Learning Many-to-Many Mapping for Unpaired Real-World Image Super-resolution and Downscaling.

Learning based single image super-resolution (SISR) for real-world images has been an active research topic yet a challenging task, due to the lack of paired low-resolution (LR) and high-resolution (HR) training images. Most of the existing unsupervised real-world SISR methods adopt a twostage training strategy by synthesizing realistic LR images from their HR counterparts first, then training the super-resolution (SR) models in a supervised manner. However, the training of image degradation and SR models in this strategy are separate, ignoring the inherent mutual dependency between downscaling and its inverse upscaling process. Additionally, the ill-posed nature of image degradation is not fully considered. In this paper, we propose an image downscaling and SR model dubbed as SDFlow, which simultaneously learns a bidirectional manyto- many mapping between real-world LR and HR images unsupervisedly. The main idea of SDFlow is to decouple image content and degradation information in the latent space, where content information distribution of LR and HR images is matched in a common latent space. Degradation information of the LR images and the high-frequency information of the HR images are fitted to an easy-to-sample conditional distribution. Experimental results on real-world image SR datasets indicate that SDFlow can generate diverse realistic LR and SR images both quantitatively and qualitatively.

Model-based bioequivalence approach for sparse pharmacokinetic bioequivalence studies: Model selection or model averaging?

Conventional pharmacokinetic (PK) bioequivalence (BE) studies aim to compare the rate and extent of drug absorption from a test (T) and reference (R) product using non-compartmental analysis (NCA) and the two one-sided test (TOST). Recently published regulatory guidance recommends alternative model-based (MB) approaches for BE assessment when NCA is challenging, as for long-acting injectables and products which require sparse PK sampling. However, our previous research on MB-TOST approaches showed that model misspecification can lead to inflated type I error. The objective of this research was to compare the performance of model selection (MS) on R product arm data and model averaging (MA) from a pool of candidate structural PK models in MBBE studies with sparse sampling. Our simulation study was inspired by a real case BE study using a two-way crossover design. PK data were simulated using three structural models under the null hypothesis and one model under the alternative hypothesis. MB-TOST was applied either using each of the five candidate models or following MS and MA with or without the simulated model in the pool. Assuming T and R have the same PK model, our simulation shows that following MS and MA, MB-TOST controls type I error rates at or below 0.05 and attains similar or even higher power than when using the simulated model. Thus, we propose to use MS prior to MB-TOST for BE studies with sparse PK sampling and to consider MA when candidate models have similar Akaike information criterion.

Safety outcomes when switching between biosimilars and reference biologics: A systematic review and meta-analysis.

Biosimilars are increasingly available for the treatment of many serious disorders, however some concerns persist about switching a patient to a biosimilar whose condition is stable while on the reference biologic. Randomized controlled studies and extension studies with a switch treatment period (STP) to or from a biosimilar and its reference biologic were identified from publicly available information maintained by the U.S. Food and Drug Administration (FDA). These findings were augmented with data from peer reviewed publications containing information not captured in FDA reviews. Forty-four STPs were identified from 31 unique studies for 21 different biosimilars. Data were extracted and synthesized following PRISMA guidelines. Meta-analysis was conducted to estimate the overall risk difference across studies. A total of 5,252 patients who were switched to or from a biosimilar and its reference biologic were identified. Safety data including deaths, serious adverse events, and treatment discontinuation showed an overall risk difference (95% CI) of -0.00 (-0.00, 0.00), 0.00 (-0.01, 0.01), -0.00 (-0.01, 0.00) across STPs, respectively. Immunogenicity data showed similar incidence of anti-drug antibodies and neutralizing antibodies in patients within a STP who were switched to or from a biosimilar to its reference biologic and patients who were not switched. Immune related adverse events such as anaphylaxis, hypersensitivity reactions, and injections site reactions were similar in switched and non-switched patients. This first systematic review using statistical methods to address the risk of switching patients between reference biologics and biosimilars finds no difference in the safety profiles or immunogenicity rates in patients who were switched and those who remained on a reference biologic or a biosimilar.

Evaluation of model-based bioequivalence approach for single sample pharmacokinetic studies.

