Search

X-ray images are vital in medical diagnostics, but their effectiveness is limited without clinical context. Radiologists often find chest X-rays insufficient for diagnosing underlying diseases, necessitating comprehensive clinical features and data integration. We present a novel framework to enhance the clinical context through augmentation techniques with clinical tabular data, thereby improving its applicability and reliability in AI medical diagnostics. We introduce a pioneering approach to clinical data augmentation that employs large language models to generate patient contextual synthetic data. This methodology is crucial for training more robust deep learning models in healthcare. It preserves the integrity of real patient data while enriching the dataset with contextually relevant synthetic features, significantly enhancing model performance. Our methodology, termed DALL-M, uses a three-phase feature generation process: (i)clinical context storage, (ii)expert query generation, and (iii)context-aware feature augmentation. DALL-M generates new, clinically relevant features by synthesizing chest X-ray images and reports. Applied to 799 cases using nine features from the MIMIC-IV dataset, it created an augmented set of 91 features. This is the first work to generate contextual values for patients' X-ray reports. Specifically, we provide (i)the capacity of LLMs to generate contextual synthetic values for existing clinical features and (ii)their ability to create entirely new clinically relevant features. Empirical validation with machine learning models showed significant performance improvements. Incorporating augmented features increased the F1 score by 16.5% and Precision and Recall by approximately 25%. DALL-M addresses a critical gap in clinical data augmentation, offering a robust framework for generating contextually enriched datasets., Comment: we introduce a pioneering approach to clinical data augmentation that employs large language models (LLMs) to generate patient contextual synthetic data. It preserves the integrity of real patient data while enriching the dataset with contextually relevant synthetic features, significantly enhancing model performance

The next generation of space-based networks will contain optical clocks embedded within satellites. To fully realize the capabilities of such clocks, high-precision clock synchronization across the networks will be necessary. Current experiments have shown the potential for classical frequency combs to synchronize remote optical clocks over free-space. However, these classical combs are restricted in precision to the standard quantum limit. Quantum frequency combs, however, which exhibit quantum properties such as squeezing and entanglement, provide pathways for going beyond the standard quantum limit. Here, we present our perspective on the prospects for practical clock synchronization in space using both classical and quantum frequency combs. We detail the current outcomes achievable with a classical frequency comb approach to synchronization, before quantifying the potential outcomes offered by quantum frequency combs. Challenges to be overcome in deploying frequency combs in space are presented, and the implications of almost-perfect synchronization for future space-based applications and experiments discussed., Comment: 23 pages, 11 figures, perspective, comments welcome

Precision navigation and timing, very-long-baseline interferometry, next-generation communication, sensing, and tests of fundamental physics all require a highly synchronized network of clocks. With the advance of highly-accurate optical atomic clocks, the precision requirements for synchronization are reaching the limits of classical physics (i.e. the standard quantum limit, SQL). Efficiently overcoming the SQL to reach the fundamental Heisenberg limit can be achieved via the use of squeezed or entangled light. Although approaches to the Heisenberg limit are well understood in theory, a practical implementation, such as in space-based platforms, requires that the advantage outweighs the added costs and complexity. Here, we focus on the question: can entanglement yield a quantum advantage in clock synchronization over lossy satellite-to-satellite channels? We answer in the affirmative, showing that the redundancy afforded by the two-mode nature of entanglement allows recoverability even over asymmetrically lossy channels. We further show this recoverability is an improvement over single-mode squeezing sensing, thereby illustrating a new complexity-performance trade-off for space-based sensing applications., Comment: 6 pages, 4 figures, comments welcome

