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: We report the observation of the rare charm decay D-0 -> K-pi(+)e(+)e(-), based on 468 fb(-1) of e(+)e(-) annihilation data collected at or close to the center-of-mass energy of the (sic)(4S) resonance with the BABAR detector at the SLAC National Accelerator Laboratory. We find the branching fraction in the invariant mass range 0.675 < m(e(+)e(-)) < 0.875 GeV/c(2) of the electron-positron pair to be B(D-0 -> K-pi(+)e(+)e(-)) = (4.0 +/- 0.5 +/- 0.2 +/- 0.1) x 10(-6), where the first uncertainty is statistical, the second systematic, and the third due to the uncertainty in the branching fraction of the decay D-0 -> K-pi(+)pi(+)pi(-) used as a normalization mode. The significance of the observation corresponds to 9.7 standard deviations including systematic uncertainties. This result is consistent with the recently reported D-0 -> K-pi(+)mu(+)mu(-) branching fraction, measured in the same invariant mass range, and with the value expected in the standard model. In a set of regions of m(e(+)e(-)), where long-distance effects are potentially small, we determine a 90% confidence level upper limit on the branching fraction B(D-0 -> K-pi(+)e(+)e(-)) < 3.1 x 10(-6).

Abstract: A search for supersymmetry in events with large missing transverse momentum, jets, and at least one hadronically decaying tau-lepton is presented. Two exclusive final states with either exactly one or at least two tau-leptons are considered. The analysis is based on proton-proton collisions at root s = 13 TeV corresponding to an integrated luminosity of 36.1 fb(-1) delivered by the Large Hadron Collider and recorded by the ATLAS detector in 2015 and 2016. No significant excess is observed over the Standard Model expectation. At 95% confidence level, model-independent upper limits on the cross section are set and exclusion limits are provided for two signal scenarios: a simplified model of gluino pair production with tau-rich cascade decays, and a model with gauge-mediated supersymmetry breaking (GMSB). In the simplified model, gluino masses up to 2000 GeV are excluded for low values of the mass of the lightest supersymmetric particle (LSP), while LSP masses up to 1000 GeV are excluded for gluino masses around 1400 GeV. In the GMSB model, values of the supersymmetry-breaking scale are excluded below 110 TeV for all values of tan beta in the range 2 <= tan beta <= 60, and below 120 TeV for tan beta > 30.

Abstract: Even mass neutron-rich niobium isotopes are among the principal contributors to the reactor antineutrino energy spectrum. They are also among the most challenging to measure due to the refractory nature of niobium, and because they exhibit isomeric states lying very close in energy. The beta-intensity distributions of Nb-100gs,Nb-100m and Nb-102gs,Nb-02m beta decays have been determined using the total absorption.-ray spectroscopy technique. The measurements were performed at the upgraded Ion Guide Isotope Separator On-Line facility at the University of Jyvaskyla. Here, the double Penning trap system JYFLTRAP was employed to disentangle the beta decay of the isomeric states. The new data obtained in this challenging measurement have a large impact in antineutrino summation calculations. For the first time the discrepancy between the summation model and the reactor antineutrino measurements in the region of the shape distortion has been reduced.

Abstract: Large numbers of muons will be produced at facilities developed to probe the lepton-flavor-violating process μ-> e gamma. We show that by constructing a suitable asymmetry, radiative muon decay μ-> e gamma nu(mu)(nu) over bar (e) can also be used to test the WW gamma vertex at such facilities. The process has two missing neutrinos in the final state, and upon integrating their momenta the partial differential decay rate shows no radiation-amplitude zero. However, we establish that an easily separable part of the normalized differential decay rate that is odd under the exchange of photon and electron energies does have a zero in the case of the standard model (SM). This new type of zero has hitherto not been studied in the literature. A suitably constructed asymmetry using this fact enables a sensitive probe for the WW gamma vertex beyond the SM. With a simplistic analysis, we find that the C- and P-conserving dimension-four WW gamma vertex can be probed at O(10(-2)) with a satisfactory significance level.