Barbara Di Eugenio

Baris Karacan, Vaibhav Bhargava, Barbara Di Eugenio et al.

Effective "all-team" summarization in high-complexity settings like the Neonatal Intensive Care Unit (NICU) requires aggregating insights from diverse disciplines (physicians, nurses, therapists) spread across hundreds of clinical free-text notes. Simply pooling heterogeneous text often leads to incoherent outputs. Structured summarization therefore first requires accurate categorization of sentence-level provenance across multi-source notes. This pilot study introduces a clinical provenance categorization pipeline using supervised fine-tuning (SFT) of large language models (LLMs). We adapted two Llama-3 models (8B and 70B) to MedSecId, a corpus of 2,002 MIMIC-III (Adult ICU) notes annotated with clinical provenance headers, achieving in-domain Macro F1 scores above 92% for both models. To evaluate cross-domain generalization, we assessed model capacity (8B vs. 70B) and quantization on a gold-standard dataset of 227 sentence-level spans derived from three multi-disciplinary NICU summaries. Experimental results demonstrate a scale-dependent transfer effect: while SFT produced only marginal changes for the 8B model, it substantially improved the 70B model, increasing Macro F1 by 7%. Notably, the quantized fine-tuned 70B model outperformed its full-precision baseline while substantially reducing computational requirements. These findings suggest that sufficient model capacity is critical for preserving semantic flexibility during cross-domain clinical transfer and that efficient quantized adaptation can enable structured provenance modeling for downstream summarization.

Abhinav Kumar, Barbara Di Eugenio, Abari Bhattacharya et al.

Reference resolution, which aims to identify entities being referred to by a speaker, is more complex in real world settings: new referents may be created by processes the agents engage in and/or be salient only because they belong to the shared physical setting. Our focus is on resolving references to visualizations on a large screen display in multimodal dialogue; crucially, reference resolution is directly involved in the process of creating new visualizations. We describe our annotations for user references to visualizations appearing on a large screen via language and hand gesture and also new entity establishment, which results from executing the user request to create a new visualization. We also describe our reference resolution pipeline which relies on an information-state architecture to maintain dialogue context. We report results on detecting and resolving references, effectiveness of contextual information on the model, and under-specified requests for creating visualizations. We also experiment with conventional CRF and deep learning / transformer models (BiLSTM-CRF and BERT-CRF) for tagging references in user utterance text. Our results show that transfer learning significantly boost performance of the deep learning methods, although CRF still out-performs them, suggesting that conventional methods may generalize better for low resource data.

Yue Zhou, Henry Peng Zou, Barbara Di Eugenio et al.

We find that language models have difficulties generating fallacious and deceptive reasoning. When asked to generate deceptive outputs, language models tend to leak honest counterparts but believe them to be false. Exploiting this deficiency, we propose a jailbreak attack method that elicits an aligned language model for malicious output. Specifically, we query the model to generate a fallacious yet deceptively real procedure for the harmful behavior. Since a fallacious procedure is generally considered fake and thus harmless by LLMs, it helps bypass the safeguard mechanism. Yet the output is factually harmful since the LLM cannot fabricate fallacious solutions but proposes truthful ones. We evaluate our approach over five safety-aligned large language models, comparing four previous jailbreak methods, and show that our approach achieves competitive performance with more harmful outputs. We believe the findings could be extended beyond model safety, such as self-verification and hallucination.

Baris Karacan, Barbara Di Eugenio, Patrick Thornton

Clinical free-text notes contain vital patient information. They are structured into labelled sections; recognizing these sections has been shown to support clinical decision-making and downstream NLP tasks. In this paper, we advance clinical section segmentation through three key contributions. First, we curate a new de-identified, section-labeled obstetrics notes dataset, to supplement the medical domains covered in public corpora such as MIMIC-III, on which most existing segmentation approaches are trained. Second, we systematically evaluate transformer-based supervised models for section segmentation on a curated subset of MIMIC-III (in-domain), and on the new obstetrics dataset (out-of-domain). Third, we conduct the first head-to-head comparison of supervised models for medical section segmentation with zero-shot large language models. Our results show that while supervised models perform strongly in-domain, their performance drops substantially out-of-domain. In contrast, zero-shot models demonstrate robust out-of-domain adaptability once hallucinated section headers are corrected. These findings underscore the importance of developing domain-specific clinical resources and highlight zero-shot segmentation as a promising direction for applying healthcare NLP beyond well-studied corpora, as long as hallucinations are appropriately managed.

