Improving Lung Cancer Screening Selection: The HUNT Lung Cancer Risk Model for Ever-Smokers Versus the NELSON and 2021 United States Preventive Services Task Force Criteria in the Cohort of Norway: A Population-Based Prospective Study.

Background: Improving the method for selecting participants for lung cancer (LC) screening is an urgent need. Here, we compared the performance of the Helseundersøkelsen i Nord-Trøndelag (HUNT) Lung Cancer Model (HUNT LCM) versus the Dutch-Belgian lung cancer screening trial (Nederlands-Leuvens Longkanker Screenings Onderzoek (NELSON)) and 2021 United States Preventive Services Task Force (USPSTF) criteria regarding LC risk prediction and efficiency. Methods: We used linked data from 10 Norwegian prospective population-based cohorts, Cohort of Norway. The study included 44,831 ever-smokers, of which 686 (1.5%) patients developed LC; the median follow-up time was 11.6 years (0.01-20.8 years). Results: Within 6 years, 222 (0.5%) individuals developed LC. The NELSON and 2021 USPSTF criteria predicted 37.4% and 59.5% of the LC cases, respectively. By considering the same number of individuals as the NELSON and 2021 USPSTF criteria selected, the HUNT LCM increased the LC prediction rate by 41.0% and 12.1%, respectively. The HUNT LCM significantly increased sensitivity (p < 0.001 and p = 0.028), and reduced the number needed to predict one LC case (29 versus 40, p < 0.001 and 36 versus 40, p = 0.02), respectively. Applying the HUNT LCM 6-year 0.98% risk score as a cutoff (14.0% of ever-smokers) predicted 70.7% of all LC, increasing LC prediction rate with 89.2% and 18.9% versus the NELSON and 2021 USPSTF, respectively (both p < 0.001). Conclusions: The HUNT LCM was significantly more efficient than the NELSON and 2021 USPSTF criteria, improving the prediction of LC diagnosis, and may be used as a validated clinical tool for screening selection.

Long-term effects of myo-inositol on traumatic brain injury: Epigenomic and transcriptomic studies.

Background and purpose: Traumatic brain injury (TBI) and its consequences remain great challenges for neurology. Consequences of TBI are associated with various alterations in the brain but little is known about long-term changes of epigenetic DNA methylation patterns. Moreover, nothing is known about potential treatments that can alter these epigenetic changes in beneficial ways. Therefore, we have examined myo-inositol (MI), which has positive effects on several pathological conditions. Methods: TBI was induced in mice by controlled cortical impact (CCI). One group of CCI animals received saline injections for two months (TBI+SAL), another CCI group received MI treatment (TBI+MI) for the same period and one group served as a sham-operated control. Mice were sacrificed 4 months after CCI and changes in DNA methylome and transcriptomes were examined. Results: For the first time we: (i) provide comprehensive map of long-term DNA methylation changes after CCI in the hippocampus; (ii) identify differences by methylation sites between the groups; (iii) characterize transcriptome changes; (iv) provide association between DNA methylation sites and gene expression. MI treatment is linked with upregulation of genes covering 33 biological processes, involved in immune response and inflammation. In support of these findings, we have shown that expression of BATF2, a transcription factor involved in immune-regulatory networks, is upregulated in the hippocampus of the TBI+MI group where the BATF2 gene is demethylated. Conclusion: TBI is followed by long-term epigenetic and transcriptomic changes in hippocampus. MI treatment has a significant effect on these processes by modulation of immune response and biological pathways of inflammation.

Src-NADH dehydrogenase subunit 2 complex and recognition memory of imprinting in domestic chicks.

Src is a non-receptor tyrosine kinase participating in a range of neuronal processes, including synaptic plasticity. We have recently shown that the amounts of total Src and its two phosphorylated forms, at tyrosine-416 (activated) and tyrosine-527 (inhibited), undergoes time-dependent, region-specific learning-related changes in the domestic chick forebrain after visual imprinting. These changes occur in the intermediate medial mesopallium (IMM), a site of memory formation for visual imprinting, but not the posterior pole of the nidopallium (PPN), a control brain region not involved in imprinting. Src interacts with mitochondrial genome-coded NADH dehydrogenase subunit 2 (NADH2), a component of mitochondrial respiratory complex I. This interaction occurs at brain excitatory synapses bearing NMDA glutamate receptors. The involvement of Src-NADH2 complexes in learning and memory is not yet explored. We show for the first time that, independently of changes in total Src or total NADH2, NADH2 bound to Src immunoprecipitated from the P2 plasma membrane-mitochondrial fraction: (i) is increased in a learning-related manner in the left IMM 1 h after the end of training; (ii), is decreased in the right IMM in a learning-related way 24 h after training. These changes occurred in the IMM but not the PPN. They are attributable to learning occurring during training rather than a predisposition to learn. Learning-related changes in Src-bound NADH2 are thus time- and region-dependent.

