: Insights from a Portuguese Single Centre Study.

As SARS-CoV-2 continues to circulate globally and new variants emerge, it remains relevant to gather data on the affected patients’ clinical characteristics and outcomes to understand how individual factors and public health measures affect prognosis. Thus, we analyzed data of 870 ICU patients admitted for COVID-19 across two distinct phases of the pandemic: before and after the introduction of immunization. Experimental results showed that vaccination significantly impacted patient demographics after the third wave, and that waves number two and three, dominated by the EU1 and Alpha variants, had higher mortality. Older age, the need for invasive mechanical ventilation, and hematologic cancer were significantly associated with an increased risk of death in the adjusted multivariable model (AUC: 0.778, 95% CI 0.746-0.810, p<0.001). As the pandemic progressed, while some public health interventions influenced the observed trends, individual patient characteristics had a more substantial impact on their outcome.

COVID-19 WAVES, COVID-19 VACCINATION, INTENSIVE CARE UNIT, MORTALITY

Conference: 4th ROME International Conference on Challenges in Engineering, Medical, Economics and Education: Research & Solutions (CEMEERS-24b) December 4-6, 2024 Rome (Italy)

Predicting Critically Ill Patients’ Outcome in the ICU using UHPLC-HRMS Data.

The available scores to predict patients’ outcomes in specific settings generally present low sensitivities and specificities when applied to intensive care units’ (ICUs) populations. Advancements in analytical techniques, notably Ultra-High Performance Liquid Chromatography- Mass Spectrometry (UHPLC-HRMS) transformed biomarker identification, enabling a comprehensive profiling of biofluids, including serum. In the current work, untargeted metabolomics, utilizing UHPLC-HRMS serum analysis, was performed on 16 ICU patients, categorized as either discharged (n=8), or deceased (n=8) in average seven days post sample collection. Linear discriminant analysis (LDA) or principal component analysis (PCA)-LDA models involving different metabolite sets were developed, enabling to predict patients’ outcomes in the ICU with 92% accuracy and 83% sensitivity on validation datasets. These results highlight the advantages of UHPLC-HRMS as a platform capable of providing a set of clinically significant biomarkers to predict patients’ outcome.

BIOMARKERS, INTENSIVE CARE UNIT, PREDICTIVE MODELS, METABOLOMICS, MASS SPECTROMETRY

Conference: 37th European Immunogenetics & Histocompatibility Conference (EFI) May 20-23, 2024 Geneva (Switzerland)

Immunogenetic profiling in living donor kidney transplantation: Insights from DSAs and pronase-treated flow cytometry crossmatch.

Highly sensitized patients have very limited access to kidney transplantation, resulting in a long waiting time on dialysis. Delisting of prohibited HLA antigens should allow the performance of DSA-positive, crossmatch-negative transplants. We describe the experience of four Spanish transplant centers after delisting prohibited HLA antigens to reduce cPRA below 99%. Delisting was gradually performed, allowing HLA antigens with MFI < 5000, avoiding repeated antigens from previous transplants. If the cPRA did not decrease, a more aggressive HLA delisting was performed up to a maximum MFI of 10,000 except for DP. In some cases, the capacity to activate complement (C3d or C1q) and the 1/16 dilution were performed to individually decide the delisting. Forty-eight patients underwent delisting from May 2022 to August 2023, with total time on the waiting list of 5.6 [3.3–9.1] years and time on dialysis of 9.8 [5.7–13.5] years. Baseline cPRA was 100.0 [99.9–100.0]%. After delisting, it dropped to 98.3 [96.0–99.0]%. Thirty patients obtained an offer within the Spanish Highly Sensitized Program (PATHI) after a period of 98 [52–154] days of which 18 had a negative CDC and flow cytometry crossmatch and underwent kidney transplant. The number of DSAs at the time of transplant was 2 [1–4], with MFI of the dominant DSA of 8036 [3857–20,951]. 55.6% of recipients received post-transplant desensitization. Rejection developed in 7 patients (38.9%), in all cases humoral and in 2 cases (11.1%) mixed, after 43 [13–91] days post-transplant. In only two cases rejection could not be controlled with treatment and, in one case, it progressed to chronic antibody-mediated rejection. All grafts, except one, are functional at 186 [67–384] days post-transplant and, in 7 patients, at 1 year of follow-up. A delisting strategy can be considered for hypersensitized patients who have no other options to find a compatible donor on the waiting list.

Conference: 37th European Immunogenetics & Histocompatibility Conference (EFI) May 20-23, 2024 Geneva (Switzerland)

Deciphering HLA antibody reactivity patterns: A cluster-based analysis of SAB assay data.

