-
A simple stacked ensemble machine learning model to predict naturalized catchment hydrology and allocation status
Authors:
Michael J. Friedel,
Dave Stewart,
Xiao Feng Lu,
Pete Stevenson,
Helen Manly,
Tom Dyer
Abstract:
New Zealand legislation requires that Regional Councils set limits for water resource usage to manage the effects of abstractions in over-allocated catchments. We propose a simple stacked ensemble machine learning model to predict the probable naturalized hydrology and allocation status across 317 anthropogenically stressed gauged catchments and across 18,612 ungauged river reaches in Otago. The t…
▽ More
New Zealand legislation requires that Regional Councils set limits for water resource usage to manage the effects of abstractions in over-allocated catchments. We propose a simple stacked ensemble machine learning model to predict the probable naturalized hydrology and allocation status across 317 anthropogenically stressed gauged catchments and across 18,612 ungauged river reaches in Otago. The training and testing of ensemble machine learning models provides unbiased results characterized as very good (R2 > 0.8) to extremely good (R2 > 0.9) when predicting naturalized mean annual low flow and Mean flow. Statistical 5-fold stacking identifies varying levels of risk for managing water-resource sustainability in over-allocated catchments; for example, at the respective 5th, 25th, 50th, 75th, and 95th percentiles the number of overallocated catchments are 73, 57, 44, 23, and 22. The proposed model can be applied to inform sustainable stream management in other regional catchments across New Zealand and worldwide.
△ Less
Submitted 4 December, 2023;
originally announced December 2023.
-
Optical pum** enhancement of a free-induction-decay magnetometer
Authors:
Dominic Hunter,
Marcin S. Mrozowski,
Allan McWilliam,
Stuart J. Ingleby,
Terry E. Dyer,
Paul F. Griffin,
Erling Riis
Abstract:
Spin preparation prior to a free-induction-decay (FID) measurement can be adversely affected by transverse bias fields, particularly in the geophysical field range. A strategy that enhances the spin polarization accumulated before readout is demonstrated, by synchronizing optical pum** with a magnetic field pulse that supersedes any transverse fields by over two order of magnitude. The pulsed ma…
▽ More
Spin preparation prior to a free-induction-decay (FID) measurement can be adversely affected by transverse bias fields, particularly in the geophysical field range. A strategy that enhances the spin polarization accumulated before readout is demonstrated, by synchronizing optical pum** with a magnetic field pulse that supersedes any transverse fields by over two order of magnitude. The pulsed magnetic field is generated along the optical pum** axis using a compact electromagnetic coil pair encompassing a micro-electromechanical systems (MEMS) vapor cell. The coils also resistively heat the cesium (Cs) vapor to the optimal atomic density without spurious magnetic field contributions as they are rapidly demagnetized to approximately zero field during spin readout. The demagnetization process is analyzed electronically, and directly with a FID measurement, to confirm that the residual magnetic field is minimal during detection. The sensitivity performance of this technique is compared to existing optical pum** modalities across a wide magnetic field range. A noise floor sensitivity of $238\,\mathrm{fT/\surd{Hz}}$ was achieved in a field of approximately $\mathrm{50\,μ{T}}$, in close agreement with the Cramér-Rao lower bound (CRLB) predicted noise density of $258\,\mathrm{fT/\surd{Hz}}$.
△ Less
Submitted 29 September, 2023; v1 submitted 21 July, 2023;
originally announced July 2023.
-
Robustness of an Artificial Intelligence Solution for Diagnosis of Normal Chest X-Rays
Authors:
Tom Dyer,
Jordan Smith,
Gaetan Dissez,
Nicole Tay,
Qaiser Malik,
Tom Naunton Morgan,
Paul Williams,
Liliana Garcia-Mondragon,
George Pearse,
Simon Rasalingham
Abstract:
Purpose: Artificial intelligence (AI) solutions for medical diagnosis require thorough evaluation to demonstrate that performance is maintained for all patient sub-groups and to ensure that proposed improvements in care will be delivered equitably. This study evaluates the robustness of an AI solution for the diagnosis of normal chest X-rays (CXRs) by comparing performance across multiple patient…
▽ More
Purpose: Artificial intelligence (AI) solutions for medical diagnosis require thorough evaluation to demonstrate that performance is maintained for all patient sub-groups and to ensure that proposed improvements in care will be delivered equitably. This study evaluates the robustness of an AI solution for the diagnosis of normal chest X-rays (CXRs) by comparing performance across multiple patient and environmental subgroups, as well as comparing AI errors with those made by human experts.
