Representativeness Research Articles

Fair patient model: Mitigating bias in the patient representation learned from the electronic health records

ObjectiveTo pre-train fair and unbiased patient representations from Electronic Health Records (EHRs) using a novel weighted loss function that reduces bias and improves fairness in deep representation learning models. MethodsWe defined a new loss function, called weighted loss function, in the deep representation learning model to balance the importance of different groups of patients and features. We applied the proposed model, called Fair Patient Model (FPM), to a sample of 34,739 patients from the MIMIC-III dataset and learned patient representations for four clinical outcome prediction tasks. ResultsFPM outperformed the baseline models in terms of three fairness metrics: demographic parity, equality of opportunity difference, and equalized odds ratio. FPM also achieved comparable predictive performance with the baselines, with an average accuracy of 0.7912. Feature analysis revealed that FPM captured more information from clinical features than the baselines. ConclusionFPM is a novel method to pre-train fair and unbiased patient representations from the EHR data using a weighted loss function. The learned representations can be used for various downstream tasks in healthcare and can be extended to other domains where fairness is important.

Explicit Timing Differently Predicts Implicit Timing Performance in Younger and Older Adults

Abstract Temporal processing can be divided into explicit timing and implicit timing. Explicit timing tasks require participants to attend to the temporal aspects of the task, whereas in implicit timing tasks, temporal information affects performance without explicit instruction to process time. Compared to younger adults, older adults have been shown to exhibit greater variability in explicit timing, while in implicit timing, they have been shown to rely more on temporal predictions formed by the hazard function. However, the relationship between explicit and implicit timing and its age-related changes have yet to be explored. To address this issue, we collected data in which younger and older adults performed a time bisection task (i.e., explicit timing task) and a foreperiod task (i.e., implicit timing task) in a within-subjects design. Based on a Bayesian optimization framework, we hypothesized that individuals with higher variability in explicit timing would show a stronger foreperiod effect, which is an index of the degree of reliance on temporal predictions. Results showed a different relation between explicit and implicit timing in younger and older adults. In older adults, results were consistent with the hypothesis that increased variability in explicit timing was associated with a stronger foreperiod effect. In contrast, bias in the temporal representation but not variability was associated with the foreperiod effect in younger adults. Implications for the age-related difference in the relation between explicit and implicit timing are discussed.

Evaluating whether the proportional odds models to analyse ordinal outcomes in COVID-19 clinical trials is providing clinically interpretable treatment effects: A systematic review.

After an initial recommendation from the World Health Organisation, trials of patients hospitalised with COVID-19 often include an ordinal clinical status outcome, which comprises a series of ordered categorical variables, typically ranging from 'Alive and discharged from hospital' to 'Dead'. These ordinal outcomes are often analysed using a proportional odds model, which provides a common odds ratio as an overall measure of effect, which is generally interpreted as the odds ratio for being in a higher category. The common odds ratio relies on the assumption of proportional odds, which implies an identical odds ratio across all ordinal categories; however, there is generally no statistical or biological basis for which this assumption should hold; and when violated, the common odds ratio may be a biased representation of the odds ratios for particular categories within the ordinal outcome. In this study, we aimed to evaluate to what extent the common odds ratio in published COVID-19 trials differed to simple binary odds ratios for clinically important outcomes. We conducted a systematic review of randomised trials evaluating interventions for patients hospitalised with COVID-19, which used a proportional odds model to analyse an ordinal clinical status outcome, published between January 2020 and May 2021. We assessed agreement between the common odds ratio and the odds ratio from a standard logistic regression model for three clinically important binary outcomes: 'Alive', 'Alive without mechanical ventilation', and 'Alive and discharged from hospital'. Sixteen randomised clinical trials, comprising 38 individual comparisons, were included in this study; of these, only 6 trials (38%) formally assessed the proportional odds assumption. The common odds ratio differed by more than 25% compared to the binary odds ratios in 55% of comparisons for the outcome 'Alive', 37% for 'Alive without mechanical ventilation', and 24% for 'Alive and discharged from hospital'. In addition, the common odds ratio systematically underestimated the odds ratio for the outcome 'Alive' by -16.8% (95% confidence interval: -28.7% to -2.9%, p = 0.02), though differences for the other outcomes were smaller and not statistically significant (-8.4% for 'Alive without mechanical ventilation' and 3.6% for 'Alive and discharged from hospital'). The common odds ratio was statistically significant for 18% of comparisons, while the binary odds ratio was significant in 5%, 16%, and 3% of comparisons for the outcomes 'Alive', 'Alive without mechanical ventilation', and 'Alive and discharged from hospital', respectively. The common odds ratio from proportional odds models often differs substantially to odds ratios from clinically important binary outcomes, and similar to composite outcomes, a beneficial common OR from a proportional odds model does not necessarily indicate a beneficial effect on the most important categories within the ordinal outcome.

