Loss Of Fidelity Research Articles

Micro-Heterogeneity of Antibody Molecules.

Therapeutic monoclonal antibodies (mAbs) are mostly of the IgG class and constitute highly efficacious biopharmaceuticals for a wide range of clinical indications. Full-length IgG mAbs are large proteins that are subject to multiple posttranslational modifications (PTMs) during biosynthesis, purification, or storage, resulting in micro-heterogeneity. The production of recombinant mAbs in nonhuman cell lines may result in loss of structural fidelity and the generation of variants having altered stability, biological activities, and/or immunogenic potential. Additionally, even fully human therapeutic mAbs are of unique specificity, by design, and, consequently, of unique structure; therefore, structural elements may be recognized as non-self by individuals within an outbred human population to provoke an anti-therapeutic/anti-drug antibody (ATA/ADA) response. Consequently, regulatory authorities require that the structure of a potential mAb drug product is comprehensively characterized employing state-of-the-art orthogonal analytical technologies; the PTM profile may define a set of critical quality attributes (CQAs) for the drug product that must be maintained, employing quality by design parameters, throughout the lifetime of the drug. Glycosylation of IgG-Fc, at Asn297 on each heavy chain, is an established CQA since its presence and fine structure can have a profound impact on efficacy and safety. The glycoform profile of serum-derived IgG is highly heterogeneous while mAbs produced in mammalian cells in vitro is less heterogeneous and can be "orchestrated" depending on the cell line employed and the culture conditions adopted. Thus, the gross structure and PTM profile of a given mAb, established for the drug substance gaining regulatory approval, have to be maintained for the lifespan of the drug. This review outlines our current understanding of common PTMs detected in mAbs and endogenous IgG and the relationship between a variant's structural attribute and its impact on clinical performance.

Uncertainty-Aware Blind Image Quality Assessment in the Laboratory and Wild.

Performance of blind image quality assessment (BIQA) models has been significantly boosted by end-to-end optimization of feature engineering and quality regression. Nevertheless, due to the distributional shift between images simulated in the laboratory and captured in the wild, models trained on databases with synthetic distortions remain particularly weak at handling realistic distortions (and vice versa). To confront the cross-distortion-scenario challenge, we develop a unified BIQA model and an approach of training it for both synthetic and realistic distortions. We first sample pairs of images from individual IQA databases, and compute a probability that the first image of each pair is of higher quality. We then employ the fidelity loss to optimize a deep neural network for BIQA over a large number of such image pairs. We also explicitly enforce a hinge constraint to regularize uncertainty estimation during optimization. Extensive experiments on six IQA databases show the promise of the learned method in blindly assessing image quality in the laboratory and wild. In addition, we demonstrate the universality of the proposed training strategy by using it to improve existing BIQA models.

Designing excitonic circuits for the Deutsch-Jozsa algorithm: mitigating fidelity loss by merging gate operations.

In this manuscript, we examine design strategies for the development of excitonic circuits that are capable of performing simple 2-qubit multi-step quantum algorithms. Specifically, we compare two different strategies for designing dye-based systems that prescribe exciton evolution encoding a particular quantum algorithm. A serial strategy implements the computation as a step-by-step series of circuits, with each carrying out a single operation of the quantum algorithm, and a combined strategy implements the entire computation in a single circuit. We apply these two approaches to the well-studied Deutsch-Jozsa algorithm and evaluate circuit fidelity in an idealized system under a model harmonic bath, and also for a bath that is parameterized to reflect the thermal fluctuations of an explicit molecular environment. We find that the combined strategy tends to yield higher fidelity and that the harmonic bath approximation leads to lower fidelity than a model molecular bath. These results imply that the programming of excitonic circuits for quantum computation should favor hard-coded modules that incorporate multiple algorithmic steps and should represent the molecular nature of the circuit environment.