Learning biologically-interpretable latent representations for gene expression data: Pathway Activity Score Learning Algorithm.

Molecular gene-expression datasets consist of samples with tens of thousands of measured quantities (i.e., high dimensional data). However, lower-dimensional representations that retain the useful biological information do exist. We present a novel algorithm for such dimensionality reduction called Pathway Activity Score Learning (PASL). The major novelty of PASL is that the constructed features directly correspond to known molecular pathways (genesets in general) and can be interpreted as pathway activity scores. Hence, unlike PCA and similar methods, PASL's latent space has a fairly straightforward biological interpretation. PASL is shown to outperform in predictive performance the state-of-the-art method (PLIER) on two collections of breast cancer and leukemia gene expression datasets. PASL is also trained on a large corpus of 50000 gene expression samples to construct a universal dictionary of features across different tissues and pathologies. The dictionary validated on 35643 held-out samples for reconstruction error. It is then applied on 165 held-out datasets spanning a diverse range of diseases. The AutoML tool JADBio is employed to show that the predictive information in the PASL-created feature space is retained after the transformation. The code is available at https://github.com/mensxmachina/PASL.

Automated machine learning for genome wide association studies.

MOTIVATION: Genome-wide association studies (GWAS) present several computational and statistical challenges for their data analysis, including knowledge discovery, interpretability, and translation to clinical practice. RESULTS: We develop, apply, and comparatively evaluate an automated machine learning (AutoML) approach, customized for genomic data that delivers reliable predictive and diagnostic models, the set of genetic variants that are important for predictions (called a biosignature), and an estimate of the out-of-sample predictive power. This AutoML approach discovers variants with higher predictive performance compared to standard GWAS methods, computes an individual risk prediction score, generalizes to new, unseen data, is shown to better differentiate causal variants from other highly correlated variants, and enhances knowledge discovery and interpretability by reporting multiple equivalent biosignatures. AVAILABILITY AND IMPLEMENTATION: Code for this study is available at: https://github.com/mensxmachina/autoML-GWAS. JADBio offers a free version at: https://jadbio.com/sign-up/. SNP data can be downloaded from the EGA repository (https://ega-archive.org/). PRS data are found at: https://www.aicrowd.com/challenges/opensnp-height-prediction. Simulation data to study population structure can be found at: https://easygwas.ethz.ch/data/public/dataset/view/1/.

Thymic function and survival at advance ages in nursing home residents from Southern Italy.

BACKGROUND: Immunosenescence is a complex process characterized by an age-related remodelling of immune system. The prominent effects of the immunosenescence process is the thymic involution and, consequently, the decreased numbers and functions of T cells. Since thymic involution results in a collapse of the T-cell receptor (TCR) repertoire, a reliable biomarker of its activity is represented by the quantification of signal joint T-cell receptor rearrangement excision circles (sjTRECs) levels. Although it is reasonable to think that thymic function could play a crucial role on elderly survival, only a few studies investigated the relationship between an accurate measurement of human thymic function and survival at old ages. METHODS AND FINDINGS: By quantifying the amount sjTRECs by real-time polymerase chain reaction (PCR), the decrease in thymic output in 241 nursing home residents from Calabria (Southern Italy) was evaluated to investigate the relationship between thymic function and survival at old ages. We found that low sjTREC levels were associated with a significant increased risk of mortality at older ages. Nursing home residents with lower sjTREC exhibit a near 2-fold increase in mortality risk compared to those with sjTREC levels in a normal range. CONCLUSION: Thymic function failure is an independent predictor of mortality among elderly nursing home residents. sjTREC represents a biomarker of effective ageing as its blood levels could anticipate individuals at high risk of negative health outcomes. The identification of these subjects is crucial to manage older people's immune function and resilience, such as, for instance, to plan more efficient vaccinal campaigns in older populations.

scAEGAN: Unification of single-cell genomics data by adversarial learning of latent space correspondences.