Methods: A total of 4,060 CXRs were sampled to represent a diverse dataset of NHS patients and care settings. Ground-truth labels were assigned by a 3-radiologist panel. AI performance was evaluated against assigned labels and sub-groups analysis was conducted against patient age and sex, as well as CXR view, modality, device manufacturer and hospital site.
Results: The AI solution was able to remove 18.5% of the dataset by classification as High Confidence Normal (HCN). This was associated with a negative predictive value (NPV) of 96.0%, compared to 89.1% for diagnosis of normal scans by radiologists. In all AI false negative (FN) cases, a radiologist was found to have also made the same error when compared to final ground-truth labels. Subgroup analysis showed no statistically significant variations in AI performance, whilst reduced normal classification was observed in data from some hospital sites.
Conclusion: We show the AI solution could provide meaningful workload savings by diagnosis of 18.5% of scans as HCN with a superior NPV to human readers. The AI solution is shown to perform well across patient subgroups and error cases were shown to be subjective or subtle in nature.
△ Less
Submitted 31 August, 2022;
originally announced September 2022.
-
Enhancing Early Lung Cancer Detection on Chest Radiographs with AI-assistance: A Multi-Reader Study
Authors:
Gaetan Dissez,
Nicole Tay,
Tom Dyer,
Matthew Tam,
Richard Dittrich,
David Doyne,
James Hoare,
Jackson J. Pat,
Stephanie Patterson,
Amanda Stockham,
Qaiser Malik,
Tom Naunton Morgan,
Paul Williams,
Liliana Garcia-Mondragon,
Jordan Smith,
George Pearse,
Simon Rasalingham
Abstract:
Objectives: The present study evaluated the impact of a commercially available explainable AI algorithm in augmenting the ability of clinicians to identify lung cancer on chest X-rays (CXR).
Design: This retrospective study evaluated the performance of 11 clinicians for detecting lung cancer from chest radiographs, with and without assistance from a commercially available AI algorithm (red dot,…
▽ More
Objectives: The present study evaluated the impact of a commercially available explainable AI algorithm in augmenting the ability of clinicians to identify lung cancer on chest X-rays (CXR).
Design: This retrospective study evaluated the performance of 11 clinicians for detecting lung cancer from chest radiographs, with and without assistance from a commercially available AI algorithm (red dot, Behold.ai) that predicts suspected lung cancer from CXRs. Clinician performance was evaluated against clinically confirmed diagnoses.
Setting: The study analysed anonymised patient data from an NHS hospital; the dataset consisted of 400 chest radiographs from adult patients (18 years and above) who had a CXR performed in 2020, with corresponding clinical text reports.
Participants: A panel of readers consisting of 11 clinicians (consultant radiologists, radiologist trainees and reporting radiographers) participated in this study.
Main outcome measures: Overall accuracy, sensitivity, specificity and precision for detecting lung cancer on CXRs by clinicians, with and without AI input. Agreement rates between clinicians and performance standard deviation were also evaluated, with and without AI input.
Results: The use of the AI algorithm by clinicians led to an improved overall performance for lung tumour detection, achieving an overall increase of 17.4% of lung cancers being identified on CXRs which would have otherwise been missed, an overall increase in detection of smaller tumours, a 24% and 13% increased detection of stage 1 and stage 2 lung cancers respectively, and standardisation of clinician performance.
Conclusions: This study showed great promise in the clinical utility of AI algorithms in improving early lung cancer diagnosis and promoting health equity through overall improvement in reader performances, without impacting downstream imaging resources.
△ Less
Submitted 31 August, 2022;
originally announced August 2022.