Scalable Fabrication and Metrology of Silicon Nanowire Arrays Made by Metal Assisted Chemical Etch

A scalable fabrication technique for silicon nanowires based on integrating nanoimprint lithography, metal assisted chemical etching (MACE), and spectroscopic scatterometry is presented in this article. The resulting wafer-scale process has demonstrated reliable and repeatable fabrication of high aspect ratio silicon nanostructures, and can provide cost-effective fabrication to enable applications in electronics, energy, point-of-use healthcare and sensing. Traditional pattern transfer using plasma etching suffers from etch taper and loss of feature fidelity at high aspect ratios, unlike MACE. However, MACE has largely been demonstrated in the literature over small areas, with scarce information on full wafer etching, critical dimension control, and etch depth uniformity. In this paper, we demonstrate a 100mm wafer high yield process to fabricate silicon nanowires. A large-area characterization of the process has been developed using imaging spectroscopic scatterometry. This scatterometry technique provides full wafer data on critical dimension control, etch depth uniformity, and presence of nanowire collapse defects. This work shows the promise of MACE as a next generation etch technology.

The effect of age on the acquisition and selection of cancer driver mutations in sun-exposed normal skin

The accumulation of somatic mutations contributes to ageing and cancer. Sunlight is the principal aetiological factor associated with skin cancer development. However, genetic and phenotypic factors also contribute to skin cancer risk. This study aimed at exploring the role of photoaging, as well as other well-known epidemiological risk factors, in the accumulation of somatic mutations in cancer-free human epidermis. We deeply sequenced 46 genes in normal skin biopsies from 123 healthy donors, from which phenotypic data (including age, pigmentation-related genotype and phenotype) and sun exposure habits were collected. We determined the somatic mutational burden, mutational signatures, clonal selection and frequency of driver mutations in all samples. Our results reveal an exponential accumulation of UV-related somatic mutations with age, matching skin cancer incidence. The increase of mutational burden is in turn modified by an individual's skin phototype. Somatic mutations preferentially accumulated in cutaneous squamous cell carcinoma cancer genes and clonally expanded with age, with distinct mutational processes underpinning different age groups. Our results suggest a loss of fidelity in transcription-coupled repair later in life. Our findings reveal that ageing is not only associated with an exponential increase in the number of somatic mutations accumulated in normal epidermis, but also with selection and expansion of cancer-associated mutations. Aged, sun-exposed normal skin is thus an extended mosaic of multiple clones with driver mutations, poised for the acquisition of transforming events.

Development of a multicellular 3D-bioprinted microtissue model of human periodontal ligament-alveolar bone biointerface: Towards a pre-clinical model of periodontal diseases and personalized periodontal tissue engineering

While periodontal (PD) disease is among principal causes of tooth loss worldwide, regulation of concomitant soft and mineralized PD tissues, and PD pathogenesis have not been completely clarified yet. Besides, relevant pre-clinical models and in vitro platforms have limitations in simulating human physiology. Here, we have harnessed three-dimensional bioprinting (3DBP) technology for developing a multi-cellular microtissue model resembling PD ligament-alveolar bone (PDL-AB) biointerface for the first time. 3DBP parameters were optimized; the physical, chemical, rheological, mechanical, and thermal properties of the constructs were assessed. Constructs containing gelatin methacryloyl (Gel-MA) and hydroxyapatite-magnetic iron oxide nanoparticles showed higher level of compressive strength when compared with that of Gel-MA constructs. Bioprinted self-supporting microtissue was cultured under flow in a microfluidic platform for >10 days without significant loss of shape fidelity. Confocal microscopy analysis indicated that encapsulated cells were homogenously distributed inside the matrix and preserved their viability for >7 days under microfluidic conditions. Immunofluorescence analysis showed the cohesion of stromal cell surface marker-1+ human PDL fibroblasts containing PDL layer with the osteocalcin+ human osteoblasts containing mineralized layer in time, demonstrating some permeability of the printed constructs to cell migration. Preliminary tetracycline interaction study indicated the uptake of model drug by the cells inside the 3D-microtissue. Also, the non-toxic levels of tetracycline were determined for the encapsulated cells. Thus, the effects of tetracyclines on PDL-AB have clinical significance for treating PD diseases. This 3D-bioprinted multi-cellular periodontal/osteoblastic microtissue model has potential as an in vitro platform for studying processes of the human PDL.