Recent progress in Single-Cell Genomics has produced different library protocols and techniques for molecular profiling. We formulate a unifying, data-driven, integrative, and predictive methodology for different libraries, samples, and paired-unpaired data modalities. Our design of scAEGAN includes an autoencoder (AE) network integrated with adversarial learning by a cycleGAN (cGAN) network. The AE learns a low-dimensional embedding of each condition, whereas the cGAN learns a non-linear mapping between the AE representations. We evaluate scAEGAN using simulated data and real scRNA-seq datasets, different library preparations (Fluidigm C1, CelSeq, CelSeq2, SmartSeq), and several data modalities as paired scRNA-seq and scATAC-seq. The scAEGAN outperforms Seurat3 in library integration, is more robust against data sparsity, and beats Seurat 4 in integrating paired data from the same cell. Furthermore, in predicting one data modality from another, scAEGAN outperforms Babel. We conclude that scAEGAN surpasses current state-of-the-art methods and unifies integration and prediction challenges.

An ELOVL2-Based Epigenetic Clock for Forensic Age Prediction: A Systematic Review.

The prediction of chronological age from methylation-based biomarkers represents one of the most promising applications in the field of forensic sciences. Age-prediction models developed so far are not easily applicable for forensic caseworkers. Among the several attempts to pursue this objective, the formulation of single-locus models might represent a good strategy. The present work aimed to develop an accurate single-locus model for age prediction exploiting ELOVL2, a gene for which epigenetic alterations are most highly correlated with age. We carried out a systematic review of different published pyrosequencing datasets in which methylation of the ELOVL2 promoter was analysed to formulate age prediction models. Nine of these, with available datasets involving 2298 participants, were selected. We found that irrespective of which model was adopted, a very strong relationship between ELOVL2 methylation levels and age exists. In particular, the model giving the best age-prediction accuracy was the gradient boosting regressor with a prediction error of about 5.5 years. The findings reported here strongly support the use of ELOVL2 for the formulation of a single-locus epigenetic model, but the inclusion of additional, non-redundant markers is a fundamental requirement to apply a molecular model to forensic applications with more robust results.

A new approach to broaden the range of eye colour identifiable by IrisPlex in DNA phenotyping.

IrisPlex system represents the most popular model for eye colour prediction. Based on six polymorphisms this model provides very accurate predictions that strongly depend on the definition of eye colour phenotypes. The aim of the present study was to introduce a new approach to improve eye colour prediction using the well-validated IrisPlex system. A sample of 238 individuals from a Southern Italian population was collected and for each of them a high-resolution image of eye was obtained. By quantifying eye colour variation into CIELAB space several clustering algorithms were applied for eye colour classification. Predictions with the IrisPlex model were obtained using eye colour categories defined by both visual inspection and clustering algorithms. IrisPlex system predicted blue and brown eye colour with high accuracy while it was inefficient in the prediction of intermediate eye colour. Clustering-based eye colour resulted in a significantly increased accuracy of the model especially for brown eyes. Our results confirm the validity of the IrisPlex system for forensic purposes. Although the quantitative approach here proposed for eye colour definition slightly improves its prediction accuracy, further research is still required to improve the model particularly for the intermediate eye colour prediction.

Just Add Data: automated predictive modeling for knowledge discovery and feature selection.

Fully automated machine learning (AutoML) for predictive modeling is becoming a reality, giving rise to a whole new field. We present the basic ideas and principles of Just Add Data Bio (JADBio), an AutoML platform applicable to the low-sample, high-dimensional omics data that arise in translational medicine and bioinformatics applications. In addition to predictive and diagnostic models ready for clinical use, JADBio focuses on knowledge discovery by performing feature selection and identifying the corresponding biosignatures, i.e., minimal-size subsets of biomarkers that are jointly predictive of the outcome or phenotype of interest. It also returns a palette of useful information for interpretation, clinical use of the models, and decision making. JADBio is qualitatively and quantitatively compared against Hyper-Parameter Optimization Machine Learning libraries. Results show that in typical omics dataset analysis, JADBio manages to identify signatures comprising of just a handful of features while maintaining competitive predictive performance and accurate out-of-sample performance estimation.

Clinical and Prognostic Implications of Estimating Glomerular Filtration Rate by Three Different Creatinine-Based Equations in Older Nursing Home Residents.