This study investigates the effects of including patients' clinical information on the performance of deep learning (DL) classifiers for disease location in chest X-ray images. Although current classifiers achieve high performance using chest X-ray images alone, our interviews with radiologists indicate that clinical data is highly informative and essential for interpreting images and making proper diagnoses. In this work, we propose a novel architecture consisting of two fusion methods that enable the model to simultaneously process patients' clinical data (structured data) and chest X-rays (image data). Since these data modalities are in different dimensional spaces, we propose a spatial arrangement strategy, spatialization, to facilitate the multimodal learning process in a Mask R-CNN model. We performed an extensive experimental evaluation using MIMIC-Eye, a dataset comprising modalities: MIMIC-CXR (chest X-ray images), MIMIC IV-ED (patients' clinical data), and REFLACX (annotations of disease locations in chest X-rays). Results show that incorporating patients' clinical data in a DL model together with the proposed fusion methods improves the disease localization in chest X-rays by 12\% in terms of Average Precision compared to a standard Mask R-CNN using only chest X-rays. Further ablation studies also emphasize the importance of multimodal DL architectures and the incorporation of patients' clinical data in disease localization. The architecture proposed in this work is publicly available to promote the scientific reproducibility of our study (https://github.com/ChihchengHsieh/multimodal-abnormalities-detection)

Disaster victim identification (DVI) entails a protracted process of evidence collection and data matching to reconcile physical remains with victim identity. Technology is critical to DVI by enabling the linkage of physical evidence to information. However, labelling physical remains and collecting data at the scene are dominated by low-technology paper-based practices. We ask, how can technology help us tag and track the victims of disaster? Our response has two parts. First, we conducted a human-computer interaction led investigation into the systematic factors impacting DVI tagging and tracking processes. Through interviews with Australian DVI practitioners, we explored how technologies to improve linkage might fit with prevailing work practices and preferences; practical and social considerations; and existing systems and processes. Using insights from these interviews and relevant literature, we identified four critical themes: protocols and training; stress and stressors; the plurality of information capture and management systems; and practicalities and constraints. Second, we applied the themes identified in the first part of the investigation to critically review technologies that could support DVI practitioners by enhancing DVI processes that link physical evidence to information. This resulted in an overview of candidate technologies matched with consideration of their key attributes. This study recognises the importance of considering human factors that can affect technology adoption into existing practices. We provide a searchable table (Supplementary Information) that relates technologies to the key attributes relevant to DVI practice, for the reader to apply to their own context. While this research directly contributes to DVI, it also has applications to other domains in which a physical/digital linkage is required, particularly within high-stress environments., Comment: 27 pages, 2 figures

Direct numerical simulations were performed to characterize fully developed supersonic turbulent channel flows over isothermal rough walls. The effect of roughness was incorporated using a level-set/volume-of-fluid immersed boundary method. Turbulence statistics of five channel flows are compared, including one reference case with both walls smooth and four cases with smooth top walls and bottom walls with two-dimensional (2D) and three-dimensional (3D) sinusoidal roughnesses. Results reveal a strong dependence of the turbulence on the roughness topography and the associated shock patterns. Specifically, the 2D geometries generate strong oblique shock waves that propagate across the channel and are reflected back to the rough-wall side. These strong shocks are absent in the smooth-wall channel and are significantly weaker in cases with 3D roughness geometries, replaced by weak shocklets. At the impingement locations of the shocks on the top wall in the 2D roughness cases, localized augmentations of turbulence shear production are observed. Such regions of augmented production also exist for the 3D cases, at a much weaker level. The oblique shock waves are thought to be responsible for a more significant entropy generation for cases with 2D surfaces than those with 3D ones, leading to a higher irreversible heat generation and consequently higher temperature values in 2D roughness cases. In the present supersonic channels, the effects of roughness extend beyond the near-wall layer due to the shocks. This suggests that outer layer similarity may not fully apply to a rough-wall supersonic turbulent flow.

Silicon Carbide is a promising host material for spin defect based quantum sensors owing to its commercial availability and established techniques for electrical and optical microfabricated device integration. The negatively charged silicon vacancy is one of the leading spin defects studied in silicon carbide owing to its near telecom photoemission, high spin number, and nearly temperature independent ground state zero field splitting. We report the realization of nanoTesla shot-noise limited ensemble magnetometry based on optically detected magnetic resonance with the silicon vacancy in 4H silicon carbide. By coarsely optimizing the anneal parameters and minimizing power broadening, we achieved a sensitivity of 3.5 nT/$\sqrt{Hz}$. This was accomplished without utilizing complex photonic engineering, control protocols, or applying excitation powers greater than a Watt. This work demonstrates that the silicon vacancy in silicon carbide provides a low-cost and simple approach to quantum sensing of magnetic fields.

