Partial Information Research Articles

Combining radiomics and molecular biomarkers: a novel economic tool to improve diagnostic ability in papillary thyroid cancer.

A preoperative diagnosis to distinguish malignant from benign thyroid nodules accurately and sensitively is urgently important. However, existing clinical methods cannot solve this problem satisfactorily. The aim of this study is to establish a simple, economic approach for preoperative diagnosis in eastern population. Our retrospective study included 86 patients with papillary thyroid cancer and 29 benign cases. The ITK-SNAP software was used to draw the outline of the area of interest (ROI), and Ultrosomics was used to extract radiomic features. Whole-transcriptome sequencing and bioinformatic analysis were used to identify candidate genes for thyroid nodule diagnosis. RT-qPCR was used to evaluate the expression levels of candidate genes. SVM diagnostic model was established based on the METLAB 2022 platform and LibSVM 3.2 language package. The radiomic model was first established. The accuracy is 73.0%, the sensitivity is 86.1%, the specificity is 17.6%, the PPV is 81.6%, and the NPV is 23.1%. Then, CLDN10, HMGA2, and LAMB3 were finally screened for model building. All three genes showed significant differential expressions between papillary thyroid cancer and normal tissue both in our cohort and TCGA cohort. The molecular model was established based on these genetic data and partial clinical information. The accuracy is 85.9%, the sensitivity is 86.1%, the specificity is 84.6%, the PPV is 96.9%, and the NPV is 52.4%. Considering that the above two models are not very effective, We integrated and optimized the two models to construct the final diagnostic model (C-thyroid model). In the training set, the accuracy is 96.7%, the sensitivity is 100%, the specificity is 93.8%, the PPV is 93.3%, and the NPV is 100%. In the validation set, the accuracy is 97.6%, the sensitivity remains 100%, the specificity is 84.6%, the PPV is 97.3%, and the NPV is 100%. A diagnostic panel is successfully established for eastern population through a simple, economic approach using only four genes and clinical data.

Two-Level method for blind image deblurring problems

Blind image deblurring (BID) is a procedure for reducing blur and noise in a deteriorated image. In this process, the estimation of the original image, as well as the blurring kernel of the degraded image, is done without or with only partial information about the imaging system and degradation. This is an inverse problem (ill-posed) that corresponds to the direct problem of deblurring. To overcome the ill-posedness of BID and attain useful solutions, the regularization models based on mean curvature (MC) are utilized. The discretization of MC-based models often leads to a large ill-conditioned nonlinear system of equations, which is computationally expensive. Moreover, the existence of MC functionals in the governing equations of the BID model complicates the calculation of the nonlinear system. To overcome these problems, in this paper, we propose the Two-Level blind image deblurring method (TLBID). First, on the coarse-grid, we solve a small nonlinear system (with a small number of pixels) for a mesh size of H, followed by solving a large linear system of equations on the finer grid (with a large number of pixels) of size h (h≤H). On the coarse mesh, we solve the BID problem utilizing the computationally expensive MC regularization functional. After this, we interpolate the results to the finer mesh. On the finer mesh, we solve the BID problem with less computationally expensive regularization functionals such as total variation (TV) or Tikhonov. This approach produces an approximate solution of the BID equations with high accuracy, which is cost-effective. The TLBID algorithm is implemented with MATLAB, and verification and validation are carried out using benchmark problems and medical digital images.

Uncovering nonequilibrium from unresolved events.

Closely related to the laws of thermodynamics, the detection and quantification of disequilibria are crucial in unraveling the complexities of nature, particularly those beneath observable layers. Theoretical developments in nonequilibrium thermodynamics employ coarse-graining methods to consider a diversity of partial information scenarios that mimic experimental limitations, allowing the inference of properties such as the entropy production rate. A ubiquitous but rather unexplored scenario involves observing events that can possibly arise from many transitions in the underlying Markov process-which we dub multifilar events-as in the cases of exchanges measured at particle reservoirs, hidden Markov models, mixed chemical and mechanical transformations in biological function, composite systems, and more. We relax one of the main assumptions in a previously developed framework, based on first-passage problems, to assess the non-Markovian statistics of multifilar events. By using the asymmetry of event distributions and their waiting times, we put forward model-free tools to detect nonequilibrium behavior and estimate entropy production, while discussing their suitability for different classes of systems and regimes where they provide no new information, evidence of nonequilibrium, a lower bound for entropy production, or even its exact value. The results are illustrated in reference models through analytics and numerics.