The aim of this study was the estimation of the effect of long-term kidney transplantation on T and B regulatory lymphocytes. Patients who had received a transplant 17+ years ago (group A) and 1 year ago (group B) took part in the study. The subpopulations of CD4 + CD25 + FOXP3 (Tregs), CD19 + CD24++CD38++ (Bregs1), CD19 + CD27 + CD24++ (Bregs2) and naïve, switched, non-switched memory and double negative (DN) B-lymphocytes were measured by flow cytometry. Patients of group A (N = 20, M/F = 11/9) did not differ with patients of group B (N = 40, M/F = 23/17) regarding age (57 vs 56 years, p = 0.065) and eGFR (63.5 ± 19.2 vs 63.6 ± 17, 1 mL/min/1,73 m², p = 0.132). Significant differences between groups were observed in respect of the percentage of Tregs (2.55 vs 4.7%, p < 0.001), the percentage and number of Bregs1 (1.6 vs 0.0%, p < 0.001 and 0.53 vs 0.0/μL, p = 0.001) and Bregs2 (0.0 vs 2.55%, p < 0.001 and 0.0 vs 2.0/μL, p = 0.001). Group A also exhibited a significant reduction of B-lymphocytes (55 vs 99/μL, p = 0.034), which mainly referred to naïve (18 vs 41/μL, p = 0.005), switched (7.26 vs 14/μL, p = 0.005), and non-switched memory (1.22 vs 13/μL, p < 0.001) B-cells, with a simultaneous increase of DN cells (22 vs 11/μL, p = 0.028). Long-term kidney transplant recipients show a reduction of the percentage of Tregs, a rise of Bregs1 and a decline of Bregs2 as well as significant alterations of B-lymphocyte subpopulations.

Conference: 2023 IEEE 7th Portuguese Meeting on Bioengineering (ENBENG), June 22-23, 2023 Lisbon (Portugal)

Alternative Serum Biomarkers of Bacteraemia for Intensive Care Unit Patients.

The diagnosis of infections in hospital or clinical settings usually involves a series of time-consuming steps, including biological sample collection, culture growth of the organism isolation and subsequent characterization. For this, there are diverse infection biomarkers based on blood analysis, however, these are of limited use in patients presenting confound processes as inflammatory process as occurring at intensive care units. In this preliminary study, the application of serum analysis by FTIR spectroscopy, to predict bacteraemia in 102 critically ill patients in an ICU was evaluated. It was analysed the effect of spectra pre-processing methods and spectral sub-regions on t-distributed stochastic neighbour embedding. By optimizing Support Vector Machine (SVM) models, based on normalised second derivative spectra of a smaller subregion, it was possible to achieve a good bacteraemia predictive model with a sensitivity and specificity of 76%. Since FTIR spectra of serum is acquired in a simple, economic and rapid mode, the technique presents the potential to be a cost-effective methodology of bacteraemia identification, with special relevance in critically ill patients, where a rapid infection diagnostic will allow to avoid the unnecessary use of antibiotics, which ultimately will ease the load on already fragile patients' metabolism. Read more

FTIR SPECTROSCOPY, INFECTION, BIOMARKERS, INTENSIVE CARE UNIT

Conference: 2023 IEEE 7th Portuguese Meeting on Bioengineering (ENBENG), June 22-23, 2023 Lisbon (Portugal)

Laboratory biomarkers associated to death in the first three COVID-19 waves in Portugal.

Besides the pandemic being over, new SARS-CoV-2 lineages, and sub-lineages, still pose risks to global health. Thus, in this preliminary study, to better understand the characteristics of COVID-19 patients and the effect of certain hematologic biomarkers on their outcome, we analyzed data from 337 patients admitted to the ICU of a single-center hospital in Lisbon, Portugal, in the first three waves of the pandemic. Most patients belonged to the second (40.4%) and third (41.2%) waves. The ones from the first wave were significantly older and relied more on respiratory techniques like invasive mechanic ventilation and extracorporeal membrane oxygenation. There were no significant differences between waves regarding mortality in the ICU. In general, non-survivors had worse laboratory results. Biomarkers significantly associated with death changed depending on the waves. Increased high-sensitivity cardiac troponin I results, and lower eosinophil counts were associated to death in all waves. In the second and third waves, the international normalized ratio, lymphocyte counts, and neutrophil counts were also associated to mortality. A higher risk of death was linked to increased myoglobin results in the first two waves, as well as increased creatine kinase results, and lower platelet counts in the third wave.