Performance of neural network models relies on the availability of large datasets with minimal levels of uncertainty. Transfer Learning (TL) models have been proposed to resolve the issue of small dataset size by letting the model train on a bigger, task-related reference dataset and then fine-tune on a smaller, task-specific dataset. In this work, we apply a transfer learning approach to improve predictive power in noisy data systems with large variable confidence datasets. We propose a deep neural network method called Filtered Transfer Learning (FTL) that defines multiple tiers of data confidence as separate tasks in a transfer learning setting. The deep neural network is fine-tuned in a hierarchical process by iteratively removing (filtering) data points with lower label confidence, and retraining. In this report we use FTL for predicting the interaction of drugs and proteins. We demonstrate that using FTL to learn stepwise, across the label confidence distribution, results in higher performance compared to deep neural network models trained on a single confidence range. We anticipate that this approach will enable the machine learning community to benefit from large datasets with uncertain labels in fields such as biology and medicine.

The prediction of absorption, distribution, metabolism, excretion, and toxicity (ADMET) of small molecules from their molecular structure is a central problem in medicinal chemistry with great practical importance in drug discovery. Creating predictive models conventionally requires substantial trial-and-error for the selection of molecular representations, machine learning (ML) algorithms, and hyperparameter tuning. A generally applicable method that performs well on all datasets without tuning would be of great value but is currently lacking. Here, we describe Pareto-Optimal Embedded Modeling (POEM), a similarity-based method for predicting molecular properties. POEM is a non-parametric, supervised ML algorithm developed to generate reliable predictive models without need for optimization. POEMs predictive strength is obtained by combining multiple different representations of molecular structures in a context-specific manner, while maintaining low dimensionality. We benchmark POEM relative to industry-standard ML algorithms and published results across 17 classifications tasks. POEM performs well in all cases and reduces the risk of overfitting., Comment: 37 pages, 9 figures, supplemental included, submitted to "Machine Learning: Science and Technology"

This paper investigates a long-standing question about the effect of surface roughness on turbulent flow: what is the equivalent roughness sand-grain height for a given roughness topography? Deep Neural Network (DNN) and Gaussian Process Regression (GPR) machine learning approaches are used to develop a high-fidelity prediction approach of the Nikuradse equivalent sand-grain height $k_s$ for turbulent flows over a wide variety of different rough surfaces. To this end, 45 surface geometries were generated and the flow over them simulated at $\hbox{Re}_\tau=1000$ using direct numerical simulations. These surface geometries differed significantly in moments of surface height fluctuations, effective slope, average inclination, porosity and degree of randomness. Thirty of these surfaces were considered fully-rough and they were supplemented with experimental data for fully-rough flows over 15 more surfaces available from previous studies. The DNN and GPR methods predicted $k_s$ with an average error of less than 10% and a maximum error of less than 30%, which appears to be significantly more accurate than existing prediction formulas. They also identified the surface porosity and the effective slope of roughness in the spanwise direction as important factors in drag prediction.

Numerous lung diseases, such as idiopathic pulmonary fibrosis (IPF), exhibit dilation of the airways. Accurate measurement of dilatation enables assessment of the progression of disease. Unfortunately the combination of image noise and airway bifurcations causes high variability in the profiles of cross-sectional areas, rendering the identification of affected regions very difficult. Here we introduce a noise-robust method for automatically detecting the location of progressive airway dilatation given two profiles of the same airway acquired at different time points. We propose a probabilistic model of abrupt relative variations between profiles and perform inference via Reversible Jump Markov Chain Monte Carlo sampling. We demonstrate the efficacy of the proposed method on two datasets; (i) images of healthy airways with simulated dilatation; (ii) pairs of real images of IPF-affected airways acquired at 1 year intervals. Our model is able to detect the starting location of airway dilatation with an accuracy of 2.5mm on simulated data. The experiments on the IPF dataset display reasonable agreement with radiologists. We can compute a relative change in airway volume that may be useful for quantifying IPF disease progression. The code is available at https://github.com/quan14/Modelling_Airway_Geometry_as_Stock_Market_Data, Comment: 14 pages, 7 figures, Accepted to The 10th International Workshop on Machine Learning in Medical Imaging (MLMI 2019). In conjunction with MICCAI 2019, Shenzhen, China

