Search

We report on the complete computation of the mixed QCD-electroweak corrections to the neutral-current Drell-Yan process. Our calculation holds in the entire range of dilepton invariant masses. We present phenomenological results for several kinematical distributions in the case of bare muons both in the resonant region and for high invariant masses. We also consider the forward-backward asymmetry, which is a key observable to measure the weak mixing angle. We finally extend our calculation to dressed leptons and compare our results in the massless limit to those available in the literature., Comment: 27 pages, 7 Figures, and 2 Tables

The development of domain-specific language models has significantly advanced natural language processing applications in various specialized fields, particularly in biomedicine. However, the focus has largely been on English-language models, leaving a gap for less-resourced languages such as Italian. This paper introduces Igea, the first decoder-only language model designed explicitly for biomedical text generation in Italian. Built on the Minerva model and continually pretrained on a diverse corpus of Italian medical texts, Igea is available in three model sizes: 350 million, 1 billion, and 3 billion parameters. The models aim to balance computational efficiency and performance, addressing the challenges of managing the peculiarities of medical terminology in Italian. We evaluate Igea using a mix of in-domain biomedical corpora and general-purpose benchmarks, highlighting its efficacy and retention of general knowledge even after the domain-specific training. This paper discusses the model's development and evaluation, providing a foundation for future advancements in Italian biomedical NLP., Comment: 6 pages, 1 figure, 3 tables

We present a generator for lepton nucleon collisions in the DIS regime, focusing in particular on processes with a massive lepton and/or a massive quark in the final state. We have built a full code matching NLO QCD corrections to parton shower Monte Carlo programs in the POWHEG-BOX framework. Our code can be used to compute NLO+PS accurate fully differential predictions for neutral current and charged current processes, including processes with an incoming tau neutrino, and/or including charm quarks in the final state. We also made comparisons with available data and predictions for the new neutrino experiments at CERN., Comment: 51 pages, 29 figures, 2 tables, LaTeX

We compute differential distributions for Drell-Yan processes at the LHC and the Tevatron colliders at next-to-next-to-leading order in perturbative QCD, including fiducial cuts on the decay leptons in the final state. The comparison of predictions obtained with four different codes shows excellent agreement, once linear power corrections from the fiducial cuts are included in those codes that rely on phase-space slicing subtraction schemes. For $Z$-boson production we perform a detailed study of the symmetric cuts on the transverse momenta of the decay leptons. Predictions at fixed order in perturbative QCD for those symmetric cuts, typically imposed in experiments, suffer from an instability. We show how this can be remedied by an all-order resummation of the fiducial transverse momentum spectrum, and we comment on the choice of cuts for future experimental analyses., Comment: 31 pages LaTeX, 13 figures

We consider neutral- and charged-current Drell Yan lepton-pair production at hadron colliders, and include dominant classes of electroweak and mixed QCD-electroweak corrections to all orders in perturbation theory. The accurate description of these physical effects is vital for a precise determination of fundamental Standard Model parameters, such as the $W$-boson mass and the electroweak mixing angle, as well as for a solid assessment of the associated theoretical uncertainties. Our state-of-the-art resummation reaches next-to-leading-logarithmic accuracy in both the electroweak and the mixed QCD-electroweak perturbative expansions, including constant terms at first order beyond Born level in both couplings, i.e. at order $\alpha$ and $\alpha_s \alpha$. These effects are incorporated on top of QCD predictions at next-to-next-to-next-to-leading-logarithmic accuracy, which include constant terms at third order in the strong coupling. Our results retain, for the first time at this accuracy, full dependence on the kinematics of the final-state leptons, thereby enabling a realistic comparison with experimental analyses at the differential level in presence of fiducial cuts. We present a phenomenological analysis of the impact of electroweak corrections in relevant observables at the LHC. We find visible shape distortions in resummation-dominated kinematical regions with respect to pure-QCD predictions, highlighting the importance of a complete description, not limited to QCD, for precision Drell Yan physics., Comment: 23 pages, 13 figures

BACKGROUND: Radiology reports are typically written in a free-text format, making clinical information difficult to extract and use. Recently the adoption of structured reporting (SR) has been recommended by various medical societies thanks to the advantages it offers, e.g. standardization, completeness and information retrieval. We propose a pipeline to extract information from free-text radiology reports, that fits with the items of the reference SR registry proposed by a national society of interventional and medical radiology, focusing on CT staging of patients with lymphoma. METHODS: Our work aims to leverage the potential of Natural Language Processing (NLP) and Transformer-based models to deal with automatic SR registry filling. With the availability of 174 radiology reports, we investigate a rule-free generative Question Answering approach based on a domain-specific version of T5 (IT5). Two strategies (batch-truncation and ex-post combination) are implemented to comply with the model's context length limitations. Performance is evaluated in terms of strict accuracy, F1, and format accuracy, and compared with the widely used GPT-3.5 Large Language Model. A 5-point Likert scale questionnaire is used to collect human-expert feedback on the similarity between medical annotations and generated answers. RESULTS: The combination of fine-tuning and batch splitting allows IT5 to achieve notable results; it performs on par with GPT-3.5 albeit its size being a thousand times smaller in terms of parameters. Human-based assessment scores show a high correlation (Spearman's correlation coefficients>0.88, p-values<0.001) with AI performance metrics (F1) and confirm the superior ability of LLMs (i.e., GPT-3.5, 175B of parameters) in generating plausible human-like statements.