COVID-19, WAVES, BIOMARKERS, ICU, RISK OF DEATH

Conference: 2023 IEEE 7th Portuguese Meeting on Bioengineering (ENBENG), June 22-23, 2023 Lisbon (Portugal)

Comparison of Analytical Methods Of Serum Untargeted Metabolomics.

Metabolomics has emerged as a powerful tool in the discovery of new biomarkers for medical diagnosis and prognosis. However, there are numerous challenges, such as the methods used to characterize the system metabolome. In the present work, the comparison of two analytical platforms to acquire the serum metabolome of critically ill patients was conducted. The untargeted serum metabolome analysis by ultraperformance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS) enabled to identify a set of metabolites statistically different between deceased and discharged patients. This set of metabolites also enabled to develop a very good predictive model, based on linear discriminant analysis (LDA) with a sensitivity and specificity of 80% and 100%, respectively. Fourier Transform Infrared (FTIR) spectroscopy was also applied in a high-throughput, simple and rapid mode to analyze the serum metabolome. Despite this technique not enabling the identification of metabolites, it allowed to identify molecular fingerprints associated to each patient group, while leading to a good predictive model, based on principal component analysis-LDA, with a sensitivity and specificity of 100% and 90%, respectively. Therefore, both analytical techniques presented complementary characteristics, that should be further explored for metabolome characterization and application as for biomarkers discovery for medical diagnosis and prognosis.

METABOLOMICS, MASS SPECTROMETRY, LIQUID CHROMATOGRAPHY, FOURIER TRANSFORM INFRARED SPECTROSCOPY

Conference: 2019 IEEE 6th Portuguese Meeting on Bioengineering (ENBENG), February 22-23, 2019 Lisbon (Portugal)

Effect of consumption of green tea extracts on the plasma molecular signature.

Epigallocatechin-3-gallate (EGCG), the major catechin present in green tea, displays diverse biological activities as anti-oxidation, anti-inflammation, anti-proliferation and anti-microbial among others. In the present work it was evaluated the effect of the consumption of EGCG along 90 days on healthy human volunteers (n=30) on plasma molecular signature acquired by mid-infrared (MIR) spectroscopy. It was observed by principal component analysis of spectra that plasma samples presented a significant different molecular profile after 90 days of EGCG consumption. Based on the corresponding loading vector, it was observed that EGCG consumption affected the profile of the major molecules as proteins and lipids. Were identified diverse ratios of spectral bands statistically different (p <; 0.01) after EGCG consumption, according to a high impact of EGCG on the general metabolism. MIR spectroscopy enabled to acquire the plasma whole molecular signature in a highly sensitive and specific mode. Since the MIR spectra is also acquired in an economic, simple, fast and high-throughput mode, the technique presents promising characteristics to acquire information in large-scale epidemiological studies towards a better understand of the in vivo effect of EGCG.

GREEN TEA, EPIGALLOCATECHIN, POLYPHENOLS, METABOLOMICS

Conference: 10th Edition Forum of Chemical & Biological Engineering, May 7-10, 2023 Lisbon (Portugal)

Predicting Septic Patient Outcomes Based on Serum Cytokines.

Background: Cytokine storms significantly influence systemic inflammatory responses and septic patient outcome, highlighting the critical role of cytokine profiling in understanding and predicting patient’s outcome [1], which holds significant potential for future clinical and therapeutic applications. The present work aims to evaluate how serum cytokines can predict the outcome of septic patients at intensive care units (ICU). Methods: Serum samples from 16 septic patients (7 discharged, 9 deceased) were analyzed using a Milliplex 384-Well High Sensitivity Human T Cell Magnetic Bead Panel to profile a comprehensive set of 21 cytokines: ITAC, GM-CSF, Fractalkine, IFN-g, IL-10, MIP-3a, IL-12p70, IL-13, IL-17a, IL-1b, IL-2, IL-21, IL-4, IL-23, IL-5, IL-6, IL-7, IL-8, MIP-1a, MIP-1b, and TNF-a. Machine learning tools coupled to a Gini decrease ranking method were employed to identify cytokines critical for discriminating between deceased and discharged patients. Results: Preliminary findings indicate that IL-6, MIP-3a, IL-8, IL-12p70, and IL-10 are notably linked to sepsis outcomes. The relevance of these cytokines aligns with existing literature on their roles in inflammation and immune responses [2]. Predictive models constructed using the Naive Bayes models showed promising results, with sensitivity and specificity exceeding 86%. Conclusion: This study underscores the importance of cytokine profiling in sepsis, highlighting key biomarkers for predicting patient outcomes.