Non-Random Sister Chromatid Segregation in Human Tissue Stem Cells

The loss of genetic fidelity in tissue stem cells is considered a significant cause of human aging and carcinogenesis. Many cellular mechanisms are well accepted for limiting mutations caused by replication errors and DNA damage. However, one mechanism, non-random sister chromatid segregation, remains controversial. This atypical pattern of chromosome segregation is restricted to asymmetrically self-renewing cells. Though first confirmed in murine cells, non-random segregation was originally proposed by Cairns as an important genetic fidelity mechanism in human tissues. We investigated human hepatic stem cells expanded by suppression of asymmetric cell kinetics (SACK) for evidence of non-random sister chromatid segregation. Cell kinetics and time-lapse microscopy analyses established that an ex vivo expanded human hepatic stem cell strain possessed SACK agent-suppressible asymmetric cell kinetics. Complementary DNA strand-labeling experiments revealed that cells in hepatic stem cell cultures segregated sister chromatids non-randomly. The number of cells cosegregating sister chromatids with the oldest “immortal DNA strands” was greater under conditions that increased asymmetric self-renewal kinetics. Detection of this mechanism in a human tissue stem cell strain increases support for Cairns’ proposal that non-random sister chromatid segregation operates in human tissue stem cells to limit carcinogenesis.

Single-image deblurring with neural networks: A comparative survey

Neural networks (NNs) are becoming the tool of choice for sharpening blurred images. We discuss and categorize deblurring NNs. Then we evaluate seven NNs for non-blind deblurring (NBD), and seven NNs and four optimization techniques for blind deblurring (BD). To do this we use several current datasets containing pairs of sharp and blurred images, synthesized either by convolving sharp images with blur kernels or by averaging consecutive sharp images, so as to produce both uniform and non-uniform blurs. We also introduce a newly reorganized benchmark dataset in which blurred images have been classified using attributes that depend on the extent of the blur. We use this dataset to compare the effectiveness of single and multi-scale training in coping with large blurs. On NBD, NNs that use regularization with a denoising prior network outperform other denoising NNs; and NNs that use a deep image prior network outperform other deconvolution NNs. On BD, NNs outperform optimizations in signal-difference terms, but not in terms of perceptual fidelity. We found that multi-scale training helps NNs to deal with large blurs, and RNNs outperform CNNs. We also observed that GANs using a perceptual loss function produce artifacts; but also that some form of perceptual fidelity loss is required to get the best results from NNs. We contend that the domain bias of current datasets works against robustness and generality. And we discuss the potential of more sophisticated perceptual loss functions, attention techniques, and unsupervised learning.

Increased PD-L1 Expression in Acute Myeloid Leukemia with Fanconi Anemia/BRCA Mutations

