Integration Of Prior Information Research Articles

Vision-Language Model for Generating Textual Descriptions From Clinical Images: Model Development and Validation Study.

The automatic generation of radiology reports, which seeks to create a free-text description from a clinical radiograph, is emerging as a pivotal intersection between clinical medicine and artificial intelligence. Leveraging natural language processing technologies can accelerate report creation, enhancing health care quality and standardization. However, most existing studies have not yet fully tapped into the combined potential of advanced language and vision models. The purpose of this study was to explore the integration of pretrained vision-language models into radiology report generation. This would enable the vision-language model to automatically convert clinical images into high-quality textual reports. In our research, we introduced a radiology report generation model named ClinicalBLIP, building upon the foundational InstructBLIP model and refining it using clinical image-to-text data sets. A multistage fine-tuning approach via low-rank adaptation was proposed to deepen the semantic comprehension of the visual encoder and the large language model for clinical imagery. Furthermore, prior knowledge was integrated through prompt learning to enhance the precision of the reports generated. Experiments were conducted on both the IU X-RAY and MIMIC-CXR data sets, with ClinicalBLIP compared to several leading methods. Experimental results revealed that ClinicalBLIP obtained superior scores of 0.570/0.365 and 0.534/0.313 on the IU X-RAY/MIMIC-CXR test sets for the Metric for Evaluation of Translation with Explicit Ordering (METEOR) and the Recall-Oriented Understudy for Gisting Evaluation (ROUGE) evaluations, respectively. This performance notably surpasses that of existing state-of-the-art methods. Further evaluations confirmed the effectiveness of the multistage fine-tuning and the integration of prior information, leading to substantial improvements. The proposed ClinicalBLIP model demonstrated robustness and effectiveness in enhancing clinical radiology report generation, suggesting significant promise for real-world clinical applications.

MachNet, a general Deep Learning architecture for Predictive Maintenance within the industry 4.0 paradigm

In the Industry 4.0 era, a myriad of sensors of diverse nature (temperature, pressure, etc.) is spreading throughout the entire value chain of industries, being potentially exploitable for multiple purposes, such as Predictive Maintenance (PdM): the just-in-time maintenance of industrial assets, which results in reduced operating costs, increased operator safety, etc. Nowadays, industrial processes require to be highly configurable, in order to proactively adapt their operation to diverse factors such as user needs, product updates or supply chain uncertainties. This limits current Industry 4.0-PdM solutions, typically consisting of ad-hoc developments intended for specific scenarios, i.e. they are designed to operate under certain conditions (configurations, employed sensors, etc.), being unable to manage changes in their setup.This paper presents a general Deep Learning (DL) architecture, MachNet, which deals with such heterogeneity and is able to address PdM problems of a diverse nature. The modularity of the proposed architecture enables it to deal with an arbitrary number of sensors of different types, also allowing the integration of prior information (age of assets, material type, etc.), which clearly affects performance and is often neglected. In practice, our architecture effortlessly adapts to the assets’ specifications and to different PdM problems. That is, MachNet becomes an architectural template that can be instantiated for a given scenario. We tested our proposal in two different PdM-related problems: Health State (HS) and Remaining-useful-Life (RuL) estimation, achieving in both cases comparable or superior performance to other state-of-the-art approaches, with the additional advantage of the generality that MachNet offers.

Preserved intention understanding during moral judgments in schizophrenia.

Although there is convincing evidence for socio-cognitive impairments in schizophrenia spectrum disorder (SSD), little evidence is found for deficient moral cognition. We investigated whether patients with SSD showed altered moral judgments in a story task where the protagonist either had a neutral or malicious intention towards another person. This paradigm examined whether SSD relates to altered moral cognition in general or specifically to impaired integration of prior information (such as beliefs) in moral judgments. 23 patients and 32 healthy controls read vignettes created in a 2 x 2 design. The protagonist in each story either had a neutral or negative intention towards another person which, as a result, either died (negative outcome) or did not die (neutral outcome). Participants rated the moral permissibility of the protagonist's action. Standard null hypothesis significance testing and equivalent Bayes analyses are reported. Schizophrenia patients did not differ significantly in permissibility ratings from healthy controls. This finding was supported by the Bayes analyses which favoured the null hypothesis. Task performance was not related to symptom severity or medication. The current findings do not support the notion that moral judgments are deficient in schizophrenia. Furthermore, the current study shows that patients do not have observable difficulties in integrating the protagonist's belief in the rating of the moral permissibility of the action-outcome.