Background: According to the international literature, the percentage of nursing home (NH) residents with renal insufficiency is very high, ranging between 22 and 78%. Diminished kidney function represents a risk factor for drug overdosage, adverse drug reactions, end-stage renal disease, disability, morbidity, and mortality. Several studies suggested that screening for chronic kidney disease (CKD) in high-risk and older populations may represent a cost-effective approach to reducing progression to renal failure and CKD mortality. Objective: This study aimed (i) to investigate to what extent CKD may be staged interchangeably by three different creatinine-based estimated glomerular filtration rate (eGFR) equations in a sample of older adults living in long-term care facilities; (ii) to investigate factors explaining differences among eGFR equations; and (iii) to compare the predictivity of different creatinine-based eGFR equations with respect to all-cause mortality. Methods: A total of 522 residents aged 65 years and older participated in a prospective cohort study of 9 long-term care facilities in Calabria. eGFR was calculated by Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI), Berlin initiative study (BIS), and full age spectrum (FAS) equations. Disability in at least one activity of daily living (ADL), depression, cognitive impairment, comorbidity, and malnutrition was considered in the analysis. Statistical analysis was carried out by Bland-Altman analysis, and 2-year mortality was investigated by Kaplan-Meier curves and Cox regression analysis. Results: Depending on the adopted equation, the prevalence of NH residents with impaired renal function (eGFR < 60 ml/min/1.73 m2) ranged between 58.2% for the CKD-EPI and 79.1% for the BIS1 equation. The average difference between BIS and FAS was nearly negligible (0.45 ml/min/1.73 m2), while a significant bias was detected between CKD-EPI and BIS and also between CKD-EPI and FAS (6.21 ml/min/1.73 m2 and 6.65 ml/min/1.73 m2, respectively). Although the eGFR study equations had comparable prognostic accuracy in terms of mortality risk, BIS and FAS were able to reclassify NH residents pertaining to a low-risk group with CKD-EPI, and this reclassification improves the discriminative capacity of CKD-EPI with respect to overall mortality. Conclusion: Despite the relatively good correlation between eGFRs calculated using all adopted equations, the findings in this study reported clearly demonstrated that CKD-EPI and BIS/FAS equations are not interchangeable to assess eGFR among older people and particularly in institutionalized and frail older subjects.

Age-related changes in medial septal cholinergic and GABAergic projection neurons and hippocampal neurotransmitter receptors: relationship with memory impairment.

The hippocampus, which provides cognitive functions, has been shown to become highly vulnerable during aging. One important modulator of the hippocampal neural network is the medial septum (MS). The present study attempts to determine how age-related mnemonic dysfunction is associated with neurochemical changes in the septohippocampal (SH) system, using behavioral and immunochemical experiments performed on young-adult, middle-aged and aged rats. According to these behavioral results, the aged and around 52.8% of middle-aged rats (within the "middle-aged-impaired" sub-group) showed both impaired spatial reference memory in the Morris water maze and habituation in the open field. Immunohistochemical studies revealed a significant decrease in the number of MS choline acetyltransferase immunoreactive cells in the aged and all middle-aged rats, in comparison to the young; however the number of gamma-aminobutyric acid-ergic (GABAergic) parvalbumin immunoreactive cells was higher in middle-aged-impaired and older rats compared to young and middle-aged-unimpaired rats. Western Blot analysis moreover showed a decrease in the level of expression of cholinergic, GABAergic and glutamatergic receptors in the hippocampus of middle-aged-impaired and aged rats in contrast to middle-aged-unimpaired and young rats. The present results demonstrate for the first time that a decrease in the expression level of hippocampal receptors in naturally aged rats with impaired cognitive abilities occurs in parallel with an increase in the number of GABAergic neurons in the MS, and it highlights the particular importance of inhibitory signaling in the SH network for memory function.

Myo-Inositol Limits Kainic Acid-Induced Epileptogenesis in Rats.

