Fabricio Murai

Oluseun Olulana, Kathleen Cachel, Fabricio Murai et al.

As learning-to-rank models are increasingly deployed for decision-making in areas with profound life implications, the FairML community has been developing fair learning-to-rank (LTR) models. These models rely on the availability of sensitive demographic features such as race or sex. However, in practice, regulatory obstacles and privacy concerns protect this data from collection and use. As a result, practitioners may either need to promote fairness despite the absence of these features or turn to demographic inference tools to attempt to infer them. Given that these tools are fallible, this paper aims to further understand how errors in demographic inference impact the fairness performance of popular fair LTR strategies. In which cases would it be better to keep such demographic attributes hidden from models versus infer them? We examine a spectrum of fair LTR strategies ranging from fair LTR with and without demographic features hidden versus inferred to fairness-unaware LTR followed by fair re-ranking. We conduct a controlled empirical investigation modeling different levels of inference errors by systematically perturbing the inferred sensitive attribute. We also perform three case studies with real-world datasets and popular open-source inference methods. Our findings reveal that as inference noise grows, LTR-based methods that incorporate fairness considerations into the learning process may increase bias. In contrast, fair re-ranking strategies are more robust to inference errors. All source code, data, and experimental artifacts of our experimental study are available here: https://github.com/sewen007/hoiltr.git

Henrique S. Assumpção, Fabrício Souza, Leandro Lacerda Campos et al.

Money laundering has become one of the most relevant criminal activities in modern societies, as it causes massive financial losses for governments, banks and other institutions. Detecting such activities is among the top priorities when it comes to financial analysis, but current approaches are often costly and labor intensive partly due to the sheer amount of data to be analyzed. Hence, there is a growing need for automatic anti-money laundering systems to assist experts. In this work, we propose DELATOR, a novel framework for detecting money laundering activities based on graph neural networks that learn from large-scale temporal graphs. DELATOR provides an effective and efficient method for learning from heavily imbalanced graph data, by adapting concepts from the GraphSMOTE framework and incorporating elements of multi-task learning to obtain rich node embeddings for node classification. DELATOR outperforms all considered baselines, including an off-the-shelf solution from Amazon AWS by 23% with respect to AUC-ROC. We also conducted real experiments that led to the discovery of 7 new suspicious cases among the 50 analyzed ones, which have been reported to the authorities.

Gavin Butts, Pegah Emdad, Jethro Lee et al.

There have been growing concerns around high-stake applications that rely on models trained with biased data, which consequently produce biased predictions, often harming the most vulnerable. In particular, biased medical data could cause health-related applications and recommender systems to create outputs that jeopardize patient care and widen disparities in health outcomes. A recent framework titled Fairness via AI posits that, instead of attempting to correct model biases, researchers must focus on their root causes by using AI to debias data. Inspired by this framework, we tackle bias detection in medical curricula using NLP models, including LLMs, and evaluate them on a gold standard dataset containing 4,105 excerpts annotated by medical experts for bias from a large corpus. We build on previous work by coauthors which augments the set of negative samples with non-annotated text containing social identifier terms. However, some of these terms, especially those related to race and ethnicity, can carry different meanings (e.g., "white matter of spinal cord"). To address this issue, we propose the use of Word Sense Disambiguation models to refine dataset quality by removing irrelevant sentences. We then evaluate fine-tuned variations of BERT models as well as GPT models with zero- and few-shot prompting. We found LLMs, considered SOTA on many NLP tasks, unsuitable for bias detection, while fine-tuned BERT models generally perform well across all evaluated metrics.

Kai Zhang, Rui Zhu, Shutian Ma et al.

Drug discovery is a critical task in biomedical natural language processing (NLP), yet explainable drug discovery remains underexplored. Meanwhile, large language models (LLMs) have shown remarkable abilities in natural language understanding and generation. Leveraging LLMs for explainable drug discovery has the potential to improve downstream tasks and real-world applications. In this study, we utilize open-source drug knowledge graphs, clinical trial data, and PubMed publications to construct a comprehensive dataset for the explainable drug discovery task, named \textbf{expRxRec}. Furthermore, we introduce \textbf{KEDRec-LM}, an instruction-tuned LLM which distills knowledge from rich medical knowledge corpus for drug recommendation and rationale generation. To encourage further research in this area, we will publicly release\footnote{A copy is attached with this submission} both the dataset and KEDRec-LM.