Abstract cognitive maps of social network structure aid adaptive inference.

Social navigation-such as anticipating where gossip may spread, or identifying which acquaintances can help land a job-relies on knowing how people are connected within their larger social communities. Problematically, for most social networks, the space of possible relationships is too vast to observe and memorize. Indeed, people's knowledge of these social relations is well known to be biased and error-prone. Here, we reveal that these biased representations reflect a fundamental computation that abstracts over individual relationships to enable principled inferences about unseen relationships. We propose a theory of network representation that explains how people learn inferential cognitive maps of social relations from direct observation, what kinds of knowledge structures emerge as a consequence, and why it can be beneficial to encode systematic biases into social cognitive maps. Leveraging simulations, laboratory experiments, and "field data" from a real-world network, we find that people abstract observations of direct relations (e.g., friends) into inferences of multistep relations (e.g., friends-of-friends). This multistep abstraction mechanism enables people to discover and represent complex social network structure, affording adaptive inferences across a variety of contexts, including friendship, trust, and advice-giving. Moreover, this multistep abstraction mechanism unifies a variety of otherwise puzzling empirical observations about social behavior. Our proposal generalizes the theory of cognitive maps to the fundamental computational problem of social inference, presenting a powerful framework for understanding the workings of a predictive mind operating within a complex social world.

Smiling women pitching down: auditing representational and presentational gender biases in image-generative AI

Abstract Generative Artificial Intelligence (AI) models like DALL·E 2 can interpret prompts and generate high-quality images that exhibit human creativity. Though public enthusiasm is booming, systematic auditing of potential gender biases in AI-generated images remains scarce. We addressed this gap by examining the prevalence of two occupational gender biases (representational and presentational biases) in 15,300 DALL·E 2 images spanning 153 occupations. We assessed potential bias amplification by benchmarking against the 2021 U.S. census data and Google Images. Our findings reveal that DALL·E 2 underrepresents women in male-dominated fields while overrepresenting them in female-dominated occupations. Additionally, DALL·E 2 images tend to depict more women than men with smiles and downward-pitching heads, particularly in female-dominated (versus male-dominated) occupations. Our algorithm auditing study demonstrates more pronounced representational and presentational biases in DALL·E 2 compared to Google Images and calls for feminist interventions to curtail the potential impacts of such biased AI-generated images on the media ecology.

Ideology in Advertisement: A Critical Discourse Analysis of Selected Nepali Advertisements

This paper explores the way ideology is embedded in the underlying structure of advertisements, applying Critical Discourse Analysis (CDA) as a theoretical tool focusing on the ideas of Michael Foucault and Norman Fairclough. Four advertisements telecasted in Nepal Television are selected for the analysis. They represent four different fields such as jewelry shop, bathing soap, food supplement and electric cable. Following qualitative research design, this study has used critical, analytical and interpretative method for analyzing the textual data. Critical analysis of these representative advertisements has led to the conclusion that mass media discourse is a fertile playground for ideological games. Keeping in mind the social structure of Nepalese society, they all in a united form impart patriarchal ideology to the audiences. This article proposes a solution to such biased representation. Language is being used as a means of imparting the patriarchal ideology and we should seek a solution from there. We should create critical linguistic awareness for the audience and empower them to resist all such misrepresentations. Resistance is essential for gender equality and only critical linguistic awareness can help create a just society.

Justice in Educational Content: A Guide to Racial and Cultural Representation in Academic and Clinical Teaching and Assessment.

