Abstract: Importance Mammography screening currently relies on subjective human interpretation. Artificial intelligence (AI) advances could be used to increase mammography screening accuracy by reducing missed cancers and false positives. Objective To evaluate whether AI can overcome human mammography interpretation limitations with a rigorous, unbiased evaluation of machine learning algorithms. Design, Setting, and Participants In this diagnostic accuracy study conducted between September 2016 and November 2017, an international, crowdsourced challenge was hosted to foster AI algorithm development focused on interpreting screening mammography. More than 1100 participants comprising 126 teams from 44 countries participated. Analysis began November 18, 2016. Main Outcomes and Measurements Algorithms used images alone (challenge 1) or combined images, previous examinations (if available), and clinical and demographic risk factor data (challenge 2) and output a score that translated to cancer yes/no within 12 months. Algorithm accuracy for breast cancer detection was evaluated using area under the curve and algorithm specificity compared with radiologists' specificity with radiologists' sensitivity set at 85.9% (United States) and 83.9% (Sweden). An ensemble method aggregating top-performing AI algorithms and radiologists' recall assessment was developed and evaluated. Results Overall, 144231 screening mammograms from 85580 US women (952 cancer positive <= 12 months from screening) were used for algorithm training and validation. A second independent validation cohort included 166578 examinations from 68008 Swedish women (780 cancer positive). The top-performing algorithm achieved an area under the curve of 0.858 (United States) and 0.903 (Sweden) and 66.2% (United States) and 81.2% (Sweden) specificity at the radiologists' sensitivity, lower than community-practice radiologists' specificity of 90.5% (United States) and 98.5% (Sweden). Combining top-performing algorithms and US radiologist assessments resulted in a higher area under the curve of 0.942 and achieved a significantly improved specificity (92.0%) at the same sensitivity. Conclusions and Relevance While no single AI algorithm outperformed radiologists, an ensemble of AI algorithms combined with radiologist assessment in a single-reader screening environment improved overall accuracy. This study underscores the potential of using machine learning methods for enhancing mammography screening interpretation. Question How do deep learning algorithms perform compared with radiologists in screening mammography interpretation? Findings In this diagnostic accuracy study using 144231 screening mammograms from 85580 women from the United States and 166578 screening mammograms from 68008 women from Sweden, no single artificial intelligence algorithm outperformed US community radiologist benchmarks; including clinical data and prior mammograms did not improve artificial intelligence performance. However, combining best-performing artificial intelligence algorithms with single-radiologist assessment demonstrated increased specificity. Meaning Integrating artificial intelligence to mammography interpretation in single-radiologist settings could yield significant performance improvements, with the potential to reduce health care system expenditures and address resource scarcity experienced in population-based screening programs. This diagnostic accuracy study evaluates whether artificial intelligence can overcome human mammography interpretation limits with a rigorous, unbiased evaluation of machine learning algorithms.

Abstract: Dark matter (DM) is currently searched for with a variety of detection strategies. Accelerator searches are particularly promising, but even if weakly interacting massive particles are found at the Large Hadron Collider (LHC), it will be difficult to prove that they constitute the bulk of the DM in the Universe Omega(DM). We show that a significantly better reconstruction of the DM properties can be obtained with a combined analysis of LHC and direct detection data, by making a simple Ansatz on the weakly interacting massive particles local density rho(0)((chi) over bar1), i.e., by assuming that the local density scales with the cosmological relic abundance, (rho(0)((chi) over bar1)/rho(DM)) = (Omega(0)((chi) over bar1)/Omega(DM)). We demonstrate this method in an explicit example in the context of a 24-parameter supersymmetric model, with a neutralino lightest supersymmetric particle in the stau coannihilation region. Our results show that future ton-scale direct detection experiments will allow to break degeneracies in the supersymmetric parameter space and achieve a significantly better reconstruction of the neutralino composition and its relic density than with LHC data alone.

Abstract: We provide a scenario where naturally small and calculable neutrino masses arise from a supersymmetry-breaking renormalization-group-induced vacuum expectation value. The construction consists of an extended version of the next-to-minimal supersymmetric standard model and the mechanism is illustrated for a universal choice of the soft supersymmetry-breaking parameters. The lightest supersymmetric particle can be an isosinglet scalar neutrino state, potentially viable as WIMP dark matter through its Higgs new boson coupling. The scenario leads to a plethora of new phenomenological implications at accelerators including the Large Hadron Collider.

Abstract: The gauge-invariant generation of an effective gluon mass proceeds through the well-known Schwinger mechanism, whose key dynamical ingredient is the nonperturbative formation of longitudinally coupled massless bound-state excitations. These excitations introduce poles in the vertices of the theory, in such a way as to maintain the Slavnov-Taylor identities intact in the presence of massive gluon propagators. In the present work we first focus on the modifications induced to the nonperturbative three-gluon vertex by the inclusion of massless two-gluon bound states into the kernels appearing in its skeleton expansion. Certain general relations between the basic building blocks of these bound states and the gluon mass are then obtained from the Slavnov-Taylor identities and the Schwinger-Dyson equation governing the gluon propagator. The homogeneous Bethe-Salpeter equation determining the wave function of the aforementioned bound state is then derived, under certain simplifying assumptions. It is then shown, through a detailed analytical and numerical study, that this equation admits nontrivial solutions, indicating that the QCD dynamics support indeed the formation of such massless bound states. These solutions are subsequently used, in conjunction with the aforementioned relations, to determine the momentumdependence of the dynamical gluon mass. Finally, further possibilities and open questions are briefly discussed.

Abstract: Even if supersymmetric particles are found at the Large Hadron Collider (LHC), it will be difficult to prove that they constitute the bulk of the dark matter (DM) in the Universe using LHC data alone. We study the complementarity of LHC and DM indirect searches, working out explicitly the reconstruction of the DM properties for a specific benchmark model in the coannihilation region of a 24-parameters supersymmetric model. Combining mock high-luminosity LHC data with presentday null searches for gamma rays from dwarf galaxies with the Fermi Large Area Telescope, we show that current Fermi Large Area Telescope limits already have the capability of ruling out a spurious wino-like solution which would survive using LHC data only, thus leading to the correct identification of the cosmological solution. We also demonstrate that upcoming Planck constraints on the reionization history will have a similar constraining power and discuss the impact of a possible detection of gamma rays from DM annihilation in the Draco dwarf galaxy with a Cherenkov-Telescope-Array-like experiment. Our results indicate that indirect searches can be strongly complementary to the LHC in identifying the DM particles, even when astrophysical uncertainties are taken into account.