Epilepsy is a severe neurological disease characterized by spontaneous recurrent seizures (SRS). A complex pathophysiological process referred to as epileptogenesis transforms a normal brain into an epileptic one. Prevention of epileptogenesis is a subject of intensive research. Currently, there are no clinically approved drugs that can act as preventive medication. Our previous studies have revealed highly promising antiepileptogenic properties of a compound-myo-inositol (MI) and the present research broadens previous results and demonstrates the long-term disease-modifying effect of this drug, as well as the amelioration of cognitive comorbidities. For the first time, we show that long-term treatment with MI: (i) decreases the frequency and duration of electrographic SRS in the hippocampus; (ii) has an ameliorating effect on spatial learning and memory deficit associated with epileptogenesis, and (iii) attenuates cell loss in the hippocampus. MI treatment also alters the expression of the glial fibrillary acidic protein, LRRC8A subunit of volume-regulated anion channels, and protein tyrosine phosphatase receptor type R, all expected to counteract the epileptogenesis. All these effects are still present even 4 weeks after MI treatment ceased. This suggests that MI may exert multiple actions on various epileptogenesis-associated changes in the brain and, therefore, could be considered as a candidate target for prevention of epileptogenesis.

Association between IGF-1 levels ranges and all-cause mortality: A meta-analysis.

The association between IGF-1 levels and mortality in humans is complex with low levels being associated with both low and high mortality. The present meta-analysis investigates this complex relationship between IGF-1 and all-cause mortality in prospective cohort studies. A systematic literature search was conducted in PubMed/MEDLINE, Scopus, and Cochrane Library up to September 2019. Published studies were eligible for the meta-analysis if they had a prospective cohort design, a hazard ratio (HR) and 95% confidence interval (CI) for two or more categories of IGF-1 and were conducted among adults. A random-effects model with a restricted maximum likelihood heterogeneity variance estimator was used to find combined HRs for all-cause mortality. Nineteen studies involving 30,876 participants were included. Meta-analysis of the 19 eligible studies showed that with respect to the low IGF-1 category, higher IGF-1 was not associated with increased risk of all-cause mortality (HR = 0.84, 95% CI = 0.68-1.05). Dose-response analysis revealed a U-shaped relation between IGF-1 and mortality HR. Pooled results comparing low vs. middle IGF-1 showed a significant increase of all-cause mortality (HR = 1.33, 95% CI = 1.14-1.57), as well as comparing high vs. middle IGF-1 categories (HR = 1.23, 95% CI = 1.06-1.44). Finally, we provide data on the association between IGF-1 levels and the intake of proteins, carbohydrates, certain vitamins/minerals, and specific foods. Both high and low levels of IGF-1 increase mortality risk, with a specific 120-160 ng/ml range being associated with the lowest mortality. These findings can explain the apparent controversy related to the association between IGF-1 levels and mortality.

A Blood-Based Molecular Clock for Biological Age Estimation.

In the last decade, extensive efforts have been made to identify biomarkers of biological age. DNA methylation levels of ELOVL fatty acid elongase 2 (ELOVL2) and the signal joint T-cell receptor rearrangement excision circles (sjTRECs) represent the most promising candidates. Although these two non-redundant biomarkers echo important biological aspects of the ageing process in humans, a well-validated molecular clock exploiting these powerful candidates has not yet been formulated. The present study aimed to develop a more accurate molecular clock in a sample of 194 Italian individuals by re-analyzing the previously obtained EVOLV2 methylation data together with the amount of sjTRECs in the same blood samples. The proposed model showed a high prediction accuracy both in younger individuals with an error of about 2.5 years and in older subjects where a relatively low error was observed if compared with those reported in previously published studies. In conclusion, an easy, cost-effective and reliable model to measure the individual rate and the quality of aging in human population has been proposed. Further studies are required to validate the model and to extend its use in an applicative context.

Short and Long Time Bloodstains Age Determination by Colorimetric Analysis: A Pilot Study.

Bloodstains found at crime scenes represent a crucial source of information for investigative purposes. However, in forensic practice, no technique is currently used to estimate the time from deposition of bloodstains. This preliminary study focuses on the age estimation of bloodstains by exploiting the color variations over time due to the oxidation of the blood. For this purpose, we used a colorimetric methodology in order to easily obtain objective, univocal and reproducible results. We developed two bloodstain age prediction algorithms: a short-term and a long-term useful model for the first 24h and 60 days, respectively. Both models showed high levels of classification accuracy, particularly for the long-term model. Although a small-scale study, these results improve the potential application of colorimetric analysis in the time-line reconstruction of violent criminal events.

Src and Memory: A Study of Filial Imprinting and Predispositions in the Domestic Chick.