Case-based instruction is broadly used in health professions education, including physical therapy education. Case-based instruction can support achievement of higher-order, applied, learning objectives and clinical reasoning. Instructors strive to represent the diversity of the clinical population in case studies and may have explicit intercultural competency objectives. The inclusion of cultural, racial, and ethnic characteristics in cases or assessments can potentially reinforce stereotypes or inaccurately emphasize these characteristics as direct predictors of health profile. Furthermore, as most physical therapy faculty creating cases are from a White majority stance, there is a risk that inclusion of cultural elements risks inappropriate and biased representation. Well-intentioned instructors risk substituting cultural, racial, and ethnic characteristics for social and structural determinants of health. Race is a social, not biologic construction and should not be confused. Informed instructors guided by evidence-based strategies can achieve rich case depictions that do not convey inaccurate risk or alienate learners. A curriculum design strategy is offered for case development that brings explicit attention to representation of race and culture. This tool serves as a self-reflective and improvement tool. Continued community and student engagement is necessary to achieve high-quality and instructive case studies.

A Man’s World? The Policy Representation of Women and Men in a Comparative Perspective

Are the preferences of women and men unequally represented in public policies? This simple yet fundamental question has remained largely unexplored in the fast-growing fields of women’s representation and inequality in the opinion-policy link. Our study analyzes gender biases in policy representation using an original dataset covering 43 countries and four decades, with citizens’ preferences regarding more than 4,000 country-year policies linked to information about actual policy change. Our analysis reveals clear and robust evidence that women’s policy preferences are underrepresented compared to those of men. While this skew is fairly modest in terms of congruence, women’s representation is driven mostly by the high correlation of preferences with men. When there is disagreement, policy is more likely to align with men’s preferences. Our analyses further suggest that women’s substantive underrepresentation is mitigated in contexts with high levels of female descriptive representation and labor market participation. In sum, our study shows that gender inequality extends to the important realm of policy representation, but there is also meaningful variation in unequal representation across contexts.

How Not to Be a Fallibilist

Abstract I develop one partial explanation of the origins of our fallibilist intuitions about knowledge in ordinary language fallibilism and argue that this explanation indicates that our epistemic methodology should be more impartial and theory-neutral. First, I explain why the so-called Moorean constraint (cf. Hawthorne 2005, 111) that encapsulates fallibilist intuitions is fallibilism’s cornerstone. Second, I describe a pattern of fallibilist reasoning in light of the influential dual processing and heuristics and biases approach to cognition (cf. Kahneman 2011; Thaler and Sunstein 2008; Evans 2017). I suggest that this pattern of reasoning involves the question-substitution heuristic, the availability and representativeness heuristics, the focusing bias as well as framing effects, priming and the anchoring and adjusting heuristic. Third, I argue that this fallibilist pattern of reasoning is methodologically dubious because it involves a vicious circularity and briefly outline an alternative, more impartial and theory-neutral abductive methodology for the theory of knowledge. Finally, I briefly explain how this analysis sheds light on the ordinary language fallibilism of Moore (1939), Austin (1961), Wittgenstein (1969) and Chisholm (1982).

BANKA MÜŞTERİLERİNİN SERGİLEDİKLERİ DAVRANIŞSAL EĞİLİMLERİN BULANIK AHP YÖNTEMİ İLE ANALİZİ

The aim of the study was to weigh the behavioral biases of commercial bank and participation bank customers in Turkey in order of significance and to reveal whether there is a difference in behavioral biases of individuals. The questionnaire form prepared in this context was delivered to the individuals and the data obtained were evaluated using the fuzzy Analytical Hierarchy Process (AHP) method. As a result of the study, it was found that the three most effective main criteria for commercial bank customers are overconfidence, representation, and herding biases. The most important biases for the participation bank customers were determined as conservatism, overconfidence, and representation biases. Overconfidence, conservatism, and representation biases were determined as important for customers who prefer both commercial and participation banks in banking transactions. The results show that there are differences between the biases exhibited by individuals who transacted in different types of banking.

KSFinder-a knowledge graph model for link prediction of novel phosphorylated substrates of kinases.