Henrique Assumpção, Diego Ferreira, Leandro Campos et al.

In this work, we introduce CodeEvolve, an open-source evolutionary coding agent that unites Large Language Models (LLMs) with genetic algorithms to solve complex computational problems. Our framework adapts powerful evolutionary concepts to the LLM domain, building upon recent methods for generalized scientific discovery. CodeEvolve employs an island-based genetic algorithm to maintain population diversity and increase throughput, introduces a novel inspiration-based crossover mechanism that leverages the LLMs context window to combine features from successful solutions, and implements meta-prompting strategies for dynamic exploration of the solution space. We conduct a rigorous evaluation of CodeEvolve on a subset of the mathematical benchmarks used to evaluate Google DeepMind's closed-source AlphaEvolve. Our findings show that our method surpasses AlphaEvolve's performance on several challenging problems. To foster collaboration and accelerate progress, we release our complete framework as an open-source repository.

Yiqing Zhang, Xiaozhong Liu, Fabricio Murai

Many existing models for clinical trial outcome prediction are optimized using task-specific loss functions on trial phase-specific data. While this scheme may boost prediction for common diseases and drugs, it can hinder learning of generalizable representations, leading to more false positives/negatives. To address this limitation, we introduce CLaDMoP, a new pre-training approach for clinical trial outcome prediction, alongside the Successful Clinical Trials dataset(SCT), specifically designed for this task. CLaDMoP leverages a Large Language Model-to encode trials' eligibility criteria-linked to a lightweight Drug-Molecule branch through a novel multi-level fusion technique. To efficiently fuse long embeddings across levels, we incorporate a grouping block, drastically reducing computational overhead. CLaDMoP avoids reliance on task-specific objectives by pre-training on a "pair matching" proxy task. Compared to established zero-shot and few-shot baselines, our method significantly improves both PR-AUC and ROC-AUC, especially for phase I and phase II trials. We further evaluate and perform ablation on CLaDMoP after Parameter-Efficient Fine-Tuning, comparing it to state-of-the-art supervised baselines, including MEXA-CTP, on the Trial Outcome Prediction(TOP) benchmark. CLaDMoP achieves up to 10.5% improvement in PR-AUC and 3.6% in ROC-AUC, while attaining comparable F1 score to MEXA-CTP, highlighting its potential for clinical trial outcome prediction. Code and SCT dataset can be downloaded from https://github.com/murai-lab/CLaDMoP.

Yiqing Zhang, Xiaozhong Liu, Fabricio Murai

Trustworthy biomedical question answering (QA) systems must not only provide accurate answers but also justify them with current, verifiable evidence. Retrieval-augmented approaches partially address this gap but lack mechanisms to iteratively refine poor queries, whereas self-reflection methods kick in only after full retrieval is completed. In this context, we introduce PubMed Reasoner, a biomedical QA agent composed of three stages: self-critic query refinement evaluates MeSH terms for coverage, alignment, and redundancy to enhance PubMed queries based on partial (metadata) retrieval; reflective retrieval processes articles in batches until sufficient evidence is gathered; and evidence-grounded response generation produces answers with explicit citations. PubMed Reasoner with a GPT-4o backbone achieves 78.32% accuracy on PubMedQA, slightly surpassing human experts, and showing consistent gains on MMLU Clinical Knowledge. Moreover, LLM-as-judge evaluations prefer our responses across: reasoning soundness, evidence grounding, clinical relevance, and trustworthiness. By orchestrating retrieval-first reasoning over authoritative sources, our approach provides practical assistance to clinicians and biomedical researchers while controlling compute and token costs.

Geovana S. de Oliveira, Ana P. C. Silva, Fabricio Murai et al.