Integration of prior information in Kaplan Meier estimator using Bayesian approach

Integration of prior information in Kaplan Meier estimator using Bayesian approach

Integration of prior information in Kaplan Meier estimator using Bayesian approach

As part of this contribution, we will illustrate the effectiveness of the Bayesian approach in estimating durations; we suggest a new definition of the Kaplan Meier Bayesian estimator based on a stochastic approximation under an informative prior. For this reason, based on the lognormal distribution, we have unconjugated a priori distributions. This method of processing makes it possible to assume that the use of the a priori data with the various suggested methods is sensitive to the choices of the parameters added.

Exploring how material cues drive sensorimotor prediction across different levels of autistic-like traits

Recent research proposes that sensorimotor difficulties, such as those experienced by many autistic people, may arise from atypicalities in prediction. Accordingly, we examined the relationship between non-clinical autistic-like traits and sensorimotor prediction in the material-weight illusion, where prior expectations derived from material cues typically bias one’s perception and action. Specifically, prediction-related tendencies in perception of weight, gaze patterns, and lifting actions were probed using a combination of self-report, eye-tracking, motion-capture, and force-based measures. No prediction-related associations between autistic-like traits and sensorimotor control emerged for any of these variables. Follow-up analyses, however, revealed that greater autistic-like traits were correlated with reduced adaptation of gaze with changes in environmental uncertainty. These findings challenge proposals of gross predictive atypicalities in autistic people, but suggest that the dynamic integration of prior information and environmental statistics may be related to autistic-like traits. Further research into this relationship is warranted in autistic populations, to assist the development of future movement-based coaching methods.

A stage-structured hierarchical Bayes model for the babassu palm tree population dynamics – Estimated from anthropogenic open area data sets

The babassu palm tree (Attalea speciosa Mart. ex Spreng.) is an endemic species of Amazon forests, and has social and economical impact/utility. Deforestation highlights this palm tree in anthropogenic open areas (pastures and cultivated fields). Simultaneously, knowledge concerning the sustainable functioning of the species within these manmade environments is sorely lacking: its life cycle is not well known, and its population dynamics remains unstudied. In this study, our objective was to generate a model of the population dynamics of the babassu palm tree, validated by in situ analysis, to understand how babassu, a forest species, adapts to pastureland and, under certain conditions, becomes invasive. We propose a random matrix model with aggregated variables based on the biological stages of the species as the input. The probabilities of the between-stage transition matrix were modelled using a Dirichlet-multinomial model with a hierarchy taking geographical organization, i.e. transect level, into account. The integration of prior information was formulated through a Bayesian approach. This Bayesian hierarchical matrix model enabled us to demonstrate a bottleneck in the population dynamics and a high year-dependent mortality rate at an early stage.

Uncertainty reduction through geologically conditioned petrophysical constraints in joint inversion

We have developed a joint geophysical inversion workflow that aims to improve subsurface imaging and decrease uncertainty by integrating petrophysical constraints and geologic data. In this framework, probabilistic geologic modeling is used as a source of information to condition the petrophysical constraints spatially and to derive starting models. The workflow then uses petrophysical measurements to constrain the values retrieved by geophysical joint inversion. The different sources of constraints are integrated into a least-squares framework to capture and integrate information related to geophysical, petrophysical, and geologic data. This allows us to quantify the posterior state of knowledge and to calculate posterior statistical indicators. To test this workflow, using geologic field data, we have generated a set of geologic models, which we used to derive a probabilistic geologic model. In this synthetic case study, we found that the integration of geologic information and petrophysical constraints in geophysical joint inversion could reduce uncertainty and improve imaging. In particular, the use of petrophysical constraints retrieves sharper boundaries and better reproduces the statistics of the observed petrophysical measurements. The integration of probabilistic geologic modeling permits more accurate retrieval of model geometry, and it better constrains the solution while still satisfying the statistics derived from geologic data. The analysis of statistical indicators at each step of the workflow indicates that (1) the inversion methodology is effective when applied to complex geology and (2) the integration of prior information and constraints from geology and petrophysics significantly improves the inversion results while decreasing uncertainty. Finally, the analysis of uncertainty to the integration of the conditioned petrophysical constraints also indicates that, for this example, the best results are obtained for joint inversion using petrophysical constraints spatially conditioned by geologic modeling.