Baris Karacan, Barbara Di Eugenio, Patrick Thornton et al.

Clinical framing -- the linguistic manner in which clinical information is presented -- can influence patient understanding and decision-making, with important implications for healthcare outcomes. Obstetrics is a high-stakes domain in which physicians counsel patients on delivery mode choices such as vaginal birth after cesarean (VBAC) and repeat cesarean section (RCS), yet counseling language remains underexplored in large-scale clinical text analysis. In this work, we analyze physician counseling language in 2,024 obstetric history and physical narratives for a rigorously defined cohort of patients for whom both VBAC and RCS were clinically viable options. To control for confounding due to medical contraindications, we first construct a VBAC-eligible cohort using structured clinical data supplemented by a large language model (LLM)-based extraction pipeline constrained to grounded, verbatim evidence from free-text narratives. We then apply a zero-shot LLM framework to categorize counseling segments into predefined framing categories capturing how physicians linguistically present delivery options. Our analysis reveals a significant difference in counseling framing distributions between VBAC and RCS notes; risk-focused language accounts for a substantially larger share of counseling segments in RCS documentation than in VBAC, with category-level differences confirmed by statistical testing, highlighting the value of controlled LLM-based framing analysis in obstetric care.

Yue Zhou, Barbara Di Eugenio, Lu Cheng

This paper studies the performance of large language models (LLMs), particularly regarding demographic fairness, in solving real-world healthcare tasks. We evaluate state-of-the-art LLMs with three prevalent learning frameworks across six diverse healthcare tasks and find significant challenges in applying LLMs to real-world healthcare tasks and persistent fairness issues across demographic groups. We also find that explicitly providing demographic information yields mixed results, while LLM's ability to infer such details raises concerns about biased health predictions. Utilizing LLMs as autonomous agents with access to up-to-date guidelines does not guarantee performance improvement. We believe these findings reveal the critical limitations of LLMs in healthcare fairness and the urgent need for specialized research in this area.

Yue Zhou, Barbara Di Eugenio, Brian Ziebart et al.

Health coaching helps patients achieve personalized and lifestyle-related goals, effectively managing chronic conditions and alleviating mental health issues. It is particularly beneficial, however cost-prohibitive, for low-socioeconomic status populations due to its highly personalized and labor-intensive nature. In this paper, we propose a neuro-symbolic goal summarizer to support health coaches in keeping track of the goals and a text-units-text dialogue generation model that converses with patients and helps them create and accomplish specific goals for physical activities. Our models outperform previous state-of-the-art while eliminating the need for predefined schema and corresponding annotation. We also propose a new health coaching dataset extending previous work and a metric to measure the unconventionality of the patient's response based on data difficulty, facilitating potential coach alerts during deployment.

Yue Zhou, Barbara Di Eugenio, Brian Ziebart et al.

Health coaching helps patients identify and accomplish lifestyle-related goals, effectively improving the control of chronic diseases and mitigating mental health conditions. However, health coaching is cost-prohibitive due to its highly personalized and labor-intensive nature. In this paper, we propose to build a dialogue system that converses with the patients, helps them create and accomplish specific goals, and can address their emotions with empathy. However, building such a system is challenging since real-world health coaching datasets are limited and empathy is subtle. Thus, we propose a modularized health coaching dialogue system with simplified NLU and NLG frameworks combined with mechanism-conditioned empathetic response generation. Through automatic and human evaluation, we show that our system generates more empathetic, fluent, and coherent responses and outperforms the state-of-the-art in NLU tasks while requiring less annotation. We view our approach as a key step towards building automated and more accessible health coaching systems.

Anuja Tayal, Barbara Di Eugenio, Devika Salunke et al.