Power utility maximization with expert opinions at fixed arrival times in a market with hidden Gaussian drift

AbstractIn this paper we study optimal trading strategies in a financial market in which stock returns depend on a hidden Gaussian mean reverting drift process. Investors obtain information on that drift by observing stock returns. Moreover, expert opinions in the form of signals about the current state of the drift arriving at fixed and known dates are included in the analysis. Drift estimates are based on Kalman filter techniques. They are used to transform a power utility maximization problem under partial information into an optimization problem under full information where the state variable is the filter of the drift. The dynamic programming equation for this problem is studied and closed-form solutions for the value function and the optimal trading strategy of an investor are derived. They allow to quantify the monetary value of information delivered by the expert opinions. We illustrate our theoretical findings by results of extensive numerical experiments.

Bayesian Methods for Quality Tolerance Limit (QTL) Monitoring.

In alignment with the ICH guideline for Good Clinical Practice [ICH E6(R2)], quality tolerance limit (QTL) monitoring has become a standard component of risk-based monitoring of clinical trials by sponsor companies. Parameters that are candidates for QTL monitoring are critical to participant safety and quality of trial results. Breaching the QTL of a given parameter could indicate systematic issues with the trial that could impact participant safety or compromise the reliability of trial results. Methods for QTL monitoring should detect potential QTL breaches as early as possible while limiting the rate of false alarms. Early detection allows for the implementation of remedial actions that can prevent a QTL breach at the end of the trial. We demonstrate that statistically based methods that account for the expected value and variability of the data generating process outperform simple methods based on fixed thresholds with respect to important operating characteristics. We also propose a Bayesian method for QTL monitoring and an extension that allows for the incorporation of partial information, demonstrating its potential to outperform frequentist methods originating from the statistical process control literature.

Vibration-based damage detection of beams using Hybrid Genetic Algorithm with combined l0 and l1 regularization

Vibration-based damage detection is an effective way of identifying beam damages before they become severe and potentially catastrophic. This paper introduces a novel and robust method for detecting multiple cracks in beams using a Hybrid Genetic Algorithm (HGA). The objective function of the proposed method, based on dynamic properties, can be employed with partial modal information and is enhanced by including the initial residuals between numerical and experimental models along with combined l0 and l1 regularization. This enhancement mitigates the impact of limited modal data, experimental noise, and modeling inaccuracies. Additionally, a method of selecting an appropriate regularization parameter by analyzing residual and solution norm curves is presented. The performance of the proposed method is evaluated and verified using both numerical simulations and experimental studies on a steel cantilever beam under four damage scenarios. In the experimental study, natural frequencies and mode shapes were determined under ambient vibration conditions using Enhanced Frequency Domain Decomposition (EFDD) and the Stochastic Subspace Identification (SSI) method. Finite element models of the beam were developed in ABAQUS software for numerical analysis, and a comparative study of numerical and experimental data was conducted. The proposed approach successfully identifies, locates and quantifies single and multiple damages in the beam, demonstrating its reliability and effectiveness, particularly in scenarios with limited modal data and significant experimental noise.