Introduction: Loss of DNA repair fidelity is a common feature of tumors and can drive genomic instability, tumor evolution, and therapy response. 1 Fanconi anemia (FA) is a DNA damage response (DDR) pathway and biallelic germline mutations in the pathway lead to a bone marrow failure syndrome with a predisposition for cancer, especially acute myeloid leukemia (AML). Unfortunately, FA-AML is a secondary AML with very poor prognosis, with an increased risk for relapse even after initial cure with bone marrow transplantation. Therefore FA patients need further options for therapy. DNA damage response (DDR) deficiency has emerged as a predictive biomarker of response to immune checkpoint inhibition. 2 In some solid tumors, defective DDR leads to genomic instability, increased mutational burden, increased neo antigen expression and increased PD-L1 expression; and respond robustly to PD-1 checkpoint inhibitor therapy. The mutational burden and PD-L1 expression in FA-mutated tumors remain to be discovered. Therefore, we hypothesize that FA mutations in AML increase mutational burden and PD-L1 expression. This work can set the stage for future studies involving PD-1 blocking therapies for FA-mutated tumors. Methods/Results: To determine the mutational burden in FA-mutated tumors, we didwhole exome sequencing (WES)of three AML cell lines obtained from pediatric FA patients, all were FANCD1/BRCA2 mutated.3,4 The three AML cell lines also represent a progression of AML disease- an early clone, a late clone and a relapsed AML after bone marrow transplantation. There was an average of 180,000 somatic mutations/case, pared down to 1,000 mutations after limiting calls with >10 reads and allele frequency >0.1%. Mutational burden of pediatric AML in non-FA patients has been reported to be ~10 mutations/case.5Therefore there is about a 100-fold increase in mutations in AML from FA patients compared to AML from non-FA patients.Then we determined PD-L1 mRNA and protein expression of the three FANCD1/BRCA2 mutated AML cells with quantitative PCR and flow cytometry respectively. We found increasing PD-L1 mRNA expression from early to late to relapsed clones (2^-ΔCt values 1.21x10-5, 1.54x10-5, and 7.02x10-5 respectively; p-value<0.05). Interestingly, interferon-gamma (IFN-g), a known inducer and regulator of PD-L1 expression, is detected at high levels in the bone marrow of FA patients.5 We find that PD-L1 expression significantly increases in FANCD1/BRCA2 mutated AML cells after treatment with physiologic doses of IFN-g (fold change in 2^-ΔΔCt before and after treatment with 5ng IFN-g for early clone (x16.55 fold change) late clone (x38.46 fold change) and relapsed clone (x17.57 fold change), p-value<0.05. We also found stable PD-L1 protein expression on the cell surface (% PD-L1-positive cells for early clone (15.2%), late clone (17.7%), and relapsed AML (18.9%). Conclusions: Our data show there is a high mutational load in FA-AML cell lines, which also express PD-L1. Both of these tumor characteristics are predictors of response to PD-1 checkpoint inhibitor immunotherapy. This will lead to future work on PD-1 blockade as a potential therapeutic option for FA-AML. Disclosures No relevant conflicts of interest to declare.

Improving Color Space Conversion for Camera-Captured Images via Wide-Gamut Metadata

Color space conversion is the process of converting color values in an image from one color space to another. Color space conversion is challenging because different color spaces have different sized gamuts. For example, when converting an image encoded in a medium-sized color gamut (e.g., AdobeRGB or Display-P3) to a small color gamut (e.g., sRGB), color values may need to be compressed in a many-to-one manner (i.e., multiple colors in the source gamut will map to a single color in the target gamut). If we try to convert this sRGB-encoded image back to a wider gamut color encoding, it can be challenging to recover the original colors due to the color fidelity loss. We propose a method to address this problem by embedding wide-gamut metadata inside saved images captured by a camera. Our key insight is that in the camera hardware, a captured image is converted to an intermediate wide-gamut color space (i.e., ProPhoto) as part of the processing pipeline. This wide-gamut image representation is then saved to a display color space and saved in an image format such as JPEG or HEIC. Our method proposes to include a small sub-sampling of the color values from the ProPhoto image state in the camera to the final saved JPEG/HEIC image. We demonstrate that having this additional wide-gamut metadata available during color space conversion greatly assists in constructing a color mapping function to convert between color spaces. Our experiments show our metadata-assisted color mapping method provides a notable improvement (up to 60% in terms of E) over conventional color space methods using perceptual rendering intent. In addition, we show how to extend our approach to perform adaptive color space conversion based spatially over the image for additional improvements.

Promiscuous DNA cleavage by HpyAII endonuclease is modulated by the HNH catalytic residues.

Helicobacter pylori is a carcinogenic bacterium that is responsible for 5.5% of all human gastric cancers. H. pylori codes for an unusually large number of restriction–modification (R–M) systems and several of them are strain-specific and phase-variable. HpyAII is a novel Type IIs phase-variable restriction endonuclease present in 26695 strain of H. pylori. We show that HpyAII prefers two-site substrates over one-site substrates for maximal cleavage activity. HpyAII is less stringent in metal ion requirement and shows higher cleavage activity with Ni2+ over Mg2+. Mutational analysis of the putative residues of the HNH motif of HpyAII confirms that the protein has an active HNH site for the cleavage of DNA. However, mutation of the first Histidine residue of the HNH motif to Alanine does not abolish the enzymatic activity, but instead causes loss of fidelity compared with wildtype HpyAII. Previous studies have shown that mutation of the first Histidine residue of the HNH motif of all other known HNH motif motif-containing enzymes completely abolishes enzymatic activity. We found, in the case of HpyAII, mutation of an active site residue leads to the loss of endonuclease fidelity. The present study provides further insights into the evolution of restriction enzymes.

