Search

Background: Neonatal seizures are a neurological emergency that require urgent treatment. They are hard to diagnose clinically and can go undetected if EEG monitoring is unavailable. EEG interpretation requires specialised expertise which is not widely available. Algorithms to detect EEG seizures can address this limitation but have yet to reach widespread clinical adoption. Methods: Retrospective EEG data from 332 neonates was used to develop and validate a seizure-detection model. The model was trained and tested with a development dataset ($n=202$) that was annotated with over 12k seizure events on a per-channel basis. This dataset was used to develop a convolutional neural network (CNN) using a modern architecture and training methods. The final model was then validated on two independent multi-reviewer datasets ($n=51$ and $n=79$). Results: Increasing dataset and model size improved model performance: Matthews correlation coefficient (MCC) and Pearson's correlation ($r$) increased by up to 50% with data scaling and up to 15% with model scaling. Over 50k hours of annotated single-channel EEG was used for training a model with 21 million parameters. State-of-the-art was achieved on an open-access dataset (MCC=0.764, $r=0.824$, and AUC=0.982). The CNN attains expert-level performance on both held-out validation sets, with no significant difference in inter-rater agreement among the experts and among experts and algorithm ($\Delta \kappa < -0.095$, $p>0.05$). Conclusion: With orders of magnitude increases in data and model scale we have produced a new state-of-the-art model for neonatal seizure detection. Expert-level equivalence on completely unseen data, a first in this field, provides a strong indication that the model is ready for further clinical validation.

We present a new approach to the problem of estimating the redshift of galaxies from photometric data. The approach uses a genetic algorithm combined with non-linear regression to model the 2SLAQ LRG data set with SDSS DR7 photometry. The genetic algorithm explores the very large space of high order polynomials while only requiring optimisation of a small number of terms. We find a $\sigma_{\text{rms}}=0.0408\pm 0.0006$ for redshifts in the range $0.4

A model of high scale inflation is presented where the radial part of the Peccei-Quinn (PQ) field with a non-minimal coupling to gravity plays the role of the inflaton, and the QCD axion is the dark matter. A quantum fluctuation of $\mathcal{O}(H/2\pi)$ in the axion field will result in a smaller angular fluctuation if the PQ field is sitting at a larger radius during inflation than in the vacuum. This changes the effective axion decay constant, $f_a$, during inflation and dramatically reduces the production of isocurvature modes. This mechanism opens up a new window in parameter space where an axion decay constant in the range $10^{12}\text{ GeV}\lesssim f_a\lesssim 10^{15}\text{ GeV}$ is compatible with observably large $r$. The exact range allowed for $f_a$ depends on the efficiency of reheating. This model also predicts a minimum possible value of $r=10^{-3}$. The new window can be explored by a measurement of $r$ possible with \textsc{Spider} and the proposed CASPEr experiment search for high $f_a$ axions., Comment: 7 pages, 4 figures

We investigate the possibility of using the only known fundamental scalar, the Higgs, as an inflaton with minimal coupling to gravity. The peculiar appearance of a plateau or a false vacuum in the renormalised effective scalar potential suggests that the Higgs might drive inflation. For the case of a false vacuum we use an additional singlet scalar field, motivated by the strong CP problem, and its coupling to the Higgs to lift the barrier allowing for a graceful exit from inflation by mimicking hybrid inflation. We find that this scenario is incompatible with current measurements of the Higgs mass and the QCD coupling constant and conclude that the Higgs can only be the inflaton in more complicated scenarios., Comment: 7 pages, 5 figures

We consider the effect of a period of inflation with a high energy density upon the stability of the Higgs potential in the early universe. The recent measurement of a large tensor-to-scalar ratio, $r_T \sim 0.16$, by the BICEP-2 experiment possibly implies that the energy density during inflation was very high, comparable with the GUT scale. Given that the standard model Higgs potential is known to develop an instability at $\Lambda \sim 10^{10}$ GeV this means that the resulting large quantum fluctuations of the Higgs field could destabilize the vacuum during inflation, even if the Higgs field starts at zero expectation value. We estimate the probability of such a catastrophic destabilisation given such an inflationary scenario and calculate that for a Higgs mass of $m_h=125.5$ GeV that the top mass must be less than $m_t\sim 172$ GeV. We present two possible cures: a direct coupling between the Higgs and the inflaton and a non-zero temperature from dissipation during inflation., Comment: 5 pages, 4 figures. Typos corrected, results unchanged, matches version submitted to journal

We consider a model with a gauge singlet Dirac fermion as a cold dark matter candidate. The dark matter particle communicates with the Standard Model via a gauge singlet scalar mediator that couples to the Higgs. The scalar mediator also serves to create a tree-level barrier in the scalar potential which leads to a strongly first order electroweak phase transition as required for Electroweak Baryogenesis. We find a large number of models that can account for all the dark matter and provide a strong phase transition while avoiding constraints from dark matter direct detection, electroweak precision data, and the latest Higgs data from the LHC. The next generation of direct detection experiments could rule out a large region of the parameter space but can be evaded in some regions when the Higgs-singlet mixing is very small., Comment: 9 pages, 6 figures. v2: figure captions updated, colour confusion fixed, no other changes. v3: references added, some clarification, now matches version published in JHEP