Defects in wide-bandgap semiconductors provide a pathway for applications in quantum information and sensing in solid state materials. The silicon vacancy in silicon carbide has recently emerged as a new candidate for optical control of single spin qubit with significant material benefits over nitrogen vacancies in diamond. In this work, we present a study of the coherence of silicon vacancies generated via proton irradiation as a function of implantation depth. We show clear evidence of dephasing interactions between the silicon vacancies and the spin environment of the bulk crystal. This result will inform further routes toward fabrication of scalable silicon carbide devices and studies of spin interactions of in high-density ensembles of defects., Comment: 5 pages, 3 figures

In this paper we analyze the derivative nonlinear Schr\"odinger equation on $\mathbb{T}$ with randomized initial data in $\cap_{s < \frac{1}{2}} H^{s}(\mathbb{T})$ according to a Wiener measure. We construct an invariant measure at each sufficiently small, fixed mass $m$ through an argument that emulates the divergence theorem in infinitely many dimensions. We also prove that the density function needed to construct the Wiener measure is in $L^p$, even after scaling of the Fourier coefficients of the intial data., Comment: A previous version contained a serious error in Lemma 5.9. This is the correct version

This paper studies the quintic nonlinear Schr\"odinger equation on $\mathbb{R}^d$ with randomized initial data below the critical regularity $H^{\frac{d-1}{2}}$. The main result is a proof of almost sure local well-posedness given a Wiener Randomization of the data in $H^s$ for $s \in (\frac{d-2}{2}, \frac{d-1}{2})$. The argument further develops the techniques introduced in the work of \'A. B\'enyi, T. Oh and O. Pocovnicu on the cubic problem. The paper concludes with a condition for almost sure global well-posedness., Comment: 21 pages

The analysis of polycrystalline materials benefits greatly from accurate quantitative descriptions of their grain structures. Laguerre tessellations approximate such grain structures very well. However, it is a quite challenging problem to fit a Laguerre tessellation to tomographic data, as a high-dimensional optimization problem with many local minima must be solved. In this paper, we formulate a version of this optimization problem that can be solved quickly using the cross-entropy method, a robust stochastic optimization technique that can avoid becoming trapped in local minima. We demonstrate the effectiveness of our approach by applying it to both artificially generated and experimentally produced tomographic data., Comment: 27 pages, 10 figures, 2 tables

Solid state quantum emitters have shown strong potential for applications in quantum information, but spectral inhomogeneity of these emitters poses a significant challenge. We address this issue in a cavity-quantum dot system by demonstrating cavity-stimulated Raman spin flip emission. This process avoids populating the excited state of the emitter and generates a photon that is Raman shifted from the laser and enhanced by the cavity. The emission is spectrally narrow and tunable over a range of at least 125 GHz, which is two orders of magnitude greater than the natural linewidth. We obtain the regime in which the Raman emission is spin-dependent, which couples the photon to a long-lived electron spin qubit. This process can enable an efficient, tunable source of indistinguishable photons and deterministic entanglement of distant spin qubits in a photonic crystal quantum network.

We present a magneto-photoluminescence study on neutral and charged excitons confined to InAs/GaAs quantum dots. Our investigation relies on a confocal microscope that allows arbitrary tuning of the angle between the applied magnetic field and the sample growth axis. First, from experiments on neutral excitons and trions, we extract the in-plane and on-axis components of the Land\'e tensor for electrons and holes in the s-shell. Then, based on the doubly negatively charged exciton magneto-photoluminescence we show that the p-electron wave function spreads significantly into the GaAs barriers. We also demonstrate that the p-electron g-factor depends on the presence of a hole in the s-shell. The magnetic field dependence of triply negatively charged excitons photoluminescence exhibits several anticrossings, as a result of coupling between the quantum dot electronic states and the wetting layer. Finally, we discuss how the system evolves from a Kondo-Anderson exciton description to the artificial atom model when the orientation of the magnetic field goes from Faraday to Voigt geometry.