Resonator-assisted single-molecule quantum state detection

We propose a state-sensitive scheme to optically detect a single molecule without a closed transition, through strong coupling to a high-Q whispering-gallery mode high-Q resonator. A background-free signal can be obtained by detecting a molecule-induced transparency in a photon bus waveguide that is critically coupled to the resonator, with a suppressed depumping rate to other molecular states by the cooperativity parameter $C$. We numerically calculate the dynamics of the molecule-resonator coupled system using Lindblad master equations, and develop analytical solutions through the evolution of quasi-steady states in the weak-driving regime. Using Rb$_2$ triplet ground state molecules as an example, we show that high fidelity state readout can be achieved using realistic resonator parameters. We further discuss the case of multiple molecules collectively coupled to a resonator, demonstrating near-unity detection fidelity and negligible population loss.

Adversarial transfer learning for cross-domain visual recognition

In many practical visual recognition scenarios, feature distributions between source domain and the target domain are quite different, which results in the emergence of general cross-domain visual recognition problems. To address the problems of visual domain mismatch, we propose a novel shallow semi-supervised adversarial transfer learning network, which is called Coupled adversarial transfer Domain Adaptation (CatDA), for distribution alignment between two domains. The proposed CatDA approach is inspired by cycleGAN, but leveraging multiple shallow multilayer perceptrons (MLPs) instead of deep networks. Specifically, our CatDA comprises of two symmetric and slim sub-networks, such that the coupled adversarial learning framework is formulated. With such symmetry of two generators, the input data from source/target domain can be fed into the MLP network for target/source domain generation, supervised by two confrontation oriented coupled discriminators. Notably, in order to avoid the critical flaw of high-capacity of the feature extraction function during domain adversarial training, domain specific loss and domain knowledge fidelity loss are proposed in each generator, such that the effectiveness of the proposed transfer network is guaranteed. Additionally, the essential difference from cycleGAN is that our method aims to generate domain-agnostic and aligned features for domain adaptation and transfer learning rather than synthesize realistic images. We show experimentally on a number of benchmark datasets and the proposed approach achieves competitive performance over state-of-the-art domain adaptation and transfer learning approaches.

Comparison of Mobile and Fixed-Site Black Carbon Measurements for High-Resolution Urban Pollution Mapping.

Urban concentrations of black carbon (BC) and other primary pollutants vary on small spatial scales (<100m). Mobile air pollution measurements can provide information on fine-scale spatial variation, thereby informing exposure assessment and mitigation efforts. However, the temporal sparsity of these measurements presents a challenge for estimating representative long-term concentrations. We evaluate the capabilities of mobile monitoring in the represention of time-stable spatial patterns by comparing against a large set of continuous fixed-site measurements from a sampling campaign in West Oakland, California. Custom-built, low-cost aerosol black carbon detectors (ABCDs) provided 100 days of continuous measurements at 97 near-road and 3 background fixed sites during summer 2017; two concurrently operated mobile laboratories collected over 300 h of in-motion measurements using a photoacoustic extinctiometer. The spatial coverage from mobile monitoring reveals patterns missed by the fixed-site network. Time-integrated measurements from mobile lab visits to fixed-site monitors reveal modest correlation (spatial R2 = 0.51) with medians of full daytime fixed-site measurements. Aggregation of mobile monitoring data in space and time can mitigate high levels of uncertainty associated with measurements at precise locations or points in time. However, concentrations estimated by mobile monitoring show a loss of spatial fidelity at spatial aggregations greater than 100 m.

Subsampled brain MRI reconstruction by generative adversarial neural networks.