Vaccination remains a cornerstone of global public health, yet the COVID-19 pandemic exposed how online misinformation, political polarization, and declining institutional trust can undermine immunization efforts. Most of the prior computational studies that analyzed vaccine discourse on social platforms focus on English-language data, specific vaccines, or short time windows, impairing our understanding of long-term dynamics in high-impact, non-English contexts like Brazil, home to one of the world's most comprehensive immunization systems. We here present the largest longitudinal study of Brazil's vaccine discourse on YouTube, leveraging a semi-supervised stance detection framework that combines self-labeling and self-training to classify nearly 1.4 million comments. By integrating stance with temporal patterns, engagement metrics, and channel taxonomy (legacy media, science communicators, digital-native outlets), we map how pro- and anti-vaccine narratives evolve and circulate within a hybrid media ecosystem. Our results show that semi-supervised learning substantially improves stance classification robustness, enabling fine-grained tracking of public attitudes across Brazil's full immunization schedule. Polarization spikes during epidemiological crises, especially COVID-19, but becomes fragmented across vaccines and interaction patterns in the post-pandemic period. Notably, science communication and digital-native channels emerge as the primary loci of both supportive and oppositional engagement, revealing structural vulnerabilities in contemporary health communication. Thus, our work advances computational methods for large-scale stance modeling while offering actionable evidence for public health agencies, platform governance, and online information ecosystems.

Chiman Salavati, Shannon Song, Willmar Sosa Diaz et al.

Biased information (recently termed bisinformation) continues to be taught in medical curricula, often long after having been debunked. In this paper, we introduce BRICC, a firstin-class initiative that seeks to mitigate medical bisinformation using machine learning to systematically identify and flag text with potential biases, for subsequent review in an expert-in-the-loop fashion, thus greatly accelerating an otherwise labor-intensive process. A gold-standard BRICC dataset was developed throughout several years, and contains over 12K pages of instructional materials. Medical experts meticulously annotated these documents for bias according to comprehensive coding guidelines, emphasizing gender, sex, age, geography, ethnicity, and race. Using this labeled dataset, we trained, validated, and tested medical bias classifiers. We test three classifier approaches: a binary type-specific classifier, a general bias classifier; an ensemble combining bias type-specific classifiers independently-trained; and a multitask learning (MTL) model tasked with predicting both general and type-specific biases. While MTL led to some improvement on race bias detection in terms of F1-score, it did not outperform binary classifiers trained specifically on each task. On general bias detection, the binary classifier achieves up to 0.923 of AUC, a 27.8% improvement over the baseline. This work lays the foundations for debiasing medical curricula by exploring a novel dataset and evaluating different training model strategies. Hence, it offers new pathways for more nuanced and effective mitigation of bisinformation.

Yiqing Zhang, Xiaozhong Liu, Fabricio Murai

Clinical trials are the gold standard for assessing the effectiveness and safety of drugs for treating diseases. Given the vast design space of drug molecules, elevated financial cost, and multi-year timeline of these trials, research on clinical trial outcome prediction has gained immense traction. Accurate predictions must leverage data of diverse modes such as drug molecules, target diseases, and eligibility criteria to infer successes and failures. Previous Deep Learning approaches for this task, such as HINT, often require wet lab data from synthesized molecules and/or rely on prior knowledge to encode interactions as part of the model architecture. To address these limitations, we propose a light-weight attention-based model, MEXA-CTP, to integrate readily-available multi-modal data and generate effective representations via specialized modules dubbed "mode experts", while avoiding human biases in model design. We optimize MEXA-CTP with the Cauchy loss to capture relevant interactions across modes. Our experiments on the Trial Outcome Prediction (TOP) benchmark demonstrate that MEXA-CTP improves upon existing approaches by, respectively, up to 11.3% in F1 score, 12.2% in PR-AUC, and 2.5% in ROC-AUC, compared to HINT. Ablation studies are provided to quantify the effectiveness of each component in our proposed method.

Libby Tiderman, Juan Sanchez Mercedes, Fiona Romanoschi et al.

Social media may disseminate medical claims that highlight misleading correlations between social identifiers and diseases due to not accounting for structural determinants of health. Our research aims to identify biased medical claims on Twitter and measure their spread. We propose a machine learning framework that uses two models in tandem: RoBERTa to detect medical claims and DistilBERT to classify bias. After identifying original biased medical claims, we conducted a retweet cascade analysis, computing their individual reach and rate of spread. Tweets containing biased claims were found to circulate faster and further than unbiased claims.

Chiman Salavati, Shannon Song, Scott A. Hale et al.