The problem of computing saddle points is important in certain problems in numerical partial differential equations and computational chemistry, and is often solved numerically by a minimization problem over a set of mountain passes. We point out that a good global mountain pass algorithm should have good local and global properties. Next, we define the parallel distance, and show that the square of the parallel distance has a quadratic property. We show how to design algorithms for the mountain pass problem based on perturbing parameters of the parallel distance, and that methods based on the parallel distance have midrange local and global properties., Comment: 24 Pages, 5 Figures

The preparation of a coherent heavy-hole spin via ionization of a spin-polarized electron-hole pair in an InAs/GaAs quantum dot in a Voigt geometry magnetic field is investigated. For a dot with a 17 ueV bright-exciton fine-structure splitting, the fidelity of the spin preparation is limited to 0.75, with optimum preparation occurring when the effective fine-structure of the bright-exciton matches the in-plane hole Zeeman energy. In principle, higher fidelities can be achieved by minimizing the bright-exciton fine-structure splitting., Comment: 8 pages, 10 figs, published PRB 85 155310 (2012)

We demonstrate coherent optical control of a single hole spin confined to an InAs/GaAs quantum dot. A superposition of hole spin states is created by fast (10-100 ps) dissociation of a spin-polarized electron-hole pair. Full control of the hole-spin is achieved by combining coherent rotations about two axes: Larmor precession of the hole-spin about an external Voigt geometry magnetic field, and rotation about the optical-axis due to the geometric phase shift induced by a picosecond laser pulse resonant with the hole-trion transition., Comment: 6 pages, 3 figs, include supplement, final version

Preparation of a specific quantum state is a required step for a variety of proposed practical uses of quantum dynamics. We report an experimental demonstration of optical quantum state preparation in a semiconductor quantum dot with electrical readout, which contrasts with earlier work based on Rabi flopping in that the method is robust with respect to variation in the optical coupling. We use adiabatic rapid passage, which is capable of inverting single dots to a specified upper level. We demonstrate that when the pulse power exceeds a threshold for inversion, the final state is independent of power. This provides a new tool for preparing quantum states in semiconductor dots and has a wide range of potential uses., Comment: 4 pages, 4 figures

ROM-based quantum computation (QC) is an alternative to oracle-based QC. It has the advantages of being less ``magical'', and being more suited to implementing space-efficient computation (i.e. computation using the minimum number of writable qubits). Here we consider a number of small (one and two-qubit) quantum algorithms illustrating different aspects of ROM-based QC. They are: (a) a one-qubit algorithm to solve the Deutsch problem; (b) a one-qubit binary multiplication algorithm; (c) a two-qubit controlled binary multiplication algorithm; and (d) a two-qubit ROM-based version of the Deutsch-Jozsa algorithm. For each algorithm we present experimental verification using NMR ensemble QC. The average fidelities for the implementation were in the ranges 0.9 - 0.97 for the one-qubit algorithms, and 0.84 - 0.94 for the two-qubit algorithms. We conclude with a discussion of future prospects for ROM-based quantum computation. We propose a four-qubit algorithm, using Grover's iterate, for solving a miniature ``real-world'' problem relating to the lengths of paths in a network., Comment: 11 pages, 5 figures

A new approach to turbulence modeling in unsteady developed flows has recently been introduced, based on results of rapid distortion theory. The approach involves closing the k-epsilon equations for the organized unsteady component of the flow by modeling local unsteadiness as a rapid distortion of the local structure of the parent turbulent flow, in terms of an effective strain parameter alpha(sub eff). In this paper, the phase-conditioned equations of motion are developed to accommodate a new unsteady dissipation model and local effects of the slow-relaxation time scale of the parent flow. The model equations are tested against measurements of the response of a fully-developed turbulent pipe flow to the superposition of sinusoidal streamwise oscillation. Good agreement is found between measurements and predictions over a wide range of frequencies of unsteadiness, indicating that this approach may be particularly well suited to modeling of unsteady turbulent flows which are perturbations about a well characterized mean.