A main challenge in magnetic resonance imaging (MRI) is speeding up scan time. Beyond improving patient experience and reducing operational costs, faster scans are essential for time-sensitive imaging, such as fetal, cardiac, or functional MRI, where temporal resolution is important and target movement is unavoidable, yet must be reduced. Current MRI acquisition methods speed up scan time at the expense of lower spatial resolution and costlier hardware. We introduce a practical, software-only framework, based on deep learning, for accelerating MRI acquisition, while maintaining anatomically meaningful imaging. This is accomplished by MRI subsampling followed by estimating the missing k-space samples via generative adversarial neural networks. A generator-discriminator interplay enables the introduction of an adversarial cost in addition to fidelity and image-quality losses used for optimizing the reconstruction. Promising reconstruction results are obtained from feasible sampling patterns of up to a fivefold acceleration of diverse brain MRIs, from a large publicly available dataset of healthy adult scans as well as multimodal acquisitions of multiple sclerosis patients and dynamic contrast-enhanced MRI (DCE-MRI) sequences of stroke and tumor patients. Clinical usability of the reconstructed MRI scans is assessed by performing either lesion or healthy tissue segmentation and comparing the results to those obtained by using the original, fully sampled images. Reconstruction quality and usability of the DCE-MRI sequences is demonstrated by calculating the pharmacokinetic (PK) parameters. The proposed MRI reconstruction approach is shown to outperform state-of-the-art methods for all datasets tested in terms of the peak signal-to-noise ratio (PSNR), the structural similarity index (SSIM), as well as either the mean squared error (MSE) with respect to the PK parameters, calculated for the fully sampled DCE-MRI sequences, or the segmentation compatibility, measured in terms of Dice scores and Hausdorff distance. The code is available on GitHub.

Culicidae-centric metabarcoding through targeted use of D2 ribosomal DNA primers.

A practical limitation to many metabarcoding initiatives is that sampling methods tend to collect many non-target taxa, which become “amplicon noise” that can saturate Next Generation Sequencing results and lead to both financial and resource inefficiencies. An available molecular tool that can significantly decrease these non-target amplicons and decrease the need for pre-DNA-extraction sorting of bycatch is the design of PCR primers tailored to the taxa under investigation. We assessed whether the D2 extension segment of the 28S ribosomal operon can limit this shortcoming within the context of mosquito (Culicidae) monitoring. We designed PCR primers that are fully conserved across mosquitos and exclude from amplification most other taxa likely to be collected with current sampling apparatuses. We show that, given enough sequencing depth, D2 is an effective marker for the detection of mosquito sequences within mock genomic DNA pools. As few as 3,050 quality-filtered Illumina reads were able to recover all 17 species in a bulk pool containing as little as 0.2% of constituent DNA from single taxa. We also mixed these mosquito DNA pools with high concentrations of non-Culicidae bycatch DNA and show that the component mosquito species are generally still recoverable and faithful to their original relative frequencies. Finally, we show that there is little loss of fidelity in abundance parameters when pools from degraded DNA samples were sequenced using the D2 primers.

Rapid Reconstruction of Four-dimensional MR Angiography of the Thoracic Aorta Using a Convolutional Neural Network.

To implement an integrated reconstruction pipeline including a graphics processing unit (GPU)-based convolutional neural network (CNN) architecture and test whether it reconstructs four-dimensional non-Cartesian, non-contrast material-enhanced MR angiographic k-space data faster than a central processing unit (CPU)-based compressed sensing (CS) reconstruction pipeline, without significant losses in data fidelity, summed visual score (SVS), or arterial vessel-diameter measurements. Raw k-space data of 24 patients (18 men and six women; mean age, 56.8 years ± 11.8 [standard deviation]) suspected of having thoracic aortic disease were used to evaluate the proposed reconstruction pipeline derived from an open-source three-dimensional CNN. For training, 4800 zero-filled images and the corresponding CS-reconstructed images from 10 patients were used as input-output pairs. For testing, 6720 zero-filled images from 14 different patients were used as inputs to a trained CNN. Metrics for evaluating the agreement between the CNN and CS images included reconstruction times, structural similarity index (SSIM) and normalized root-mean-square error (NRMSE), SVS (3 = nondiagnostic, 9 = clinically acceptable, 15 = excellent), and vessel diameters. The mean reconstruction time was 65 times and 69 times shorter for the CPU-based and GPU-based CNN pipelines (216.6 seconds ± 40.5 and 204.9 seconds ± 40.5), respectively, than for CS (14 152.3 seconds ± 1708.6) (P < .001). Compared with CS as practical ground truth, CNNs produced high data fidelity (SSIM = 0.94 ± 0.02, NRMSE = 2.8% ± 0.4) and not significantly different (P = .25) SVS and aortic diameters, except at one out of seven locations, where the percentage difference was only 3% (ie, clinically irrelevant). The proposed integrated reconstruction pipeline including a CNN architecture is capable of rapidly reconstructing time-resolved volumetric cardiovascular MRI k-space data, without a significant loss in data quality, thereby supporting clinical translation of said non-contrast-enhanced MR angiograms. Supplemental material is available for this article. © RSNA, 2020.