Aberrant protein kinase regulation leading to abnormal substrate phosphorylation is associated with several human diseases. Despite the promise of therapies targeting kinases, many human kinases remain understudied. Most existing computational tools predicting phosphorylation cover less than 50% of known human kinases. They utilize local feature selection based on protein sequences, motifs, domains, structures, and/or functions, and do not consider the heterogeneous relationships of the proteins. In this work, we present KSFinder, a tool that predicts kinase-substrate links by capturing the inherent association of proteins in a network comprising 85% of the known human kinases. We also postulate the potential role of two understudied kinases based on their substrate predictions from KSFinder. KSFinder learns the semantic relationships in a phosphoproteome knowledge graph using a knowledge graph embedding algorithm and represents the nodes in low-dimensional vectors. A multilayer perceptron (MLP) classifier is trained to discern kinase-substrate links using the embedded vectors. KSFinder uses a strategic negative generation approach that eliminates biases in entity representation and combines data from experimentally validated non-interacting protein pairs, proteins from different subcellular locations, and random sampling. We assess KSFinder's generalization capability on four different datasets and compare its performance with other state-of-the-art prediction models. We employ KSFinder to predict substrates of 68 "dark" kinases considered understudied by the Illuminating the Druggable Genome program and use our text-mining tool, RLIMS-P along with manual curation, to search for literature evidence for the predictions. In a case study, we performed functional enrichment analysis for two dark kinases - HIPK3 and CAMKK1 using their predicted substrates. KSFinder shows improved performance over other kinase-substrate prediction models and generalized prediction ability on different datasets. We identified literature evidence for 17 novel predictions involving an understudied kinase. All of these 17 predictions had a probability score ≥0.7 (nine at >0.9, six at 0.8-0.9, and two at 0.7-0.8). The evaluation of 93,593 negative predictions (probability ≤0.3) identified four false negatives. The top enriched biological processes of HIPK3 substrates relate to the regulation of extracellular matrix and epigenetic gene expression, while CAMKK1 substrates include lipid storage regulation and glucose homeostasis. KSFinder outperforms the current kinase-substrate prediction tools with higher kinase coverage. The strategically developed negatives provide a superior generalization ability for KSFinder. We predicted substrates of 432 kinases, 68 of which are understudied, and hypothesized the potential functions of two dark kinases using their predicted substrates.

Examining the representativeness heuristic and anchoring effects in China's carbon markets

This paper proposes a novel integrated approach incorporating grey relational analysis, corresponding correlation test, with vector autoregression and panel vector autoregression models to examine the representativeness heuristic and anchoring effects in carbon markets. The grey relational analysis and corresponding correlation test measure the representativeness heuristic in eight pilot carbon markets. The vector autoregression and panel vector autoregression models detect the anchoring effects in eight pilot carbon markets and the overall carbon market. The study first explores the representativeness heuristic and anchoring effects in China's carbon markets. Taking the 2014–2021 United Nations Climate Change Conferences (COP20–COP26) as an event, the proposal of China's carbon emissions peak and carbon-natural targets, as well as the establishment of China's national carbon market as three events, the empirical results demonstrate that Hubei, Guangdong, Shenzhen, Beijing, Shanghai, and Chongqing six pilot carbon markets were found to exhibit the representativeness heuristics for the COP21. Hubei, Shenzhen, Beijing, and Chongqing carbon markets were found to exhibit the representativeness heuristics for the proposal of China's carbon emissions peak and carbon-natural targets, and the establishment of China's national carbon market. Anchoring effects were found to exist in Guangdong, Tianjin, and Chongqing carbon markets, while no anchoring effect has been found to exist in the overall carbon market. Investor sentiment is one of the crucial factors affecting carbon prices.