SEPSIS, SERUM CYTOKINES, BIOMARKERS, ICU MORTALITY

Conference: 14th Edition International Chemical and Biological Engineering Conference (CHEMPOR), September 12-15, 2023 Bragança (Portugal)

Streamlining Bacterial Infection Diagnosis: Rapid Gram Classification Using FTIR Spectroscopy.

In a hospital setting, diagnosing infections typically involves a complex process that includes the collection of biological samples and growing a culture for organism isolation, followed by its characterization. However, these methods are slow, require multiple steps and are often limited by the need of specialized equipment and skilled personnel. In this preliminary study, it was analysed the serum, by FTIR spectroscopy, of 29 critically ill COVID-19 patients in an ICU. It was analysed the effect of varied pre-processing methods and spectral sub-regions on t-SNE. Through the optimization of SVM models, it was possible to achieve a very good gram predictive model with a sensitivity and specificity of 90 and 89% respectively. As an accurate classification of bacterial strains is crucial to guide effective antimicrobial therapy and prevent the spread of multidrug-resistant bacteria, FTIR spectra, acquired in a simple, economic, and rapid mode, presents therefore the potential for development of new classification methods that would greatly enhance the ability to manage bacterial infections.

Conference: 12th International Conference on Advanced Vibrational Spectroscopy (ICAVS12), August 27-September 1, 2023 Kraków (Poland)

Impact of COVID-19 on critically ill patients’ - mortality prediction models based on serum FTIR-spectra.

Due to the relevance of mortality prediction of critically ill patients, it is common practice at intensive care units (ICU) to use physiological scores, e.g., APACHE II. However, these type of scorings don’t enable to predict individual patients’ outcome, mostly used for comparing groups of patients and ICUs [1]. It is therefore relevant to discover robust biomarkers of mortality prediction at ICU. Since FTIR spectroscopy can capture the whole molecular fingerprint of a system in a very specific and sensitive mode [2,3], in the present work, diverse mortality predictive models of support vector machines (SVM), based on FTIR-spectra of serum of critically ill patients, were developed. Serum samples from 200 patients at an ICU, with half presenting COVID-19 and the other half not presenting this infection, were considered. SVM models were optimized by combining spectral regions with diverse spectra pre-processing methods. A model cross-validation strategy, based on 10 random iterations, with 80% of samples for training and 20% for validation, was implemented. It was possible to develop very good SVM models to predict mortality based only on patients without COVID-19 (AUC=0.90), and a slightly better model was achiever for patients with COVID-19 (AUC=0.93). This difference can result from COVID-19 patients presenting a different metabolic status in relation to non-COVID-19 patients. Indeed, a very good SVM model enabled to discriminate these two populations (AUC=0.83). When considering the mixed population (i.e., with and without COVID-19), a slightly worse predictive SVM model was obtained (AUC=0.88).

FTIR, COVID-19, MORTALITY, ICU, BIOMARKERS

Conference: 12th International Conference on Advanced Vibrational Spectroscopy (ICAVS12), August 27-September 1, 2023 Kraków (Poland)

Impact Of Serum Metabolome Isolation Process On Models’ Prediction Of Critically Ill Patients’ Mortality.

Metabolomics has emerged as a powerful tool in the discovery of new biomarkers for medical diagnosis and prognosis. Metabolomics of biofluids, such as serum, can therefore potentially deliver biomarkers that may be applicable in patients’ monitoring [1]. This is especially relevant in the management of critically ill patients. However, there are numerous challenges, including the metabolome isolation process and the subsequent platform applied to analyze it. FTIR-spectroscopy presents diverse advantages for metabolome analysis, since it may be applied in rapid, economic, and high-throughput mode, while enabling to acquire the system’s metabolic status with a high sensitivity and specificity [2,3]. In the current project, two extraction protocols of the serum metabolome were evaluated. Both protocols included macromolecules precipitation induced by mixtures of methanol, acetonitrile, and water. Replicas of 5L extracted serum metabolome, from critically ill patients, were plated in 384 wells-microplates, and after a rapid dehydration step, spectra were acquired between 400 to 4000 cm-1. The impact of the two extraction procedures to isolate the serum metabolome, on reproducibility, based on FTIR-spectra principal component analysis (PCA) was studied. The impact of the two extraction procedures on the performance of predictive models, based on spectra PCA-discriminant analysis of patients’ mortality, was also conducted. Serum samples from critically ill patients were obtained according to legal and ethics requirements, including project ethics approval by the Hospital Ethics Committee (Centro Hospitalar Universitário Lisboa Central), and patients’ informed consent.

Conference: IV Jornadas de Ortoprotesia da ESTeSL, June 2, 2018 Lisbon (Portugal)

Comparison of outcomes between the DEKA Arm and conventional prothesis - an in depth review study of measures and evaluation techniques.