Development and validation of novel interpretable survival prediction models based on drug exposures for severe heart failure during vulnerable period

BackgroundSevere heart failure (HF) has a higher mortality during vulnerable period while targeted predictive tools, especially based on drug exposures, to accurately assess its prognoses remain largely unexplored. Therefore, this study aimed to utilize drug information as the main predictor to develop and validate survival models for severe HF patients during this period.MethodsWe extracted severe HF patients from the MIMIC-IV database (as training and internal validation cohorts) as well as from the MIMIC-III database and local hospital (as external validation cohorts). Three algorithms, including Cox proportional hazards model (CoxPH), random survival forest (RSF), and deep learning survival prediction (DeepSurv), were applied to incorporate the parameters (partial hospitalization information and exposure durations of drugs) for constructing survival prediction models. The model performance was assessed mainly using area under the receiver operator characteristic curve (AUC), brier score (BS), and decision curve analysis (DCA). The model interpretability was determined by the permutation importance and Shapley additive explanations values.ResultsA total of 11,590 patients were included in this study. Among the 3 models, the CoxPH model ultimately included 10 variables, while RSF and DeepSurv models incorporated 24 variables, respectively. All of the 3 models achieved respectable performance metrics while the DeepSurv model exhibited the highest AUC values and relatively lower BS among these models. The DCA also verified that the DeepSurv model had the best clinical practicality.ConclusionsThe survival prediction tools established in this study can be applied to severe HF patients during vulnerable period by mainly inputting drug treatment duration, thus contributing to optimal clinical decisions prospectively.

Proposal for a General Outcome-Based Value Attribution Framework for Combination Therapies

ObjectivesValuing and pricing the components of combination therapies can be difficult because of competition law issues, difficulty implementing different prices for the same product in alternative uses, and attributing value to each component of the combination. We propose a value attribution solution that allows all combination components to be priced according to their relative value in the combination. MethodsWe developed a value attribution solution that is universal, symmetrical, and neutral to each combination constituent, regardless of whether it is the backbone or the add-on, and complete, meaning that it will always attribute the full value of the combination between the component parts. Moreover, it can be applied to any number of components in the combination (eg, triplets or quadruplets). We compared this solution with 2 other existing approaches. ResultsThe results of the proposed value attribution solution sit between those of the 2 other value attribution approaches as it combines elements of each. As the degree of additivity moves further away from one in either direction, then our general approach ratios also move, reflecting the impact of the incremental value. ConclusionsThe proposed value attribution solution for combination therapies differs from 2 existing approaches by being universally applicable and allowing for symmetry when neutral to the constituent components of the combination. To optimally contribute to policy debate and practice, various requirements for its implementation need to be well understood, including how to overcome (1) partial information, (2) whether its assumptions can be relaxed, and (3) implementation issues.

Modeling residual-vehicle effects on uncertainty estimation of the connected vehicle penetration rate

In the transition to full deployment of connected vehicles (CVs), the CV penetration rate plays a key role in bridging the gap between partial and complete traffic information. Several innovative methods have been proposed to estimate the CV penetration rate using only CV data. However, these methods, as point estimators, may lead to biased estimations or suboptimal solutions when applied directly in modeling or system optimization. To avoid these problems, the uncertainty and variability in the CV penetration rate must be considered. Recently, a probabilistic penetration rate (PPR) model was developed for estimating such uncertainties. The key model input is a constrained queue length distribution composed exclusively of queues formed by red signals in undersaturation conditions with no residual vehicles. However, in real-world scenarios, due to random arrivals, residual vehicles are commonly carried over from one cycle to another in temporary overflow cycles in undersaturation conditions, which seriously restricts the applicability of the PPR model. To address this limitation, this paper proposes a Markov-constrained queue length (MCQL) model that can model the complex effects of residual vehicles on the CV penetration rate uncertainty. A constrained queue with residual vehicles is decomposed into four vehicle groups: observable constrained residual vehicles, unobservable constrained residual vehicles, unconstrained residual vehicles, and new arrivals. Although the first vehicle group is observable in the former cycle, the focus of this work is to model the residual vehicles from the second and third vehicle groups in combination with the new arrivals. The MCQL model includes four sub-models, namely, the residual-vehicle model, convolutional constrained queue model, constrained residual queue model, and observable residual queue model, to isolate and derive the distribution of the constrained vehicle set formed by the three latter vehicle groups. This distribution is then substituted into the PPR model to estimate the uncertainty. Comprehensive VISSIM simulations and applications to real-world datasets demonstrate that the proposed MCQL model can accurately model the residual-vehicle effect and estimate the uncertainty. Thus, the applicability of the PPR model is truly extended to real-world settings, regardless of the presence of residual vehicles. A simple stochastic CV-based adaptive signal control example illustrates the potential of the proposed model in real-world applications.