We propose a dialogue system that enables heart failure patients to inquire about salt content in foods and help them monitor and reduce salt intake. Addressing the lack of specific datasets for food-based salt content inquiries, we develop a template-based conversational dataset. The dataset is structured to ask clarification questions to identify food items and their salt content. Our findings indicate that while fine-tuning transformer-based models on the dataset yields limited performance, the integration of Neuro-Symbolic Rules significantly enhances the system's performance. Our experiments show that by integrating neuro-symbolic rules, our system achieves an improvement in joint goal accuracy of over 20% across different data sizes compared to naively fine-tuning transformer-based models.

Rochana Chaturvedi, Yue Zhou, Andrew Boyd et al.

Clinical notes in Electronic Health Records (EHRs) capture rich temporal information on events, clinician reasoning, and lifestyle factors often missing from structured data. Leveraging them for predictive modeling can be impactful for timely identification of chronic diseases. However, they present core natural language processing (NLP) challenges: long text, irregular event distribution, complex temporal dependencies, privacy constraints, and resource limitations. We present two complementary methods for temporally and contextually grounded risk prediction from longitudinal notes. First, we introduce HiTGNN, a hierarchical temporal graph neural network that integrates intra-note temporal event structures, inter-visit dynamics, and medical knowledge to model patient trajectories with fine-grained temporal granularity. Second, we propose ReVeAL, a lightweight, test-time framework that distills the reasoning of large language models into smaller verifier models. Applied to opportunistic screening for Type 2 Diabetes (T2D) using temporally realistic cohorts curated from private and public hospital corpora, HiTGNN achieves the highest predictive accuracy, especially for near-term risk, while preserving privacy and limiting reliance on large proprietary models. ReVeAL enhances sensitivity to true T2D cases and retains explanatory reasoning. Our ablations confirm the value of temporal structure and knowledge augmentation, and fairness analysis shows HiTGNN performs more equitably across subgroups.

Paul Landes, Sitara Rao, Aaron Jeremy Chaise et al.

The Achilles heel of Large Language Models (LLMs) is hallucination, which has drastic consequences for the clinical domain. This is particularly important with regards to automatically generating discharge summaries (a lengthy medical document that summarizes a hospital in-patient visit). Automatically generating these summaries would free physicians to care for patients and reduce documentation burden. The goal of this work is to discover new methods that combine language-based graphs and deep learning models to address provenance of content and trustworthiness in automatic summarization. Our method shows impressive reliability results on the publicly available Medical Information Mart for Intensive III (MIMIC-III) corpus and clinical notes written by physicians at Anonymous Hospital. rovide our method, generated discharge ary output examples, source code and trained models.

Yue Zhou, Barbara Di Eugenio

Despite LLMs' explicit alignment against demographic stereotypes, they have been shown to exhibit biases under various social contexts. In this work, we find that LLMs exhibit concerning biases in how they associate solution veracity with demographics. Through experiments across five human value-aligned LLMs on mathematics, coding, commonsense, and writing problems, we reveal two forms of such veracity biases: Attribution Bias, where models disproportionately attribute correct solutions to certain demographic groups, and Evaluation Bias, where models' assessment of identical solutions varies based on perceived demographic authorship. Our results show pervasive biases: LLMs consistently attribute fewer correct solutions and more incorrect ones to African-American groups in math and coding, while Asian authorships are least preferred in writing evaluation. In additional studies, we show LLMs automatically assign racially stereotypical colors to demographic groups in visualization code, suggesting these biases are deeply embedded in models' reasoning processes. Our findings indicate that demographic bias extends beyond surface-level stereotypes and social context provocations, raising concerns about LLMs' deployment in educational and evaluation settings.

Anuja Tayal, Devika Salunke, Barbara Di Eugenio et al.