Applying Artificial Intelligence (AI) and Machine Learning (ML) in critical contexts, such as medicine, requires the implementation of safety measures to reduce risks of harm in case of prediction errors. Spotting ML failures is of paramount importance when ML predictions are used to drive clinical decisions. ML predictive reliability measures the degree of trust of a ML prediction on a new instance, thus allowing decision-makers to accept or reject it based on its reliability. To assess reliability, we propose a method that implements two principles. First, our approach evaluates whether an instance to be classified is coming from the same distribution of the training set. To do this, we leverage Autoencoders (AEs) ability to reconstruct the training set with low error. An instance is considered Out-of-Distribution (OOD) if the AE reconstructs it with a high error. Second, it is evaluated whether the ML classifier has good performances on samples similar to the newly classified instance by using a proxy model. We show that this approach is able to assess reliability both in a simulated scenario and on a model trained to predict disease progression of Multiple Sclerosis patients. We also developed a Python package, named relAI, to embed reliability measures into ML pipelines. We propose a simple approach that can be used in the deployment phase of any ML model to suggest whether to trust predictions or not. Our method holds the promise to provide effective support to clinicians by spotting potential ML failures during deployment., Comment: 20 pages, 7 figures

We consider subleading power corrections to event shape variables in $e^+e^-$ collisions at the first order in the QCD coupling $\alpha_s$. We start from the jettiness variable $\tau_2$ and the $y_{23}$ resolution variable for the $k_T$ jet clustering algorithm and we analytically compute the corresponding cumulative cross section. We investigate the origin of the different power suppressed contributions in the two-jet limit and trace it back to their different coverage of the phase space. We extend our analysis to the case of thrust and of the $C$-parameter, and we finally discuss a class of observables that depend on a continuous parameter giving different weight to central and forward emissions and we evaluate the corresponding subleading power corrections., Comment: 16 pages, 4 figures. Minor modifications and updated references. Version published in EPJC

Scenarios with new physics particles feebly interacting with the Standard Model sector provide compelling candidates for dark matter searches. Geared with a set of new experiments for the detection of neutrinos and long-lived particles the Large Hadron Collider (LHC) has joined the hunt for these elusive states. On the theoretical side, this emerging physics program requires reliable estimates of the associated particle fluxes, in particular those arising from heavy hadron decays. In this work, we provide state-of-the-art QCD predictions for heavy hadron production including radiative corrections at next-to-leading order and using parton distribution functions including small-$x$ resummation at next-to-leading logarithmic accuracy. We match our predictions to parton showers to provide a realistic description of hadronisation effects. We demonstrate the utility of our predictions by presenting the energy spectrum of neutrinos from charm hadron decays. Furthermore, we employ our predictions to estimate, for the first time, FASER's sensitivity to electrophilic ALPs, which are predominantly generated in beauty hadron decays., Comment: 12 pages, 8 figures. Version published in EPJC

We consider the class of inclusive hadron collider processes in which one or more energetic jets are produced, possibly accompanied by colourless particles. We provide a general formulation of a slicing scheme for this class of processes, by identifying the various contributions that need to be computed up to next-to-leading order (NLO) in QCD perturbation theory. We focus on two novel observables, the one-jet resolution variable $\Delta E_t$ and the $n$-jet resolution variable $k_{T}^{\mathrm{ness}}$, and explicitly compute all the ingredients needed to carry out NLO computations using these variables. We contrast the behaviour of these variables when the slicing parameter becomes small. In the case of $k_{T}^{\mathrm{ness}}$ we also present results for the hadroproduction of multiple jets., Comment: 47 pages, 3 figures. Minor modifications. Version published in JHEP