Visual imprinting is a learning process whereby young animals come to prefer a visual stimulus after exposure to it (training). The available evidence indicates that the intermediate medial mesopallium (IMM) in the domestic chick forebrain is a site of memory formation during visual imprinting. We have studied the role of Src, an important non-receptor tyrosine kinase, in memory formation. Amounts of total Src (Total-Src) and its two phosphorylated forms, tyrosine-416 (activated, 416P-Src) and tyrosine-527 (inhibited, 527P-Src), were measured 1 and 24 h after training in the IMM and in a control brain region, the posterior pole of nidopallium (PPN). One hour after training, in the left IMM, we observed a positive correlation between the amount of 527P-Src and learning strength that was attributable to learning, and there was also a positive correlation between 416P-Src and learning strength that was attributable to a predisposition to learn readily. Twenty-four hours after training, the amount of Total-Src increased with learning strength in both the left and right IMM, and amount of 527P-Src increased with learning strength only in the left IMM; both correlations were attributable to learning. A further, negative, correlation between learning strength and 416P-Src/Total-Src in the left IMM reflected a predisposition to learn. No learning-related changes were found in the PPN control region. We suggest that there are two pools of Src; one of them in an active state and reflecting a predisposition to learn, and the second one in an inhibited condition, which increases as a result of learning. These two pools may represent two or more signaling pathways, namely, one pathway downstream of Src activated by tyrosine-416 phosphorylation and another upstream of Src, keeping the enzyme in an inactivated state via phosphorylation of tyrosine-527.

Metabolomic Fingerprint of Mecp2-Deficient Mouse Cortex: Evidence for a Pronounced Multi-Facetted Metabolic Component in Rett Syndrome.

Using unsupervised metabolomics, we defined the complex metabolic conditions in the cortex of a mouse model of Rett syndrome (RTT). RTT, which represents a cause of mental and cognitive disabilities in females, results in profound cognitive impairment with autistic features, motor disabilities, seizures, gastrointestinal problems, and cardiorespiratory irregularities. Typical RTT originates from mutations in the X-chromosomal methyl-CpG-binding-protein-2 (Mecp2) gene, which encodes a transcriptional modulator. It then causes a deregulation of several target genes and metabolic alterations in the nervous system and peripheral organs. We identified 101 significantly deregulated metabolites in the Mecp2-deficient cortex of adult male mice; 68 were increased and 33 were decreased compared to wildtypes. Pathway analysis identified 31 mostly upregulated metabolic pathways, in particular carbohydrate and amino acid metabolism, key metabolic mitochondrial/extramitochondrial pathways, and lipid metabolism. In contrast, neurotransmitter-signaling is dampened. This metabolic fingerprint of the Mecp2-deficient cortex of severely symptomatic mice provides further mechanistic insights into the complex RTT pathogenesis. The deregulated pathways that were identified-in particular the markedly affected amino acid and carbohydrate metabolism-confirm a complex and multifaceted metabolic component in RTT, which in turn signifies putative therapeutic targets. Furthermore, the deregulated key metabolites provide a choice of potential biomarkers for a more detailed rating of disease severity and disease progression.

Automated machine learning optimizes and accelerates predictive modeling from COVID-19 high throughput datasets.

COVID-19 outbreak brings intense pressure on healthcare systems, with an urgent demand for effective diagnostic, prognostic and therapeutic procedures. Here, we employed Automated Machine Learning (AutoML) to analyze three publicly available high throughput COVID-19 datasets, including proteomic, metabolomic and transcriptomic measurements. Pathway analysis of the selected features was also performed. Analysis of a combined proteomic and metabolomic dataset led to 10 equivalent signatures of two features each, with AUC 0.840 (CI 0.723-0.941) in discriminating severe from non-severe COVID-19 patients. A transcriptomic dataset led to two equivalent signatures of eight features each, with AUC 0.914 (CI 0.865-0.955) in identifying COVID-19 patients from those with a different acute respiratory illness. Another transcriptomic dataset led to two equivalent signatures of nine features each, with AUC 0.967 (CI 0.899-0.996) in identifying COVID-19 patients from virus-free individuals. Signature predictive performance remained high upon validation. Multiple new features emerged and pathway analysis revealed biological relevance by implication in Viral mRNA Translation, Interferon gamma signaling and Innate Immune System pathways. In conclusion, AutoML analysis led to multiple biosignatures of high predictive performance, with reduced features and large choice of alternative predictors. These favorable characteristics are eminent for development of cost-effective assays to contribute to better disease management.