We explore the potential of ChatGPT (3.5-turbo and 4) to generate conversations focused on self-care strategies for African-American heart failure patients -- a domain with limited specialized datasets. To simulate patient-health educator dialogues, we employed four prompting strategies: domain, African American Vernacular English (AAVE), Social Determinants of Health (SDOH), and SDOH-informed reasoning. Conversations were generated across key self-care domains of food, exercise, and fluid intake, with varying turn lengths (5, 10, 15) and incorporated patient-specific SDOH attributes such as age, gender, neighborhood, and socioeconomic status. Our findings show that effective prompt design is essential. While incorporating SDOH and reasoning improves dialogue quality, ChatGPT still lacks the empathy and engagement needed for meaningful healthcare communication.

Anuja Tayal, Devika Salunke, Barbara Di Eugenio et al.

Conversational assistants are becoming more and more popular, including in healthcare, partly because of the availability and capabilities of Large Language Models. There is a need for controlled, probing evaluations with real stakeholders which can highlight advantages and disadvantages of more traditional architectures and those based on generative AI. We present a within-group user study to compare two versions of a conversational assistant that allows heart failure patients to ask about salt content in food. One version of the system was developed in-house with a neurosymbolic architecture, and one is based on ChatGPT. The evaluation shows that the in-house system is more accurate, completes more tasks and is less verbose than the one based on ChatGPT; on the other hand, the one based on ChatGPT makes fewer speech errors and requires fewer clarifications to complete the task. Patients show no preference for one over the other.

Rochana Chaturvedi, Peyman Baghershahi, Sourav Medya et al.

Temporal information extraction from unstructured text is essential for contextualizing events and deriving actionable insights, particularly in the medical domain. We address the task of extracting clinical events and their temporal relations using the well-studied I2B2 2012 Temporal Relations Challenge corpus. This task is inherently challenging due to complex clinical language, long documents, and sparse annotations. We introduce GRAPHTREX, a novel method integrating span-based entity-relation extraction, clinical large pre-trained language models (LPLMs), and Heterogeneous Graph Transformers (HGT) to capture local and global dependencies. Our HGT component facilitates information propagation across the document through innovative global landmarks that bridge distant entities. Our method improves the state-of-the-art with 5.5% improvement in the tempeval $F_1$ score over the previous best and up to 8.9% improvement on long-range relations, which presents a formidable challenge. We further demonstrate generalizability by establishing a strong baseline on the E3C corpus. This work not only advances temporal information extraction but also lays the groundwork for improved diagnostic and prognostic models through enhanced temporal reasoning.

Paul Landes, Barbara Di Eugenio, Cornelia Caragea

Reproducing results in publications by distributing publicly available source code is becoming ever more popular. Given the difficulty of reproducing machine learning (ML) experiments, there have been significant efforts in reducing the variance of these results. As in any science, the ability to consistently reproduce results effectively strengthens the underlying hypothesis of the work, and thus, should be regarded as important as the novel aspect of the research itself. The contribution of this work is a framework that is able to reproduce consistent results and provides a means of easily creating, training, and evaluating natural language processing (NLP) deep learning (DL) models.

Mehrdad Alizadeh, Barbara Di Eugenio

Visual Question Answering (VQA) concerns providing answers to Natural Language questions about images. Several deep neural network approaches have been proposed to model the task in an end-to-end fashion. Whereas the task is grounded in visual processing, if the question focuses on events described by verbs, the language understanding component becomes crucial. Our hypothesis is that models should be aware of verb semantics, as expressed via semantic role labels, argument types, and/or frame elements. Unfortunately, no VQA dataset exists that includes verb semantic information. Our first contribution is a new VQA dataset (imSituVQA) that we built by taking advantage of the imSitu annotations. The imSitu dataset consists of images manually labeled with semantic frame elements, mostly taken from FrameNet. Second, we propose a multitask CNN-LSTM VQA model that learns to classify the answers as well as the semantic frame elements. Our experiments show that semantic frame element classification helps the VQA system avoid inconsistent responses and improves performance.

Paul Landes, Barbara Di Eugenio

To-do lists are a popular medium for personal information management. As to-do tasks are increasingly tracked in electronic form with mobile and desktop organizers, so does the potential for software support for the corresponding tasks by means of intelligent agents. While there has been work in the area of personal assistants for to-do tasks, no work has focused on classifying user intention and information extraction as we do. We show that our methods perform well across two corpora that span sub-domains, one of which we released.