Retrieving Quantum Information with Active Learning.

Active learning is a machine learning method aiming at optimal design for model training. At variance with supervised learning, which labels all samples, active learning provides an improved model by labeling samples with maximal uncertainty according to the estimation model. Here, we propose the use of active learning for efficient quantum information retrieval, which is a crucial task in the design of quantum experiments. Meanwhile, when dealing with large data output, we employ active learning for the sake of classification with minimal cost in fidelity loss. Indeed, labeling only 5% samples, we achieve almost 90% rate estimation. The introduction of active learning methods in the data analysis of quantum experiments will enhance applications of quantum technologies.

Generation and evaluation of privacy preserving synthetic health data

We develop metrics for measuring the quality of synthetic health data for both education and research. We use novel and existing metrics to capture a synthetic dataset’s resemblance, privacy, utility and footprint. Using these metrics, we develop an end-to-end workflow based on our generative adversarial network (GAN) method, HealthGAN, that creates privacy preserving synthetic health data. Our workflow meets privacy specifications of our data partner: (1) the HealthGAN is trained inside a secure environment; (2) the HealthGAN model is used outside of the secure environment by external users to generate synthetic data. This second step facilitates data handling for external users by avoiding de-identification, which may require special user training, be costly, or cause loss of data fidelity. This workflow is compared against five other baseline methods. While maintaining resemblance and utility comparable to other methods, HealthGAN provides the best privacy and footprint. We present two case studies in which our methodology was put to work in the classroom and research settings. We evaluate utility in the classroom through a data analysis challenge given to students and in research by replicating three different medical papers with synthetic data. Data, code, and the challenge that we organized for educational purposes are available.

Outcomes associated with scale-up of the Stepping On falls prevention program: A case study in redesigning for dissemination

Translating complex behavior change interventions into practice can be accompanied by a loss of fidelity and effectiveness. We present the evaluation of two sequential phases of implementation of a complex evidence-based community workshop to reduce falls, using the Replicating Effective Programs Framework. Between the two phases, workshop training and delivery were revised to improve fidelity with key elements. Stepping On program participants completed a questionnaire at baseline (phase 1: n = 361; phase 2: n = 2219) and 6 months post-workshop (phase 1: n = 232; phase 2: n = 1281). Phase 2 participants had an additional follow-up at 12 months (n = 883). Outcomes were the number of falls in the prior 6 months and the Falls Behavioral Scale (FaB) score. Workshop participation in phase 1 was associated with a 6% reduction in falls (RR = 0.94, 95% CI 0.74-1.20) and a 0.14 improvement in FaB score (95% CI, 0.11- 0.18) at 6 months. Workshop participation in phase 2 was associated with a 38% reduction in falls (RR = 0.62, 95% CI 0.57-0.68) and a 0.16 improvement in FaB score (95% CI 0.14-0.18) at 6 months, and a 28% reduction in falls (RR = 0.72, 95% CI 0.65-0.80) and a 0.19 score improvement in FaB score (95% CI 0.17-0.21) at 12-month follow-up. Effectiveness can be maintained with widespread dissemination of a complex behavior change intervention if attention is paid to fidelity of key elements. An essential role for implementation science is to ensure effectiveness as programs transition from research to practice.