An overview of a highly versatile forward and stable inverse algorithm for airborne, ground-based and borehole electromagnetic and electric data

We present an overview of a mature, robust and general algorithm providing a single framework for the inversion of most electromagnetic and electrical data types and instrument geometries. The implementation mainly uses a 1D earth formulation for electromagnetics and magnetic resonance sounding (MRS) responses, while the geoelectric responses are both 1D and 2D and the sheet’s response models a 3D conductive sheet in a conductive host with an overburden of varying thickness and resistivity. In all cases, the focus is placed on delivering full system forward modelling across all supported types of data. Our implementation is modular, meaning that the bulk of the algorithm is independent of data type, making it easy to add support for new types. Having implemented forward response routines and file I/O for a given data type provides access to a robust and general inversion engine. This engine includes support for mixed data types, arbitrary model parameter constraints, integration of prior information and calculation of both model parameter sensitivity analysis and depth of investigation. We present a review of our implementation and methodology and show four different examples illustrating the versatility of the algorithm. The first example is a laterally constrained joint inversion (LCI) of surface time domain induced polarisation (TDIP) data and borehole TDIP data. The second example shows a spatially constrained inversion (SCI) of airborne transient electromagnetic (AEM) data. The third example is an inversion and sensitivity analysis of MRS data, where the electrical structure is constrained with AEM data. The fourth example is an inversion of AEM data, where the model is described by a 3D sheet in a layered conductive host.

MVDA: a multi-view genomic data integration methodology.

BackgroundMultiple high-throughput molecular profiling by omics technologies can be collected for the same individuals. Combining these data, rather than exploiting them separately, can significantly increase the power of clinically relevant patients subclassifications.ResultsWe propose a multi-view approach in which the information from different data layers (views) is integrated at the levels of the results of each single view clustering iterations. It works by factorizing the membership matrices in a late integration manner. We evaluated the effectiveness and the performance of our method on six multi-view cancer datasets. In all the cases, we found patient sub-classes with statistical significance, identifying novel sub-groups previously not emphasized in literature. Our method performed better as compared to other multi-view clustering algorithms and, unlike other existing methods, it is able to quantify the contribution of single views on the final results.ConclusionOur observations suggest that integration of prior information with genomic features in the subtyping analysis is an effective strategy in identifying disease subgroups. The methodology is implemented in R and the source code is available online at http://neuronelab.unisa.it/a-multi-view-genomic-data-integration-methodology/.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-015-0680-3) contains supplementary material, which is available to authorized users.

Reversal learning and dopamine: a bayesian perspective.

Reversal learning has been studied as the process of learning to inhibit previously rewarded actions. Deficits in reversal learning have been seen after manipulations of dopamine and lesions of the orbitofrontal cortex. However, reversal learning is often studied in animals that have limited experience with reversals. As such, the animals are learning that reversals occur during data collection. We have examined a task regime in which monkeys have extensive experience with reversals and stable behavioral performance on a probabilistic two-arm bandit reversal learning task. We developed a Bayesian analysis approach to examine the effects of manipulations of dopamine on reversal performance in this regime. We find that the analysis can clarify the strategy of the animal. Specifically, at reversal, the monkeys switch quickly from choosing one stimulus to choosing the other, as opposed to gradually transitioning, which might be expected if they were using a naive reinforcement learning (RL) update of value. Furthermore, we found that administration of haloperidol affects the way the animals integrate prior knowledge into their choice behavior. Animals had a stronger prior on where reversals would occur on haloperidol than on levodopa (l-DOPA) or placebo. This strong prior was appropriate, because the animals had extensive experience with reversals occurring in the middle of the block. Overall, we find that Bayesian dissection of the behavior clarifies the strategy of the animals and reveals an effect of haloperidol on integration of prior information with evidence in favor of a choice reversal.

Prior Expectations Bias Sensory Representations in Visual Cortex

Perception is strongly influenced by expectations. Accordingly, perception has sometimes been cast as a process of inference, whereby sensory inputs are combined with prior knowledge. However, despite a wealth of behavioral literature supporting an account of perception as probabilistic inference, the neural mechanisms underlying this process remain largely unknown. One important question is whether top-down expectation biases stimulus representations in early sensory cortex, i.e., whether the integration of prior knowledge and bottom-up inputs is already observable at the earliest levels of sensory processing. Alternatively, early sensory processing may be unaffected by top-down expectations, and integration of prior knowledge and bottom-up input may take place in downstream association areas that are proposed to be involved in perceptual decision-making. Here, we implicitly manipulated human subjects' prior expectations about visual motion stimuli, and probed the effects on both perception and sensory representations in visual cortex. To this end, we measured neural activity noninvasively using functional magnetic resonance imaging, and applied a forward modeling approach to reconstruct the motion direction of the perceived stimuli from the signal in visual cortex. Our results show that top-down expectations bias representations in visual cortex, demonstrating that the integration of prior information and sensory input is reflected at the earliest stages of sensory processing.