The use of inappropriate language -- such as outdated, exclusionary, or non-patient-centered terms -- medical instructional materials can significantly influence clinical training, patient interactions, and health outcomes. Despite their reputability, many materials developed over past decades contain examples now considered inappropriate by current medical standards. Given the volume of curricular content, manually identifying instances of inappropriate use of language (IUL) and its subcategories for systematic review is prohibitively costly and impractical. To address this challenge, we conduct a first-in-class evaluation of small language models (SLMs) fine-tuned on labeled data and pre-trained LLMs with in-context learning on a dataset containing approximately 500 documents and over 12,000 pages. For SLMs, we consider: (1) a general IUL classifier, (2) subcategory-specific binary classifiers, (3) a multilabel classifier, and (4) a two-stage hierarchical pipeline for general IUL detection followed by multilabel classification. For LLMs, we consider variations of prompts that include subcategory definitions and/or shots. We found that both LLama-3 8B and 70B, even with carefully curated shots, are largely outperformed by SLMs. While the multilabel classifier performs best on annotated data, supplementing training with unflagged excerpts as negative examples boosts the specific classifiers' AUC by up to 25%, making them most effective models for mitigating harmful language in medical curricula.

Daniel de Mello, Renato Assunção, Fabricio Murai

Deep clustering (DC) leverages the representation power of deep architectures to learn embedding spaces that are optimal for cluster analysis. This approach filters out low-level information irrelevant for clustering and has proven remarkably successful for high dimensional data spaces. Some DC methods employ Generative Adversarial Networks (GANs), motivated by the powerful latent representations these models are able to learn implicitly. In this work, we propose HC-MGAN, a new technique based on GANs with multiple generators (MGANs), which have not been explored for clustering. Our method is inspired by the observation that each generator of a MGAN tends to generate data that correlates with a sub-region of the real data distribution. We use this clustered generation to train a classifier for inferring from which generator a given image came from, thus providing a semantically meaningful clustering for the real distribution. Additionally, we design our method so that it is performed in a top-down hierarchical clustering tree, thus proposing the first hierarchical DC method, to the best of our knowledge. We conduct several experiments to evaluate the proposed method against recent DC methods, obtaining competitive results. Last, we perform an exploratory analysis of the hierarchical clustering tree that highlights how accurately it organizes the data in a hierarchy of semantically coherent patterns.

Shiri Dori-Hacohen, Roberto Montenegro, Fabricio Murai et al.

Most Fairness in AI research focuses on exposing biases in AI systems. A broader lens on fairness reveals that AI can serve a greater aspiration: rooting out societal inequities from their source. Specifically, we focus on inequities in health information, and aim to reduce bias in that domain using AI. The AI algorithms under the hood of search engines and social media, many of which are based on recommender systems, have an outsized impact on the quality of medical and health information online. Therefore, embedding bias detection and reduction into these recommender systems serving up medical and health content online could have an outsized positive impact on patient outcomes and wellbeing. In this position paper, we offer the following contributions: (1) we propose a novel framework of Fairness via AI, inspired by insights from medical education, sociology and antiracism; (2) we define a new term, bisinformation, which is related to, but distinct from, misinformation, and encourage researchers to study it; (3) we propose using AI to study, detect and mitigate biased, harmful, and/or false health information that disproportionately hurts minority groups in society; and (4) we suggest several pillars and pose several open problems in order to seed inquiry in this new space. While part (3) of this work specifically focuses on the health domain, the fundamental computer science advances and contributions stemming from research efforts in bias reduction and Fairness via AI have broad implications in all areas of society.

Francisco Galuppo Azevedo, Fabricio Murai

Scientific knowledge cannot be seen as a set of isolated fields, but as a highly connected network. Understanding how research areas are connected is of paramount importance for adequately allocating funding and human resources (e.g., assembling teams to tackle multidisciplinary problems). The relationship between disciplines can be drawn from data on the trajectory of individual scientists, as researchers often make contributions in a small set of interrelated areas. Two recent works propose methods for creating research maps from scientists' publication records: by using a frequentist approach to create a transition probability matrix; and by learning embeddings (vector representations). Surprisingly, these models were evaluated on different datasets and have never been compared in the literature. In this work, we compare both models in a systematic way, using a large dataset of publication records from Brazilian researchers. We evaluate these models' ability to predict whether a given entity (scientist, institution or region) will enter a new field w.r.t. the area under the ROC curve. Moreover, we analyze how sensitive each method is to the number of publications and the number of fields associated to one entity. Last, we conduct a case study to showcase how these models can be used to characterize science dynamics in the context of Brazil.