Spiking attractor model of motor cortex explains modulation of neural and behavioral variability by prior target information

When preparing a movement, we often rely on partial or incomplete information, which can decrement task performance. In behaving monkeys we show that the degree of cued target information is reflected in both, neural variability in motor cortex and behavioral reaction times. We study the underlying mechanisms in a spiking motor-cortical attractor model. By introducing a biologically realistic network topology where excitatory neuron clusters are locally balanced with inhibitory neuron clusters we robustly achieve metastable network activity across a wide range of network parameters. In application to the monkey task, the model performs target-specific action selection and accurately reproduces the task-epoch dependent reduction of trial-to-trial variability in vivo where the degree of reduction directly reflects the amount of processed target information, while spiking irregularity remained constant throughout the task. In the context of incomplete cue information, the increased target selection time of the model can explain increased behavioral reaction times. We conclude that context-dependent neural and behavioral variability is a signum of attractor computation in the motor cortex.

Randomized greedy magic point selection schemes for nonlinear model reduction

An established way to tackle model nonlinearities in projection-based model reduction is via relying on partial information. This idea is shared by the methods of gappy proper orthogonal decomposition (POD), missing point estimation (MPE), masked projection, hyper reduction, and the (discrete) empirical interpolation method (DEIM). The selected indices of the partial information components are often referred to as “magic points.” The original contribution of the work at hand is a novel randomized greedy magic point selection. It is known that the greedy method is associated with minimizing the norm of an oblique projection operator, which, in turn, is associated with solving a sequence of rank-one SVD update problems. We propose simplification measures so that the resulting greedy point selection has the following main features: (1) The inherent rank-one SVD update problem is tackled in a way, such that its dimension does not grow with the number of selected magic points. (2) The approach is online efficient in the sense that the computational costs are independent from the dimension of the full-scale model. To the best of our knowledge, this is the first greedy magic point selection that features this property. We illustrate the findings by means of numerical examples. We find that the computational cost of the proposed method is orders of magnitude lower than that of its deterministic counterpart. Nevertheless, the prediction accuracy is just as good if not better. When compared to a state-of-the-art randomized method based on leverage scores, the randomized greedy method outperforms its competitor.

Data-and knowledge-driven belief rule learning for hybrid classification

In hybrid classification problems, apart from labeled data, some related expert knowledge may also be obtained. If the partial information from these two sources can be jointly used well, the performance may be effectively improved. To this end, we develop a new framework for hybrid classification by modeling the uncertain data and knowledge in the form of belief rules under the framework of belief function theory. First, a data-driven belief rule learning algorithm is proposed to learn a compact and interpretable belief rule model from labeled training data. Then, a knowledge-driven belief rule learning algorithm is proposed to learn a model which is complementary to that learned from data via active learning strategy. Finally, the genetic integration of belief rules is developed in order to integrate data-driven belief rules with knowledge-driven ones to reduce the rule redundancy and rule conflict by considering both classification accuracy and model interpretability. Experiments based on both simulated and real data sets demonstrate the superiority of the proposed model for integrating data and knowledge.

Adaptive Regularized Tri-Factor Non-Negative Matrix Factorization for Cell Type Deconvolution.