Dissecting the Gendered Diplomacy Paradigm: Delving into the Participation and Far-reaching Influence of Women in the Realm of International Relations

Gender diplomacy is a critical examination of the impact of women on global affairs, specifically focusing on their roles in key diplomatic positions such as ambassadors, foreign ministers, and heads of international organizations. This article employs a thorough analysis of statistical data and trends to assess the extent of gender parity in decision-making positions within international relations. It explores the influence of women's participation in shaping international relations, emphasizing instances where their involvement has contributed to conflict resolution and the implementation of gender-driven strategies. The article also sheds light on successful examples of gender-equal diplomacy while addressing pervasive challenges such as minimal representation, glass ceilings, and cultural biases. Structural and personal barriers are explored, with a particular emphasis on the absence of gender-sensitive policies and the delicate balance required between work and family commitments. To propel gender equality within the diplomatic arena, the article proposes a series of strategic measures, including policy recalibration, awareness initiatives, and robust international cooperation. The overarching goal is to foster a diplomatic environment characterized by genuine and comprehensive gender parity

Multi‐Approach Analysis of Baroclinic Internal Tide Perturbation in the Ionian Sea Abyssal Layer (Mediterranean Sea)

AbstractDespite being widely recognized, the importance of deep layers thermohaline and mixing processes in the ocean circulation and variability is still poorly investigated, especially in the Mediterranean Sea. This limits understanding and parametrizing deep dynamics, which result in evident biases in the global circulation representation by observations and numerical ocean simulations. Having access to hydrological datasets, collected on a whole water column, we investigated the abyssal stratification and its variability of the Ionian Sea (Central Mediterranean). Applying multiple analyses, we found a tidal‐period oscillation and the resulting activation of mixing, pointing out that the combined effect of stratification, morphology, and tides has a key role in enhancing local diapycnal diffusivity in the deepest layers, being a mechanism that connects the whole water column with a compelling impact on the vertical transport of heat and tracers.

Multi-Task Learning for UAV Aerial Object Detection in Foggy Weather Condition

Adverse weather conditions such as haze and snowfall can degrade the quality of captured images and affect performance of drone detection. Therefore, it is challenging to locate and identify targets in adverse weather scenarios. In this paper, a novel model called Object Detection in a Foggy Condition with YOLO (ODFC-YOLO) is proposed, which performs image dehazing and object detection jointly by multi-task learning approach. Our model consists of a detection subnet and a dehazing subnet, which can be trained end-to-end to optimize both tasks. Specifically, we propose a Cross-Stage Partial Fusion Decoder (CSP-Decoder) in the dehazing subnet to recover clean features of encoder from complex weather conditions, thereby reducing the feature discrepancy between hazy and clean images, thus enhancing the feature consistency between different tasks. Additionally, to increase the feature modeling and representation capabilities of our network, we also propose an efficient Global Context Enhanced Extraction (GCEE) module to extract beneficial information from blurred images by constructing global feature context long-range dependencies. Furthermore, we propose a Correlation-Aware Aggregated Loss (CAALoss) to average noise patterns and tune gradient magnitudes across different tasks, accordingly implicitly enhancing data diversity and alleviating representation bias. Finally, we verify the advantages of our proposed model on both synthetic and real-world foggy datasets, and our ODFC-YOLO achieves the highest mAP on all datasets while achieving 36 FPS real-time detection speed.

Does Heuristic Bias Matter for Long And Short-Term Investment Decision-Making During The COVID-19 Pandemic?

Introduction/Main Objectives: This study examines the effect of heuristic behavior on investment decision-making in the long- and short-term during the COVID-19 pandemic in the Indonesian capital market. Background Problems: Traditional finance cannot fully explain how investors behave in the capital market. Investors will tend to use heuristics when making investment decisions because humans have cognitive limitations, as explained in the bounded rationality theory. Especially during the COVID-19 pandemic, investors have shown their irrationality due to the high uncertainty and panic caused by the COVID-19 pandemic. This phenomenon can only be explained by behavioral finance. Novelty: This study examines the effect of bias on the invest­ment decision-making of investors who make long-term and short-term investments. Previous studies only tested the impact of bias directly, without differentiating the length of time of the investment. Research Methods: This study used partial least squares structural equation modelling (PLS-SEM) with WarpPLS tool. Testing the moderating effect was undertaken using multi-group analysis (MGA). Finding/ Results: The results of this study indicate that anchoring and availability bias have a positive effect on investment decision-making, while representativeness bias has no significant impact. Investment time moderates the effect of representativeness bias on irrational investment decision-making, while anchoring and availability bias are not supported. Conclusion: Anchoring and availability heuristics will increase irrational investment decisions, while the effect of representativeness heuristics on irrational investment decisions will decrease when investors make long-term investments.