In a traditional pharmacokinetic (PK) bioequivalence (BE) study, a two-way crossover study is conducted, PK parameters (namely the area under the time-concentration curve [AUC] and the maximal concentration [ C max ]) are obtained by noncompartmental analysis (NCA), and the BE analysis is performed using the two one-sided test (TOST) method. For ophthalmic drugs, however, only one sample of aqueous humor, in one eye, per eye can be obtained in each patient, which precludes the traditional BE analysis. To circumvent this issue, the U.S. Food and Drug Administration (FDA) has proposed an approach coupling NCA with either parametric or nonparametric bootstrap (NCA bootstrap). The model-based TOST (MB-TOST) has previously been proposed and evaluated successfully for various settings of sparse PK BE studies. In this paper, we evaluate, via simulations, MB-TOST in the specific setting of single sample PK BE study and compare its performance to NCA bootstrap. We performed BE study simulations using a published PK model and parameter values and evaluated multiple scenarios, including study design (parallel or crossover), sampling times (5 or 10 spread across the dosing interval), and geometric mean ratio (of 0.8, 0.9, 1, and 1.25). Using the simulated structural PK model, MB-TOST performed similarly to NCA bootstrap for AUC. For C max , the latter tended to be conservative and less powerful. Our research suggests that MB-TOST may be considered as an alternative BE approach for single sample PK studies, provided that the PK model is correctly specified and the test drug has the same structural model as the reference drug.

Impact of model misspecification on model-based tests in PK studies with parallel design: real case and simulation studies.

This article evaluates the performance of pharmacokinetic (PK) equivalence testing between two formulations of a drug through the Two-One Sided Tests (TOST) by a model-based approach (MB-TOST), as an alternative to the classical non-compartmental approach (NCA-TOST), for a sparse design with a few time points per subject. We focused on the impact of model misspecification and the relevance of model selection for the reference data. We first analysed PK data from phase I studies of gantenerumab, a monoclonal antibody for the treatment of Alzheimer's disease. Using the original rich sample data, we compared MB-TOST to NCA-TOST for validation. Then, the analysis was repeated on a sparse subset of the original data with MB-TOST. This analysis inspired a simulation study with rich and sparse designs. With rich designs, we compared NCA-TOST and MB-TOST in terms of type I error and study power. With both designs, we explored the impact of misspecifying the model on the performance of MB-TOST and adding a model selection step. Using the observed data, the results of both approaches were in general concordance. MB-TOST results were robust with sparse designs when the underlying PK structural model was correctly specified. Using the simulated data with a rich design, the type I error of NCA-TOST was close to the nominal level. When using the simulated model, the type I error of MB-TOST was controlled on rich and sparse designs, but using a misspecified model led to inflated type I errors. Adding a model selection step on the reference data reduced the inflation. MB-TOST appears as a robust alternative to NCA-TOST, provided that the PK model is correctly specified and the test drug has the same PK structural model as the reference drug.

Learning Discrete Representations From Reference Images for Large Scale Factor Image Super-Resolution.

Image super-resolution (SR) task aims to recover high-resolution (HR) images from degraded low-resolution (LR) images, which has achieved great progress due to the recent advances of deep neural networks. Due to severe information loss of the LR images, it is more challenging to reconstruct high quality HR images at large scale factors, i. e., higher than 4× . Traditional reference image based SR methods usually perform patch matching to locate detailed texture from HR reference images which could provide fine details from similar image contents. But it suffers from difficulties in achieving good matching in the largely downscaled image space or feature space due to the ill-posed nature between LR and HR mapping. In this paper, we tackle this problem by exploiting fine details contained in reference HR images. Inspired by vector quantization (VQ), we propose a simple yet effective auto-encoder convolutional neural network (CNN) module to learn discrete representations of images. Furthermore, we propose to progressively learn pairs of cross-scale discrete feature representations using paired LR and HR reference images. The coarser scale of the discrete representation is responsible for encoding the global image structure while the paired finer scale of the discrete representation takes charge of capturing missing details in the finer image scale. During inference, continuous features of the test LR image are used as queries to retrieve finer scale discrete representations (value) by searching the nearest coarser scale discrete representations (key). Then, the queries and retrieved values are combined to progressively recover the HR image. Experimental results indicate that when compared with the state-of-the-art image SR models, the proposed method can achieve advanced performance in terms of both objective quality and subjective quality. The code will be available on URL: https://github.com/sunwj/refsr.