The behavior of a turbulent boundary layer which has been subjected to a local ramp-like deceleration in the external velocity field, which leads to forced separation, has been studied experimentally. The data of this study are re-interpreted in light of more recent findings concerning the temporal nature of boundary layer turbulence in the presence of forced unsteady shear. In particular, the robustness of the near-wall turbulent motions to organized deformation is recognized. Their resilence during unsteady shearing action promotes continued efficient turbulent mixing and rapid redistribution of turbulent kinetic energy during forced transients. In aerodynamic problems, the rapid nature of the adjustment of the turbulence field to a new temporal boundary condition necessitates equally rapid remedial measures to be taken if means of control/prevention of forced unsteady separation are to be deployed to maximum effect. This requirement suggests exploration of the use of simple, real-time statistical forecasting techniques, based upon time-series analysis of easily-measurable features of the flow, to help assure timely deployment of mechanisms of boundary-layer control. This paper focuses upon the nature of turbulence in boundary layers undergoing forced deceleration which would lead to separation. A preliminary form of a forecasting model is presented and evaluated. Using observations of the previous two large eddies passing a detector, it forecasts the behavior of the future large eddy rather well.

A proposed baseline design for the Space Infrared Telescope Facility includes a Tertiary Mirror Assembly (TMA) which selectively redirects the telescope's converging science beam to each of several instruments. The TMA's mirror rotates on an axis coincident with the beam's axis, and is held steady during observation by a kinematic mount. A bearing has been designed whose compliance causes minimal interference with the precision of the kinematic mount, and which is well suited to the particular requirements of a cryogenic satellite such as SIRTF. The bearing suspends its rotor by taking advantage of the repulsion between a superconductor and a magnet. It potentially eliminates problems associated with mechanical bearings that arise in similar applications, such as lubricant loss or failure, bearing wear, and sensitivity to particulates, and does so without imposing the thermal load of a bearing heater or active magnetic bearing. The bearing shows promise of offering an alternative to ball bearings in cryogenic applications where some compliance is acceptable or advantageous.

Selected data sets produced by the Stanford Unsteady Turbulent Boundary Layer Program are compared to analytic solutions of the Stokes equations for different kinds of imposed, organized unsteadiness. Good agreement is found, qualifying the facility as one that faithfully produces flows of different prescribed unsteady conditions. Data sets from the facility are therefore considered especially suitable for comparison with predictions of turbulence models for unsteady flows. New data describing the development of an unsteady separation bubble are presented.

The response of a turbulent boundary layer to abrupt changes in free-stream velocity is studied. Streamwise velocity measurements are examined during the abrupt events and after them, as the flow relaxes to its equilibrium condition. The measured velocity field is found to be in good agreement with the analytic solution of the Stokes equation for boundary conditions describing abrupt changes. While the flow recovers to its equilibrium state, changes in integral, boundary-layer parameters scale on the local free-stream time of flight.

Concepts, technology, and application of ultralightweight structures in space are examined. Gossamer spacecraft represented a generic class of space vehicles or structures characterized by a low mass per unit area (approximately 50g/m2). Gossamer concepts include the solar sail, the space tether, and various two and three dimensional large lightweight structures that were deployed or assembled in space. The Gossamer Spacecraft had a high potential for use as a transportation device (solar sail), as a science instrument (reflecting or occulting antenna), or as a large structural component for an enclosure, manned platform, or other human habitats. Inflatable structures were one possible building element for large ultralightweight structures in space.

The capability to specify new instrument/mechanism technology needs, for effective remote surface exploration, prospecting and assaying (EPA), requires first, an understanding of the functions or major elements of such a task, and second an understanding of the scientific instruments and support mechanisms that may be involved. An analog or task model was developed from which the various functions, operational procedures, scientific instruments, and support mechanisms for an automated mission could be derived. The task model led to the definition of nine major functions or categories of discrete operational elements that may have to be accomplished on a mission of this type. Each major function may stand alone as an element of an EPA mission, but more probably a major function will require the support of other functions, so they are inter-related.

Lunar sinuous rilles and Mojave desert stream channels similarities, discussing presence of terraces

Geophysical experiments for manned portion of future lunar roving vehicle missions

Venus/Mercury swingby with Venus capsule - preliminary science objectives and experiments for use in advanced mission studies