Understanding patterns of stakeholder participation in public commenting on bureaucratic policymaking: Evidence from the European Union

AbstractWhat explains the levels and diversity of stakeholder participation in public commenting on bureaucratic policymaking? We examine a novel dataset on a stakeholder engagement mechanism recently introduced by the European Commission containing information about 1258 events organized between 2016 and 2019. We highlight the importance of administrative acts' characteristics and acknowledge the role of policy area type. Acts corresponding to early policy stages, broader in scope, less technical, and more explicit about feedback loop rules, that is, roadmaps, inception impact assessments and delegated acts, generate significantly more comments, from more diverse stakeholders, relative to legislative proposals, and draft implementing acts. Regulatory and distributive policies generate significantly more comments than interior and foreign policies. Diversity is significantly higher in distributive policies but only relative to foreign policies. We contribute by showing the power of administrative acts in influencing stakeholder participation and diversity across decision stages and policy areas and shaping bias in interest representation.

Misunderstanding Flight Part 2: Epistemology and the Philosophy of Science

Flight has become a common everyday occurrence. We have engineered ever more efficient and reliable aircraft, facilitating safe transportation around the world. However, from the education literature on the topic of lift, online discussions, and YouTube, it becomes apparent that there are underlying pedagogical issues. The 2003 New York Times article by Chang and the 2020 Scientific American article by Regis conclude that no one really understands flight. These claims are made without regard for the underpinning science and engineering responsible for the modern aviation industry. Although, it does beg the question, why is there confusion about how wings work? Several factors have conspired together, resulting in this confusion. Fluid mechanics is a complex topic that stumped legends of physics and mathematics for centuries. It also contains paradoxes, exacerbating the complexity. However, the central thesis of this work is that knowledge about aerodynamics is not easy to construct due to two main factors. First, there are epistemological traps that directly lead to fallacious conclusions. Second, representativeness heuristics incorrectly apply behaviors of visible water to invisible air. While many assume they know how wings work, if they do not understand why there is confusion, rather than dismissing it, confusion will endure.

Content preserving image translation with texture co-occurrence and spatial self-similarity for texture debiasing and domain adaptation

Models trained on datasets with texture bias usually perform poorly on out-of-distribution samples since biased representations are embedded into the model. Recently, various image translation and debiasing methods have attempted to disentangle texture biased representations for downstream tasks, but accurately discarding biased features without altering other relevant information is still challenging. In this paper, we propose a novel framework that leverages image translation to generate additional training images using the content of a source image and the texture of a target image with a different bias property to explicitly mitigate texture bias when training a model on a target task. Our model ensures texture similarity between the target and generated images via a texture co-occurrence loss while preserving content details from source images with a spatial self-similarity loss. Both the generated and original training images are combined to train improved classification or segmentation models robust to inconsistent texture bias. Evaluation on five classification- and two segmentation-datasets with known texture biases demonstrates the utility of our method, and reports significant improvements over recent state-of-the-art methods in all cases.

Representation Bias in Data: A Survey on Identification and Resolution Techniques

Data-driven algorithms are only as good as the data they work with, while datasets, especially social data, often fail to represent minorities adequately. Representation Bias in data can happen due to various reasons, ranging from historical discrimination to selection and sampling biases in the data acquisition and preparation methods. Given that “bias in, bias out,” one cannot expect AI-based solutions to have equitable outcomes for societal applications, without addressing issues such as representation bias. While there has been extensive study of fairness in machine learning models, including several review papers, bias in the data has been less studied. This article reviews the literature on identifying and resolving representation bias as a feature of a dataset, independent of how consumed later. The scope of this survey is bounded to structured (tabular) and unstructured (e.g., image, text, graph) data. It presents taxonomies to categorize the studied techniques based on multiple design dimensions and provides a side-by-side comparison of their properties. There is still a long way to fully address representation bias issues in data. The authors hope that this survey motivates researchers to approach these challenges in the future by observing existing work within their respective domains.