A Bayesian population physiologically based pharmacokinetic absorption modeling approach to support generic drug development: application to bupropion hydrochloride oral dosage forms.

We propose a Bayesian population modeling and virtual bioequivalence assessment approach to establishing dissolution specifications for oral dosage forms. A generalizable semi-physiologically based pharmacokinetic absorption model with six gut segments and liver, connected to a two-compartment model of systemic disposition for bupropion hydrochloride oral dosage forms was developed. Prior information on model parameters for gut physiology, bupropion physicochemical properties, and drug product properties were obtained from the literature. The release of bupropion hydrochloride from immediate-, sustained- and extended-release oral dosage forms was described by a Weibull function. In vitro dissolution data were used to assign priors to the in vivo release properties of the three bupropion formulations. We applied global sensitivity analysis to identify the influential parameters for plasma bupropion concentrations and calibrated them. To quantify inter- and intra-individual variability, plasma concentration profiles in healthy volunteers that received the three dosage forms, each at two doses, were used. The calibrated model was in good agreement with both in vitro dissolution and in vivo exposure data. Markov Chain Monte Carlo samples from the joint posterior parameter distribution were used to simulate virtual crossover clinical trials for each formulation with distinct drug dissolution profiles. For each trial, an allowable range of dissolution parameters ("safe space") in which bioequivalence can be anticipated was established. These findings can be used to assure consistent product performance throughout the drug product life-cycle and to support manufacturing changes. Our framework provides a comprehensive approach to support decision-making in drug product development.

Visual Scanpath Prediction Using IOR-ROI Recurrent Mixture Density Network.

A visual scanpath represents the human eye movements when scanning the visual field for acquiring and receiving visual information. Predicting visual scanpaths when a certain stimulus is presented plays an important role in modeling overt human visual attention and search behavior. In this paper, we presented an 'Inhibition of Return - Region of Interest' (IOR-ROI) recurrent mixture density network based framework learning to produce human-like visual scanpaths under task-free viewing conditions. The proposed model simultaneously predicts a sequence of ordered fixation positions and their corresponding fixation durations. Our model integrates bottom-up features and semantic features extracted by convolutional neural networks. Then the integrated feature maps are fed into the IOR-ROI Long Short-Term Memory (LSTM) which is the core component of the proposed model. The IOR-ROI LSTM is a dual LSTM unit, i.e., the IOR-LSTM and the ROI-LSTM, capturing IOR dynamics and gaze shift behavior simultaneously. IOR-LSTM simulates the visual working memory to adaptively maintain and update visual information regarding previously fixated regions. ROI-LSTM is responsible for predicting the next possible ROIs given the spatially inhibited image feature maps on the feature-wise basis. Fixation duration is predicted by a regression neural network given the viewing history and image feature maps corresponding to currently fixated ROI. Considering the eye movement pattern variations among subjects, a mixture density network is adopted to model the next fixation distribution as Gaussian mixtures and the fixation duration is also modeled using Gaussian distribution. Our model is evaluated on the OSIE and MIT low resolution eye-tracking datasets and experimental results indicate that the proposed method can achieve superior performance in predicting visual scanpaths. The code will be publicly available on URL: https://github.com/sunwj/scanpath.

New Model-Based Bioequivalence Statistical Approaches for Pharmacokinetic Studies with Sparse Sampling.

In traditional pharmacokinetic (PK) bioequivalence analysis, two one-sided tests (TOST) are conducted on the area under the concentration-time curve and the maximal concentration derived using a non-compartmental approach. When rich sampling is unfeasible, a model-based (MB) approach, using nonlinear mixed effect models (NLMEM) is possible. However, MB-TOST using asymptotic standard errors (SE) presents increased type I error when asymptotic conditions do not hold. In this work, we propose three alternative calculations of the SE based on (i) an adaptation to NLMEM of the correction proposed by Gallant, (ii) the a posteriori distribution of the treatment coefficient using the Hamiltonian Monte Carlo algorithm, and (iii) parametric random effects and residual errors bootstrap. We evaluate these approaches by simulations, for two-arms parallel and two-period, two-sequence cross-over design with rich (n = 10) and sparse (n = 3) sampling under the null and the alternative hypotheses, with MB-TOST. All new approaches correct for the inflation of MB-TOST type I error in PK studies with sparse designs. The approach based on the a posteriori distribution appears to be the best compromise between controlled type I errors and computing times. MB-TOST using non-asymptotic SE controls type I error rate better than when using asymptotic SE estimates for bioequivalence on PK studies with sparse sampling.