Wind speed parameter estimation from one‐month sample via Bayesian approach

AbstractSite‐specific wind potential assessment shows difficulties mainly because it needs very long on‐site anemometric monitoring. This paper proposes to reduce the long‐term monitoring by exploiting other initial information about parameters to be estimated via MCMC (Markov chain Monte Carlo). The proposed Bayesian approach allows the integration of prior information (e.g. obtained from atlases, databases and/or fluid‐dynamic assessment) with sampling data, and furnishes effective and timely posterior information about Weibull parameters of the wind speed distribution. Real sampling data, collected from a southern Italian site, are analysed in order to illustrate the main features of the methodology. Moreover, the effectiveness of both the filtering strategy adopted to deal with the high correlation that usually characterizes the anemometric data, and the seasonal adjustment proposed to obtain a sample unbiased by seasonal effect is highlighted. The results of the application show that the proposed methodology fits the applicative needs very well. A bootstrap simulation remarks that the attained precision of the Bayesian estimates carried out from a one‐month sample is comparable to the maximum likelihood estimates obtained from an actual one‐year sample. Copyright © 2010 John Wiley & Sons, Ltd.

Prior Information Biases Stimulus Representations during Vibrotactile Decision Making

Neurophysiological data suggest that the integration of prior information and incoming sensory evidence represents the neural basis of the decision-making process. Here, we aimed to identify the brain structures involved in the integration of prior information about the average magnitude of a stimulus set and current sensory evidence. Specifically, we investigated whether prior average information already biases vibrotactile decision making during stimulus perception and maintenance before the comparison process. For this purpose, we used a vibrotactile delayed discrimination task and fMRI. At the behavioral level, participants showed the time-order effect. This psychophysical phenomenon has been shown to result from the influence of prior information on the perception of and the memory for currently presented stimuli. Similarly, the fMRI signal reflected the integration of prior information about the average vibration frequency and the currently presented vibration frequency. During stimulus encoding, the fMRI signal in primary and secondary somatosensory (S2) cortex, thalamus, and ventral premotor cortex mirrored an integration process. During stimulus maintenance, only a region in the intraparietal sulcus showed this modulation by prior average information. Importantly, the fMRI signal in S2 and intraparietal sulcus correlated with individual differences in the degree to which participants integrated prior average information. This strongly suggests that these two regions play a pivotal role in the integration process. Taken together, these results support the notion that the integration of current sensory and prior average information is a major feature of how the human brain perceives, remembers, and judges magnitude stimuli.

Internal structure and conditions of permafrost mounds at Umiujaq in Nunavik, Canada, inferred from field investigation and electrical resistivity tomography

A systematic approach was used for the interpretation of the electrical resistivity tomography carried out on two permafrost mounds at Umiujaq in Nunavik, Canada, to assess their internal structure and conditions. Prior information under the form of a geocryologic model of the permafrost mounds was integrated in the inversion of the pseudo-section of apparent electrical resistivity. The geocryologic model was developed from the synthesis of previous field investigations, including shallow and deep sampling, temperature and electrical resistivity logging, and cone penetration tests performed in the permafrost mounds. Values of electrical resistivity were ascribed to the different layers making of the geocryologic model to define a synthetic resistivity model of the permafrost mounds used as a reference model to constrain the inversion. The constrained resistivity model clearly show the presence of ice-rich cores in the permafrost mounds underscored by high resistivity values in excess of 30 000 Ωm, while the unfrozen zones surrounding the permafrost mounds are characterized by resistivity values lower than 1000 Ωm. The spatial distribution of unfrozen water and ice contents in the permafrost mounds were also assessed according to empirical relationships between the electrical resistivity and water contents. The ice content is highly variable and can be as high as 80% in the ice-rich cores, while the unfrozen water content varies between 2% and 5%. The integration of prior information in the inversion process leads to a more realistic constrained resistivity model showing sharp resistivity contrasts expected at the boundaries such as the permafrost table and base.