Accurate deconvolution of cell types from bulk gene expression is crucial for understanding cellular compositions and uncovering cell-type specific differential expression and physiological states of diseased tissues. Existing deconvolution methods have limitations, such as requiring complete cellular gene expression signatures or neglecting partial biological information. Moreover, these methods often overlook varying cell-type mRNA amounts, leading to biased proportion estimates. Additionally, they do not effectively utilize valuable reference information from external studies, such as means and ranges of population cell-type proportions. To address these challenges, we introduce an Adaptive Regularized Tri-factor non-negative matrix factorization approach for deconvolution (ARTdeConv). We rigorously establish the numerical convergence of our algorithm. Through benchmark simulations, we demonstrate the superior performance of ARTdeConv compared to state-of-the-art semi-reference-based and reference-free methods. In a real-world application, our method accurately estimates cell proportions, as evidenced by the nearly perfect Pearson's correlation between ARTdeConv estimates and flow cytometry measurements in a dataset from a trivalent influenza vaccine study. Moreover, our analysis of ARTdeConv estimates in COVID-19 patients reveals patterns consistent with important immunological phenomena observed in other studies. The proposed method, ARTdeConv, is implemented as an R package and can be accessed on GitHub for researchers and practitioners.

Recovery of damaged information via scrambling in indefinite causal order

Scrambling prevents the access to local information with local operators and therefore can be used to protect quantum information from damage caused by local perturbations. Even though partial quantum information can be recovered if the type of the damage is known, the initial target state cannot be completely recovered, because the obtained state is a mixture of the initial state and a maximally mixed state. Here, we demonstrate an improved scheme to recover damaged quantum information via scrambling in indefinite causal order. We show that scheme with indefinite causal order can record information of the damage and distill the initial state from the damaged state simultaneously. It allows us to retrieve initial information versus any damage. Moreover, by iterating the schemes, the initial quantum state can be completely recovered. In addition, we experimentally demonstrate our schemes on the cloud-based quantum computer, named as Quafu. Our work proposes a feasible scheme to protect whole quantum information from damage, which is also compatible with other techniques such as quantum error corrections and entanglement purification protocols. We expect that our scheme will be useful in the both quantum information recovery from the damage and systems bench-marking.

Positivity and convexity in incomplete cooperative games

Incomplete cooperative games generalize the classical model of cooperative games by omitting the values of some of the coalitions. This allows for incorporating uncertainty into the model and studying the underlying games and possible payoff distributions based only on the partial information. In this paper, we conduct a systematic investigation of incomplete games, focusing on two important classes: positive and convex games. Regarding positivity, we generalize previous results from a special class of minimal incomplete games to a general setting. We characterize the non-extendability to a positive game by the existence of a certificate and provide a description of the set of positive extensions using its extreme games. These results also enable the construction of explicit formulas for several classes of incomplete games with special structures. The second part deals with convexity. We begin with the case of non-negative, minimal incomplete games. We establish the connection between incomplete games and the problem of completing partial functions and, consequently, provide a characterization of extendability and a full description of the set of symmetric convex extensions. This set serves as an approximation of the set of convex extensions.

Partial information decomposition for continuous variables based on shared exclusions: Analytical formulation and estimation.

Describing statistical dependencies is foundational to empirical scientific research. For uncovering intricate and possibly nonlinear dependencies between a single target variable and several source variables within a system, a principled and versatile framework can be found in the theory of partial information decomposition (PID). Nevertheless, the majority of existing PID measures are restricted to categorical variables, while many systems of interest in science are continuous. In this paper, we present a novel analytic formulation for continuous redundancy-a generalization of mutual information-drawing inspiration from the concept of shared exclusions in probability space as in the discrete PID definition of I_{∩}^{sx}. Furthermore, we introduce a nearest-neighbor-based estimator for continuous PID and showcase its effectiveness by applying it to a simulated energy management system provided by the Honda Research Institute Europe GmbH. This work bridges the gap between the measure-theoretically postulated existence proofs for a continuous I_{∩}^{sx} and its practical application to real-world scientific problems.