Learned Image Downscaling for Upscaling using Content Adaptive Resampler.

Deep convolutional neural network based image super-resolution (SR) models have shown superior performance in recovering the underlying high resolution (HR) images from low resolution (LR) images obtained from the predefined downscaling methods. In this paper, we propose a learned image downscaling method based on content adaptive resampler (CAR) with consideration on the upscaling process. The proposed resampler network generates content adaptive image resampling kernels that are applied to the original HR input to generate pixels on the downscaled image. Moreover, a differentiable upscaling (SR) module is employed to upscale the LR result into its underlying HR counterpart. By back-propagating the reconstruction error down to the original HR input across the entire framework to adjust model parameters, the proposed framework achieves a new state-of-the-art SR performance through upscaling guided image resamplers which adaptively preserve detailed information that is essential to the upscaling. Experimental results indicate that the quality of the generated LR image is comparable to that of the traditional interpolation based method and the significant SR performance gain is achieved by deep SR models trained jointly with the CAR model. The code is publicly available on: https://github.com/sunwj/CAR.

Assessing the ratio of means as a causal estimand in clinical endpoint bioequivalence studies in the presence of intercurrent events.

In clinical endpoint bioequivalence studies, the observed per-protocol (PP) population (compliers and completers in general) is usually used in the primary analysis for equivalence assessment. However, intercurrent events, ie, missingness and noncompliance, are not properly handled. The resulting estimand is not causal. Previously, we proposed the first causal framework to assess equivalence in the presence of missing data and noncompliance. We proposed a causal survivor average causal effect (SACE) estimand for the difference of means (DOM). In equivalence assessment, DOM is not as widely used as the ratio of means (ROM). However, no existing formula links the observed PP estimand to the SACE estimand for ROM as exists for DOM. Herein, we propose a similar causal framework for ROM using the principal stratification approach, one of the strategies recommended by the International Conference on Harmonisation (ICH) E9 R1 addendum. We quantify the bias of the observed ROM PP estimand for the SACE estimand, which provides a basis to identify three conditions under which the two estimands are equal. We propose a sensitivity analysis method to evaluate the robustness of the current PP estimator to estimate the SACE estimand. We extend Fieller's confidence interval for the SACE estimand using ROM, which can be applied to many settings. Simulation demonstrates that the PP estimator is biased in either directions and may inflate type 1 error and/or change power when the three identified conditions are violated. Our work can be applied to comparative clinical biosimilar studies.

Statistical considerations and impact of the FDA draft guidance for assessing adhesion with transdermal delivery systems and topical patches for ANDAs.

Until 2016, a ratio of means (ROM) non-inferiority (NI) test was recommended in FDA product-specific guidances (PSGs) to evaluate adhesion performance for prospective generic transdermal delivery systems (TDS). However, the ROM NI test had low power for well-adhering TDS, which were becoming increasingly prevalent. Mathematical proof and simulation revealed that the low power wasn't because the non-normality of adhesion data violated the normality assumption of parametric methods; it was because the ROM NI test was coupled with an adhesion scale where scores approached 0 as adhesion got better. In June 2016, FDA published a draft general guidance on TDS adhesion and recommended a new statistical approach, replacing the ROM NI test with a difference-of-means (DOM) NI test, using the same scale and primary endpoint (mean adhesion scores). An analysis of 40 TDS adhesion studies submitted in ANDAs after the publication of the 2016 draft guidance suggests that, consistent with simulation results, the new statistical approach markedly improves the low power, and thereby reduces the sample size required by the old approach for moderately to well-adhering TDS, while retaining comparable power for poorly adhering TDS. The new statistical approach thus enhances the potential approvability and patient access to well-adhering generic TDS.