The ability to measure and evaluate [1] is a major exercise in the daily lives of health professionals, regardless of their speciality, may they be physiotherapists, nurses, doctors, prosthetist and orthotisc and of biomedical engineers, who assist in the whole process of gathering the data and analysing it. This exercise must be realized in a professional, responsible and economic manner and ultimately based on backed-up evidence. Therefore it is of extreme importance to focus on a correct and accurate measurement system and be able to properly evaluate a persons’s illness or disability. To understand the results of the study case that was reviewed [2], it becomes important to understand two important and different definitions: outcome and outcome measure. A database of a few of the measures available can be consulted online. [3] [4]. An outcome is a “measurable individual, family, or community state, behavior or perception that is measured along a continuum and is responsive to clinical interventions” [5]. An outcome measure is a set of items that are used to create scores that are “intended to quantify a patient’s performance or health status based on standardized evaluation protocols or closed ended questions”. [5] It is therefore important to make sure that these patient characteristics are measured using standardized outcome instruments so that they can be adequately stored in electronic records, allowing for its use by clinical bodies, to facilitate in the identification of signal or symptoms of any healthy condition, making the diagnostic and call for action easier, all the while contributing to the development of clinical knowledge and professional education. [6] More detailed information about outcome measurements can be consulted in [7].

Conference: NOVA Summer Medical School: Lifestyle, July 3, 2025 Lisbon (Portugal)

Big Data and Artificial Intelligence in Healthcare.

At the NOVA Summer Medical School: Lifestyle (Lisbon, July 3, 2025), an oral presentation addressed the role of Big Data and Artificial Intelligence in healthcare. The session introduced core concepts of machine learning, programming, and omics, highlighting their integration into diagnostics and clinical decision-making. The discussion emphasized both the transformative potential of these technologies and their relevance for driving future innovation in medical practice.

Conference: XIV Encontro de Investigadores da Qualidade, June 28, 2024 Lisbon (Portugal)

Session moderator on "Techniques and Optimization Methodologies".

The XIV Meeting of Quality Researchers (RIQUAL), held at Instituto Superior de Engenharia de Lisboa on June 28, 2024, focused on advancing knowledge and practices in quality across diverse domains. The event covered themes such as accreditation, certification, environmental and safety management, social responsibility, sustainability, health, education, and information systems. A special session was dedicated to showcasing student research projects, dissertations, and theses aligned with these topics.

Conference: 4th Comprehensive Health Research Centre (CHRC) Annual Summit, May 25-26, 2023 Évora (Portugal)

Fungal and Bacterial Infections discrimination in ICU patients based on serum molecular fingerprint.

The rapid discrimination between fungal and bacterial infections in intensive care unit (ICU) patients is crucial for an efficient antimicrobial therapy. This work aims to develop a predictive model capable of discriminating fungal and bacterial infections among these patients based on a rapid serum analysis. Serum from 46 patients at ICU, with COVID-19, were analyzed by Fourier-Transform infrared (FTIR) spectroscopy: 17 did not present any other infection; 12 presented bacterial infections (including Enterobacter aerogenes (2), Escherichia coli (4), Enterococcus faecalis (1), Haemophilus influenzae (1) and Serratia marcescens (1), Klebsiella pneumoniae (3)), isolated from blood, tracheobronchial aspiration, bronchoalveolar lavage and urine; 17 with fungal infections (including Candida albicans (14), C. tropicalis (1), C. parapsilosis (1) and C. glabrata (1), isolated from bronchoalveolar lavage, urine and urethral exudate. The impact of a Fast Correlation Based Filter (FCBF) of normalized second derivatized spectra (between 406-1800 and 2800-3992 cm-1) was evaluated on a t-distributed Stochastic Neighbor Embedding (t-SNE), and a Naïve-Bayes model. The predictive models were based on 10 random iterations, with 80% of samples used for training and the remaining 20% for validation. The impact of FCBF was evaluated on t-SNE to select the spectral bands with high significance in detecting the infection and in discriminating the bacterial from the fungus infection. Optimized Naïve-Bayes models enabled to detect the infection with a sensitivity of 83% and a specificity of 81%, and the discrimination of the bacterial and fungal infection with a sensitivity and specificity of 86% and 96%, respectively. Serum analysis, based on FTIR spectroscopy associated to machine learning algorithms presents a high potential to detect in a rapid and economic mode either the infection either the discrimination between bacterial or fungal infection.

FTIR SPECTROSCOPY, INTENSIVE CARE UNIT, FUNGAL INFECTION, BACTERIAL INFECTION, DIAGNOSTICS