-
A Digital Alkali Spin Maser
Authors:
Stuart Ingleby,
Paul Griffin,
Terry Dyer,
Marcin Mrozowski,
Erling Riis
Abstract:
Self-oscillating atomic magnetometers, in which the precession of atomic spins in a magnetic field is driven by resonant modulation, offer high sensitivity and dynamic range. Phase-coherent feedback from the detected signal to the applied modulation creates a resonant spin maser system, highly responsive to changes in the background magnetic field. Here we show a system in which the phase conditio…
▽ More
Self-oscillating atomic magnetometers, in which the precession of atomic spins in a magnetic field is driven by resonant modulation, offer high sensitivity and dynamic range. Phase-coherent feedback from the detected signal to the applied modulation creates a resonant spin maser system, highly responsive to changes in the background magnetic field. Here we show a system in which the phase condition for resonant precession is met by digital signal processing integrated into the maser feedback loop. This system uses a modest chip-scale laser and mass-produced dual-pass caesium vapour cell and operates in a 50 microtesla field, making it a suitable technology for portable measurements of the geophysical magnetic field. We demonstrate a Cramer-Rao lower bound-limited resolution of 50 fT at 1 s sampling cadence, and a sensor bandwidth of 10 kHz. This device also represents an important class of atomic system in which low-latency digital processing forms an integral part of a coherently-driven quantum system.
△ Less
Submitted 22 July, 2022;
originally announced July 2022.
-
Surfactancy in a tadpole model of proteins
Authors:
Oliver T. Dyer,
Robin C. Ball
Abstract:
We model the environment of eukaryotic nuclei by representing macromolecules by only their entropic properties, with globular molecules represented by spherical colloids and flexible molecules by polymers. We put particular focus on proteins with both globular and intrinsically disordered regions, which we represent with `tadpole' constructed by grafting single polymers and colloids together. In M…
▽ More
We model the environment of eukaryotic nuclei by representing macromolecules by only their entropic properties, with globular molecules represented by spherical colloids and flexible molecules by polymers. We put particular focus on proteins with both globular and intrinsically disordered regions, which we represent with `tadpole' constructed by grafting single polymers and colloids together. In Monte Carlo simulations we find these tadpoles support phase separation via depletion flocculation, and demonstrate several surfactant behaviours, including being found preferentially at interfaces and forming micelles in single phase solution. Furthermore, the model parameters can be tuned to give a tadpole a preference for either bulk phase. However, we find entropy too weak to drive these behaviours by itself at likely biological concentrations.
△ Less
Submitted 26 October, 2022; v1 submitted 11 April, 2022;
originally announced April 2022.
-
Time Reversal of the Overdamped Langevin Equation and Fixman's Law
Authors:
Robin C. Ball,
Oliver T. Dyer
Abstract:
We discuss how the first order Langevin equation for the overdamped dynamics of an interacting system has a natural time reversal of simple but surprising form, with consequences for correlation functions. This leads to the correlation of interactions as a strictly restraining term in the time-dependent diffusion tensor of the system, deriving the relation first suggested by Fixman. Applying this…
▽ More
We discuss how the first order Langevin equation for the overdamped dynamics of an interacting system has a natural time reversal of simple but surprising form, with consequences for correlation functions. This leads to the correlation of interactions as a strictly restraining term in the time-dependent diffusion tensor of the system, deriving the relation first suggested by Fixman. Applying this to the time-dependent diffusion of dilute polymer coils leads to the quantitative calibration of Kirkwood's approximation for their hydrodynamic radius. We find the generalized ``Fixman Law" for dissipation with a memory kernel, which has revealing causal structure, and we also discuss the case of the second order Langevin Equation.
△ Less
Submitted 25 June, 2021;
originally announced June 2021.