Davi Pedrosa de Aguiar, Fabricio Murai

Inertial Measurement Unit (IMU) sensors are present in everyday devices such as smartphones and fitness watches. As a result, the array of health-related research and applications that tap onto this data has been growing, but little attention has been devoted to the prediction of an individual's heart rate (HR) from IMU data, when undergoing a physical activity. Would that be even possible? If so, this could be used to design personalized sets of aerobic exercises, for instance. In this work, we show that it is viable to obtain accurate HR predictions from IMU data using Recurrent Neural Networks, provided only access to HR and IMU data from a short-lived, previously executed activity. We propose a novel method for initializing an RNN's hidden state vectors, using a specialized network that attempts to extract an embedding of the physical conditioning (PCE) of a subject. We show that using a discriminator in the training phase to help the model learn whether two PCEs belong to the same individual further reduces the prediction error. We evaluate the proposed model when predicting the HR of 23 subjects performing a variety of physical activities from IMU data available in public datasets (PAMAP2, PPG-DaLiA). For comparison, we use as baselines the only model specifically proposed for this task and an adapted state-of-the-art model for Human Activity Recognition (HAR), a closely related task. Our method, PCE-LSTM, yields over 10% lower mean absolute error. We demonstrate empirically that this error reduction is in part due to the use of the PCE. Last, we use the two datasets (PPG-DaLiA, WESAD) to show that PCE-LSTM can also be successfully applied when photoplethysmography (PPG) sensors are available, outperforming the state-of-the-art deep learning baselines by more than 30%.

Bárbara Silveira, Henrique S. Silva, Fabricio Murai et al.

In recent years, Online Social Networks have become an important medium for people who suffer from mental disorders to share moments of hardship, and receive emotional and informational support. In this work, we analyze how discussions in Reddit communities related to mental disorders can help improve the health conditions of their users. Using the emotional tone of users' writing as a proxy for emotional state, we uncover relationships between user interactions and state changes. First, we observe that authors of negative posts often write rosier comments after engaging in discussions, indicating that users' emotional state can improve due to social support. Second, we build models based on SOTA text embedding techniques and RNNs to predict shifts in emotional tone. This differs from most of related work, which focuses primarily on detecting mental disorders from user activity. We demonstrate the feasibility of accurately predicting the users' reactions to the interactions experienced in these platforms, and present some examples which illustrate that the models are correctly capturing the effects of comments on the author's emotional tone. Our models hold promising implications for interventions to provide support for people struggling with mental illnesses.

Fabricio Murai, Diogo Rennó, Bruno Ribeiro et al.

Active search (AS) on graphs focuses on collecting certain labeled nodes (targets) given global knowledge of the network topology and its edge weights under a query budget. However, in most networks, nodes, topology and edge weights are all initially unknown. We introduce selective harvesting, a variant of AS where the next node to be queried must be chosen among the neighbors of the current queried node set; the available training data for deciding which node to query is restricted to the subgraph induced by the queried set (and their node attributes) and their neighbors (without any node or edge attributes). Therefore, selective harvesting is a sequential decision problem, where we must decide which node to query at each step. A classifier trained in this scenario suffers from a tunnel vision effect: without recourse to independent sampling, the urge to query promising nodes forces classifiers to gather increasingly biased training data, which we show significantly hurts the performance of AS methods and standard classifiers. We find that it is possible to collect a much larger set of targets by using multiple classifiers, not by combining their predictions as an ensemble, but switching between classifiers used at each step, as a way to ease the tunnel vision effect. We discover that switching classifiers collects more targets by (a) diversifying the training data and (b) broadening the choices of nodes that can be queried next. This highlights an exploration, exploitation, and diversification trade-off in our problem that goes beyond the exploration and exploitation duality found in classic sequential decision problems. From these observations we propose D3TS, a method based on multi-armed bandits for non-stationary stochastic processes that enforces classifier diversity, matching or exceeding the performance of competing methods on seven real network datasets in our evaluation.