Optimal Investment for Defined-Contribution Pension Plans with the Return of Premium Clause under Partial Information

The optimal investment problem for defined contribution (DC) pension plans with partial information is the subject of this paper. The purpose of the return of premium clauses is to safeguard the rights of DC pension plan participants who pass away during accumulation. We assume that the market price of risk consists of observable and unobservable factors that follow the Ornstein-Uhlenbeck processes, and the pension fund managers estimate the unobservable component from known information through Bayesian learning. Considering maximizing the expected utility of fund wealth at the terminal time, optimal investment strategies and the corresponding value function are determined using the dynamical programming principle approach and the filtering technique. Additionally, fund managers forsake learning, which results in the presentation of suboptimal strategies and utility losses for comparative analysis. Lastly, numerical analyses are included to demonstrate the impact of model parameters on the optimal strategy.

Instruct Pix-to-3D: Instructional 3D object generation from a single image

We study a challenging task, creating the instructional 3D object based on a single image and human instruction. Although existing methods utilize pre-trained text-to-image diffusion models to optimize Neural Radiance Fields (NeRF) and achieve remarkable results, they are inherently challenged in addressing this task due to two inherent obstacles. First, the written instruction is mismatched or uncorrelated with target images in the distribution of pre-trained text-to-image diffusion models, which leads to insufficient supervision. Another is editing 3D content starting from a single image, which is an ill-posed problem as the image only provides partial information. To tackle the difficulties, we propose an Instruct Pixel-to-3D framework, which exploits the supervisory signals from the given instructions and integrates the domain-specific knowledge drawn from 2D and 3D diffusion models. Specifically, to address the first challenge, we propose instruction guidance to facilitate meticulous image editing using fine-grained instructions and regulate the novel views of 3D objects using positional prompts. To tackle the second difficulty, we propose diffusion guidance for 3D generation, which optimizes the NeRF employing dual guidances: a 2D diffusion guidance derived from the score distillation sampling (SDS) loss and a 3D diffusion guidance embodied by the Zero-1-to-3 loss. Moreover, Instruct Pix-to-3D adopts a coarse-to-fine training strategy aimed at mitigating the complexities inherent in reconstructing scenes from sparse viewpoints. Qualitative and quantitative experiments on the DTU dataset and in-the-wild images demonstrate that our approach could synthesize high-quality and high-fidelity 3D content based on a single image and human instruction.

Evaluating theories of neural information integration during visual search.

The brain routes and integrates information from many sources during behavior. A number of models explain this phenomenon within the framework of mixed selectivity theory, yet it is difficult to compare their predictions to understand how neurons and circuits integrate information. In this work, we apply time-series partial information decomposition [PID] to compare models of integration on a dataset of superior colliculus [SC] recordings collected during a multi-target visual search task. On this task, SC must integrate target guidance, bottom-up salience, and previous fixation signals to drive attention. We find evidence that SC neurons integrate these factors in diverse ways, including decision-variable selectivity to expected value, functional specialization to previous fixation, and code-switching (to incorporate new visual input).

Propagation of Modelling Uncertainties for Seismic Risk Assessment: The Effect of Sampling Techniques on Low-Rise Non-Ductile RC Frames

ABSTRACT Quantifying the impact of modelling uncertainty on seismic performance assessment of existing buildings is non-trivial when considering the partial information available on material properties, construction details, and the uncertainty in the capacity models. This task is further complicated when uncertainty related to ground motion representation is considered. To address this issue, record-to-record variability, uncertainties in structural model parameters, and fragility model parameters due to limited sample size are propagated herein by employing a nonlinear dynamic analysis procedure based on recorded ground motions. A one-to-one sampling approach is adopted in which each recorded ground motion is paired up with a different structural model realization. Uncertainty propagation is explored by measuring the impact of different sampling techniques, such as Monte Carlo simulation with standard random sampling and Latin Hypercube sampling (with Simulated Annealing) in the presence of three alternative nonlinear dynamic analysis procedures: Incremental Dynamic Analysis (IDA), Modified Cloud Analysis (MCA), and Cloud to IDA (a highly efficient IDA-like procedure). This is all illustrated through application to an existing reinforced-concrete school building in southern Italy. It is shown that with a small subset of records, both MCA and Cloud to IDA can provide reliable structural fragility (and risk) estimates for three considered limit states, comparable to the results of more resource-intensive schemes.