-
Influence of thermal fluctuations on active diffusion at large Péclet numbers
Authors:
Oliver T. Dyer,
Robin C. Ball
Abstract:
Wavelet Monte Carlo dynamics simulations are used to study the dynamics of passive particles in the presence of microswimmers, taking account of the often-omitted thermal motion alongside the hydrodynamic flows generated by the swimmers. Although the Péclet numbers considered are large, we find the thermal motion to have a significant effect on the dynamics of our passive particles, and can be inc…
▽ More
Wavelet Monte Carlo dynamics simulations are used to study the dynamics of passive particles in the presence of microswimmers, taking account of the often-omitted thermal motion alongside the hydrodynamic flows generated by the swimmers. Although the Péclet numbers considered are large, we find the thermal motion to have a significant effect on the dynamics of our passive particles, and can be included as a decorrelation factor in the velocity autocorrelation with a decay time proportional to the Péclet number. Similar decorrelation factors come from swimmer rotations, e.g.~run and tumble motion, and apply to both entrainment and far field loop contributions. These decorrelation factors lead to active diffusivity having a weak apparent power law close to $Pe^{0.2}$ for small tracer-like particles at Péclet numbers appropriate for E. coli swimmers at room temperature. Meanwhile, the reduced hydrodynamic response of large particles to nearby forces has a corresponding reduction in active diffusivity in that regime. Together, they lead to a non-monotonic dependence of active diffusivity on particle size that can shed light on similar behaviour observed in experiments by Patteson et al.
△ Less
Submitted 17 March, 2021;
originally announced March 2021.
-
Resonant Very Low- and Ultra Low Frequency Digital Signal Reception Using a Portable Atomic Magnetometer
Authors:
Stuart J. Ingleby,
Iain C. Chalmers,
Terry E. Dyer,
Paul F. Griffin,
Erling Riis
Abstract:
Radio communication through attenuating media necessitates the use of very-low frequency (VLF) and ultra-low frequency (ULF) carrier bands, which are frequently used in underwater and under-ground communication applications. Quantum sensing techniques can be used to circumvent hard constraints on the size, weight and noise floor of classical signal transducers. In this low-frequency range, an opti…
▽ More
Radio communication through attenuating media necessitates the use of very-low frequency (VLF) and ultra-low frequency (ULF) carrier bands, which are frequently used in underwater and under-ground communication applications. Quantum sensing techniques can be used to circumvent hard constraints on the size, weight and noise floor of classical signal transducers. In this low-frequency range, an optically pumped atomic sample can be used to detect carrier wave modulation resonant with ground-state Zeeman splitting of alkali atoms. Using a compact, self-calibrating system we demonstrate a resonant atomic transducer for digital data encoded using binary phase- and frequency-keying of resonant carrier waves in the 200 Hz -200 kHz range. We present field trial data showing sensor noise floor, decoded data and received bit error rate, and calculate the projected range of sub-sea communication using this device.
△ Less
Submitted 6 March, 2020;
originally announced March 2020.
-
Subpicotesla Scalar Atomic Magnetometer with a Microfabricated Cell
Authors:
Rui Zhang,
Terry Dyer,
Nathan Brockie,
Roozbeh Parsa,
Rahul Mhaskar
Abstract:
We demonstrated a scalar atomic magnetometer using a micro-fabricated Cs vapor cell. The atomic spin precession is driven by an amplitude-modulated circularly-polarized pump laser resonant on D1 transition of Cs atoms and detected by an off-resonant linearly-polarized probe laser using a balanced polarimeter setup. Under a magnetic field with amplitude in the Earth's magnetic field range, the magn…
▽ More
We demonstrated a scalar atomic magnetometer using a micro-fabricated Cs vapor cell. The atomic spin precession is driven by an amplitude-modulated circularly-polarized pump laser resonant on D1 transition of Cs atoms and detected by an off-resonant linearly-polarized probe laser using a balanced polarimeter setup. Under a magnetic field with amplitude in the Earth's magnetic field range, the magnetometer in the gradiometer mode can reach sensitivities below 150 fT/rt-Hz, which shows that the magnetometer by itself can achieve sub-100 fT/rt-Hz sensitivities. In addition to its high sensitivity, the magnetometer has a bandwidth close to 1 kHz due to the broad magnetic resonance inside the small vapor cell. Our experiment suggests the feasibility of a portable, low-power and high-performance magnetometer which can be operated in the Earth's magnetic field. Such a device will greatly reduce the restrictions on the operating environment and expand the range of applications for atomic magnetometers, such as detection of nuclear magnetic resonance (NMR) in low magnetic fields.