Adaptive clinical endpoint bioequivalence studies with sample size re-estimation based on a nuisance parameter.

A clinical endpoint bioequivalence (BE) study is often used to establish bioequivalence (BE) between a locally acting generic drug (T) and an innovator drug (R), which is a double-blind, randomized three-arm (T, R and placebo: P) parallel clinical trial. BE is established if two superiority tests (T vs. P, R vs. P) and one equivalence test (T vs. R) all pass. An accurate estimate of the nuisance parameter (e.g. variance) is vital in determining an accurate sample size to attain sufficient power. However, due to potential study design variations between NDA and Abbreviated NDA (ANDA) studies and high variability of clinical endpoints, variance may be over- or under-estimated, resulting in unnecessary extra costs or underpowered studies. Traditionally, clinical endpoint BE studies use a fixed study design. In this work, we propose four sample size re-estimation approaches based on a nuisance parameter and recommend one approach after comparing various operating characteristics by simulation. The proposed adaptive design with sample size re-estimation provides a more accurate estimate of sample size without wasting resources or under-powering the study and controls the Type 1 error rate under a negligible level, both for the family-wise alpha and individual alpha for superiority and equivalence tests.

Estimation of causal effects in clinical endpoint bioequivalence studies in the presence of intercurrent events: noncompliance and missing data.

In clinical endpoint bioequivalence (BE) studies, the primary analysis for assessing equivalence between a generic and an innovator product is based on the observed per-protocol (PP) population (usually completers and compliers). However, missing data and noncompliance are post-randomization intercurrent events and may introduce selection bias. Therefore, PP analysis is generally not causal. The FDA Missing Data Working Group recommended using "causal estimands of primary interest." In this paper, we propose a principal stratification causal framework and co-primary causal estimands to test equivalence, which was also recommended by the recently published ICH E9 (R1) addendum to address intercurrent events. We identify three conditions under which the current PP estimator is unbiased for one of the proposed co-primary causal estimands - the "Survivor Average Causal Effect" (SACE) estimand. Simulation shows that when these three conditions are not met, the PP estimator is biased and may inflate Type 1 error and/or change power. We also propose a tipping point sensitivity analysis to evaluate the robustness of the current PP estimator in testing equivalence when the sensitivity parameters deviate from the three identified conditions, but stay within a clinically meaningful range. Our work is the first causal equivalence assessment in equivalence studies with intercurrent events.

Ratio of means vs. difference of means as measures of superiority, noninferiority, and average bioequivalence.

Ratio of means (ROM) and difference of means (DOM) are often used in a superiority, noninferiority (NI), or average bioequivalence (ABE) test to evaluate whether the test mean is superior, NI, or equivalent to the reference (placebo or active control) mean. The literature provides recommendations regarding how to choose between ROM and DOM, mainly for superiority testing. In this article, we evaluated these two measures from other perspectives and cautioned the potential impact of different scoring systems/transformation for the same outcome (which is not rarely seen in practice) on the power of a ROM or DOM test for superiority, NI, or ABE. 1) For superiority, with the same margin, power remains the same for a location, scale, or combined shift (no other transformations) to scoring systems for both measures; however, for NI and ABE, different shifts can change the power of the test significantly. 2) Direction of scores (larger or smaller value indicating desirable effects) does not change the power for a DOM superiority, NI, or ABE test, but it does change the power tremendously for a ROM, NI, or ABE test. Caution should be taken when defining scoring systems. Data transformation is not encouraged in general, and if needed, should be statistically justified.

Skin collagen fluorophore LW-1 versus skin fluorescence as markers for the long-term progression of subclinical macrovascular disease in type 1 diabetes.