The production of a top-antitop quark pair in association with a $W$ boson ($t\bar tW$) is one of the heaviest signatures currently probed at the Large Hadron Collider (LHC). Since the first observation reported in 2015 the corresponding rates have been found to be consistently higher than the Standard Model predictions, which are based on next-to-leading order~(NLO) calculations in the QCD and electroweak (EW) interactions. We present the first next-to-next-to-leading order (NNLO) QCD computation of $t\bar tW$ production at hadron colliders. The calculation is exact, except for the finite part of the two-loop virtual corrections, which is estimated using two different approaches that lead to consistent results within their uncertainties. We combine the newly computed NNLO QCD corrections with the complete NLO QCD+EW results, thus obtaining the most advanced perturbative prediction available to date for the \ttW inclusive cross section. The tension with the latest ATLAS and CMS results remains at the $1\sigma-2\sigma$ level., Comment: 7 pages, 3 figures, 1 table. Minor changes, version published on PRL

The introduction of computerized medical records in hospitals has reduced burdensome activities like manual writing and information fetching. However, the data contained in medical records are still far underutilized, primarily because extracting data from unstructured textual medical records takes time and effort. Information Extraction, a subfield of Natural Language Processing, can help clinical practitioners overcome this limitation by using automated text-mining pipelines. In this work, we created the first Italian neuropsychiatric Named Entity Recognition dataset, PsyNIT, and used it to develop a Transformers-based model. Moreover, we collected and leveraged three external independent datasets to implement an effective multicenter model, with overall F1-score 84.77%, Precision 83.16%, Recall 86.44%. The lessons learned are: (i) the crucial role of a consistent annotation process and (ii) a fine-tuning strategy that combines classical methods with a "low-resource" approach. This allowed us to establish methodological guidelines that pave the way for Natural Language Processing studies in less-resourced languages., Comment: 2 figures, 6 tables, Supplementary Notes included

In the era of digital healthcare, the huge volumes of textual information generated every day in hospitals constitute an essential but underused asset that could be exploited with task-specific, fine-tuned biomedical language representation models, improving patient care and management. For such specialized domains, previous research has shown that fine-tuning models stemming from broad-coverage checkpoints can largely benefit additional training rounds over large-scale in-domain resources. However, these resources are often unreachable for less-resourced languages like Italian, preventing local medical institutions to employ in-domain adaptation. In order to reduce this gap, our work investigates two accessible approaches to derive biomedical language models in languages other than English, taking Italian as a concrete use-case: one based on neural machine translation of English resources, favoring quantity over quality; the other based on a high-grade, narrow-scoped corpus natively written in Italian, thus preferring quality over quantity. Our study shows that data quantity is a harder constraint than data quality for biomedical adaptation, but the concatenation of high-quality data can improve model performance even when dealing with relatively size-limited corpora. The models published from our investigations have the potential to unlock important research opportunities for Italian hospitals and academia. Finally, the set of lessons learned from the study constitutes valuable insights towards a solution to build biomedical language models that are generalizable to other less-resourced languages and different domain settings., Comment: 8 pages, 2 figures, 6 tables. Published in Journal of Biomedical Informatics

We present the first calculation for the hadroproduction of a $W$ boson in association with a massive bottom ($b$) quark-antiquark pair at next-to-next-to-leading order (NNLO) in QCD perturbation theory. We exploit the hierarchy between the $b$ quark mass and the characteristic energy scale of the process to obtain a reliable analytic expression for the two-loop virtual amplitude with three massive legs, starting from the corresponding result available for massless bottom quarks. The use of massive $b$ quarks avoids the ambiguities associated with the correct flavour assignment in massless calculations, paving the way to a more realistic comparison with experimental data. We present phenomenological results considering proton-proton collisions at centre-of-mass energy $\sqrt{s}=13.6$ TeV for inclusive $Wb \bar b$ production and within a fiducial region relevant for the associated production of a $W$ boson and a Higgs boson decaying into a $b \bar b$ pair, for which $Wb \bar b$ production represents one of the most relevant backgrounds. We find that the NNLO corrections are substantial and that their inclusion is mandatory to obtain reliable predictions., Comment: 8 pages, 2 figures, 2 tables. Final version published in PRD

Recent progress in calculating lepton density functions inside the proton and simulating lepton showers laid the foundations for precision studies of resonant leptoquark production at hadron colliders. Direct quark-lepton fusion into a leptoquark is a novel production channel at the LHC that has the potential to probe a unique parameter space for large masses and couplings. In this work, we build the first Monte Carlo event generator for a full-fledged simulation of this process at NLO for production, followed by a subsequent decay using the POWHEG method and matching to the parton showers utilizing HERWIG. The code can handle all scalar leptoquark models with renormalisable quark-lepton interactions. We then comprehensively study the differential distributions, including higher-order effects, and asses the corresponding theoretical uncertainties. We also quantify the impact of the improved predictions on the projected (HL-)LHC sensitivities and initiate the first exploration of the potential at the FCC-hh. Our work paves the way toward performing LHC searches using this channel., Comment: 33 pages, 7 figures, 3 tables