△ Less
Submitted 3 October, 2019;
originally announced October 2019.
-
Wavelet Monte Carlo dynamics: a new algorithm for simulating the hydrodynamics of interacting Brownian particles
Authors:
Oliver T Dyer,
Robin C Ball
Abstract:
We develop a new algorithm for the Brownian dynamics of soft matter systems that evolves time by spatially correlated Monte Carlo moves. The algorithm uses vector wavelets as its basic moves and produces hydrodynamics in the low Reynolds number regime propagated according to the Oseen tensor. When small moves are removed the correlations closely approximate the Rotne-Prager tensor, itself widely u…
▽ More
We develop a new algorithm for the Brownian dynamics of soft matter systems that evolves time by spatially correlated Monte Carlo moves. The algorithm uses vector wavelets as its basic moves and produces hydrodynamics in the low Reynolds number regime propagated according to the Oseen tensor. When small moves are removed the correlations closely approximate the Rotne-Prager tensor, itself widely used to correct for deficiencies in Oseen. We also include plane wave moves to provide the longest range correlations, which we detail for both infinite and periodic systems. The computational cost of the algorithm scales competitively with the number of particles simulated, $N$, scaling as $N\ln N$ in homogeneous systems and as $N$ in dilute systems. In comparisons to established lattice Boltzmann and Brownian dynamics algorithms the wavelet method was found to be only a factor of order 1 times more expensive than the cheaper lattice Boltzmann algorithm in marginally semi-dilute simulations, while it is significantly faster than both algorithms at large $N$ in dilute simulations. We also validate the algorithm by checking it reproduces the correct dynamics and equilibrium properties of simple single polymer systems, as well as verifying the effect of periodicity on the mobility tensor.
△ Less
Submitted 6 March, 2017; v1 submitted 28 November, 2016;
originally announced November 2016.
-
Fast Genome-Wide QTL Association Map** on Pedigree and Population Data
Authors:
Hua Zhou,
John Blangero,
Thomas D. Dyer,
Kei-hang K. Chan,
Kenneth Lange,
Eric M. Sobel
Abstract:
Since most analysis software for genome-wide association studies (GWAS) currently exploit only unrelated individuals, there is a need for efficient applications that can handle general pedigree data or mixtures of both population and pedigree data. Even data sets thought to consist of only unrelated individuals may include cryptic relationships that can lead to false positives if not discovered an…
▽ More
Since most analysis software for genome-wide association studies (GWAS) currently exploit only unrelated individuals, there is a need for efficient applications that can handle general pedigree data or mixtures of both population and pedigree data. Even data sets thought to consist of only unrelated individuals may include cryptic relationships that can lead to false positives if not discovered and controlled for. In addition, family designs possess compelling advantages. They are better equipped to detect rare variants, control for population stratification, and facilitate the study of parent-of-origin effects. Pedigrees selected for extreme trait values often segregate a single gene with strong effect. Finally, many pedigrees are available as an important legacy from the era of linkage analysis. Unfortunately, pedigree likelihoods are notoriously hard to compute. In this paper we re-examine the computational bottlenecks and implement ultra-fast pedigree-based GWAS analysis. Kinship coefficients can either be based on explicitly provided pedigrees or automatically estimated from dense markers. Our strategy (a) works for random sample data, pedigree data, or a mix of both; (b) entails no loss of power; (c) allows for any number of covariate adjustments, including correction for population stratification; (d) allows for testing SNPs under additive, dominant, and recessive models; and (e) accommodates both univariate and multivariate quantitative traits. On a typical personal computer (6 CPU cores at 2.67 GHz), analyzing a univariate HDL (high-density lipoprotein) trait from the San Antonio Family Heart Study (935,392 SNPs on 1357 individuals in 124 pedigrees) takes less than 2 minutes and 1.5 GB of memory. Complete multivariate QTL analysis of the three time-points of the longitudinal HDL multivariate trait takes less than 5 minutes and 1.5 GB of memory.
△ Less
Submitted 19 December, 2014; v1 submitted 30 July, 2014;
originally announced July 2014.