BACKGROUND: Skin collagen Long Wavelength Fluorescence (LWF) is widely used as a surrogate marker for accumulation of advanced glycation end-products. Here we determined the relationship of LWF with glycemia, skin fluorescence, and the progression of complications during EDIC in 216 participants from the DCCT. METHODS: LW-1 and collagen-linked fluorescence (CLF) were measured by either High Performance Liquid Chromatography (HPLC) with fluorescence detection (LW-1) or total fluorescence of collagenase digests (CLF) in insoluble skin collagen extracted from skin biopsies obtained at the end of the DCCT (1993). Skin intrinsic fluorescence (SIF) was noninvasively measured on volar forearm skin at EDIC year 16 by the SCOUT DS instrument. RESULTS: LW-1 levels significantly increased with age and diabetes duration (P < 0.0001) and significantly decreased by intensive vs. conventional glycemic therapy in both the primary (P < 0.0001) and secondary (P < 0.037) DCCT cohorts. Levels were associated with 13-16 year progression risk of retinopathy (>3 sustained microaneurysms, P = 0.0004) and albumin excretion rate (P = 0.0038), the latter despite adjustment for HbA1c. Comparative analysis for all three fluorescent measures for future risk of subclinical macrovascular disease revealed the following significant (P < 0.05) associations after adjusting for age, diabetes duration and HbA1c: coronary artery calcium with SIF and CLF; intima-media thickness with SIF and LW-1; and left ventricular mass with LW-1 and CLF. CONCLUSIONS: LW-1 is a novel risk marker that is robustly and independently associated with the future progression of microvascular disease, intima-media thickness and left ventricular mass in type 1 diabetes. Trial registration NCT00360815 and NCT00360893 at clinicaltrials.gov.

Effects of Prior Intensive Insulin Therapy and Risk Factors on Patient-Reported Visual Function Outcomes in the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Cohort.

IMPORTANCE: Preservation of vision in patients with diabetes mellitus is critical. Interventions to improve glycemic control through early intensive treatment of diabetes reduce rates of severe retinopathy and preserve visual acuity. OBJECTIVE: To assess the effects of prior intensive insulin treatment and risk factors on patient-reported visual function in the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) cohort. DESIGN, SETTING, AND PARTICIPANTS: Cohort study of 1184 participants with type 1 diabetes from the DCCT/EDIC study (randomized clinical trial followed by an observational follow-up study) who completed the 25-item National Eye Institute Visual Function Questionnaire (NEI-VFQ-25) during EDIC years 17 through 20 (September 1, 2009, through April 30, 2014) in 28 institutions across the United States and Canada. MAIN OUTCOMES AND MEASURES: The primary outcome was the composite NEI-VFQ-25 score. Secondary outcomes were visual acuity (measured by the Early Treatment Diabetic Retinopathy Study protocol), retinopathy level (determined by masked grading of stereoscopic color fundus photographs), and NEI-VFQ-25 subscale scores. The composite NEI-VFQ-25 scale and its subscales were scored 0 to 100, corresponding to poor to excellent function, respectively. RESULTS: The overall average NEI-VFQ-25 score for 1184 DCCT/EDIC participants (mean [SD] age, 52.3 [6.9] years; 48% female) with a 30-year duration of diabetes was high (all participants: median, 91.7; interquartile range [IQR], 89.7-96.9; intensive treatment [n = 605]: median, 94.7; IQR, 91.0-97.2; conventional treatment [n = 579]: median, 94.0; IQR, 88.4-96.1; P = .006 for intensive vs conventional). After adjustment for sex, age, hemoglobin A1c level, and retinopathy level at DCCT baseline, the former intensive treatment group had a significant, albeit modest, improvement in overall NEI-VFQ-25 score compared with the former conventional diabetes treatment group (median difference, -1.0; 95% CI, -1.7 to -0.3; P = .006). This beneficial treatment effect was fully attributed to the prior glycemic control in DCCT (explained treatment effect: 100%). Those with visual acuity worse than 20/100 reported the largest decline in visual function (median difference, -21.0; 95% CI, -40.5 to -1.6; P = .03). CONCLUSIONS AND RELEVANCE: In the DCCT/EDIC cohort, patient-reported visual function remains high in both treatment groups, comparable to previous reports of overall health-related quality of life. Intensive diabetes therapy modestly improved NEI-VFQ-25 score 30 years after the start of the DCCT, the benefit underestimated owing to more nonparticipants from the conventional treatment group. Visual acuity had the greatest effect on patient-reported visual function from among all risk factors. TRIAL REGISTRATION: clinicaltrials.gov Identifiers: NCT00360815 and NCT00360893.