SPSarah Masud Preum

Sarah Masud Preum

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sarah Masud Preum has a total of 146 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Characterizing Patient Portal Usage and Implications: A Data-driven Analysis on a Rural Population (Preprint)

    by Roshnik Rahat, Jialin Yang, Parker Seegmiller, Tim Burdick, Sarah Preum

    BACKGROUND Patient portals have become a primary channel for asynchronous patient–provider communication, yet rural-specific communication patterns remain underexplored. OBJECTIVE This study characterizes portal communication patterns in a rural-serving academic medical center over 5 years and links messaging behaviors to markers of care burden. METHODS We analyzed 370,498 patient messages and 256,295 provider responses from 10,206 patients at Dartmouth Hitchcock Medical Center (2020-2024). We also analyze the linked structured electronic health record (EHR) data for each patient. We used validated large language model (LLM)–based classifiers for thematic analysis and message authorship identification at scale. RESULTS Female patients and older adults generated disproportionately high message volumes. Anxiety (47.0%), hypertension (36.1%), and lipid disorders (33.7%) were the most prevalent conditions. Information seeking dominated portal communication (33.4%). The median response time to a patient messages was 10.6 hours, with patients who had dementia or cerebrovascular conditions waiting for the longest. Care partner-authored messages were substantially elevated among patients with Alzheimer disease and dementia (approximately 55%-60%) vs 5% for those without. CONCLUSIONS Rural portal communication reflects systematic disparities linked to age, sex, and clinical complexity. LLM-based analysis enables scalable characterization of thematic patterns and clarity failures that may inform AI-assisted triage.

  • Open Access

    How Much Would a Clinician Edit This Draft? Evaluating LLM Alignment for Patient Message Response Drafting

    by Parker Seegmiller, Joseph Gatto, Sarah E. Greer, Ganza Belise Isingizwe, Rohan Ray, Timothy E. Burdick, Sarah Masud Preum

    Large language models (LLMs) show promise in drafting responses to patient portal messages, yet their integration into clinical workflows raises various concerns, including whether they would actually save clinicians time and effort in their portal workload. We investigate LLM alignment with individual clinicians through a comprehensive evaluation of the patient message response drafting task. We develop a novel taxonomy of thematic elements in clinician responses and propose a novel evaluation framework for assessing clinician editing load of LLM-drafted responses at both content and theme levels. We release an expert-annotated dataset and conduct large-scale evaluations of local and commercial LLMs using various adaptation techniques including thematic prompting, retrieval-augmented generation, supervised fine-tuning, and direct preference optimization. Our results reveal substantial epistemic uncertainty in aligning LLM drafts with clinician responses. While LLMs demonstrate capability in drafting certain thematic elements, they struggle with clinician-aligned generation in other themes, particularly question asking to elicit further information from patients. Theme-driven adaptation strategies yield improvements across most themes. Our findings underscore the necessity of adapting LLMs to individual clinician preferences to enable reliable and responsible use in patient-clinician communication workflows.


  • Open Access

    Can LLMs Solve My Grandma's Riddle? Evaluating Multilingual Large Language Models on Reasoning Traditional Bangla Tricky Riddles

    by Nurul Labib Sayeedi, Faiyaz Abdullah Sayeedi, Khushnur Binte Jahangir, Swakkhar Shatabda, Sarah Masud Preum

    Large Language Models (LLMs) show impressive performance on many NLP benchmarks, yet their ability to reason in figurative, culturally grounded, and low-resource settings remains underexplored. We address this gap for Bangla by introducing BanglaRiddleEval, a benchmark of 1,244 traditional Bangla riddles instantiated across four tasks (4,976 riddle-task artifacts in total). Using an LLM-based pipeline, we generate Chain-of-Thought explanations, semantically coherent distractors, and fine-grained ambiguity annotations, and evaluate a diverse suite of open-source and closed-source models under different prompting strategies. Models achieve moderate semantic overlap on generative QA but low correctness, MCQ accuracy peaks at only about 56% versus an 83% human baseline, and ambiguity resolution ranges from roughly 26% to 68%, with high-quality explanations confined to the strongest models. These results show that current LLMs capture some cues needed for Bangla riddle reasoning but remain far from human-level performance, establishing BanglaRiddleEval as a challenging new benchmark for low-resource figurative reasoning. All data, code, and evaluation scripts are available on GitHub: https://github.com/Labib1610/BanglaRiddleEval.

  • Open Access

    Artificial Intelligence Approximates Human Affect Ratings of Cannabis Images

    by Jacob T. Borodovsky, Richard J. Macatee, Sarah M. Preum, Caroline L. Chung, Porter Malone, Denisse P. Gonzalez-Marquez

    Cannabis imagery is proliferating online and can elicit affective responses related to use. Scalable tools are needed to evaluate how this proliferation could influence population health. This pilot study tested whether multimodal generative artificial intelligence (MGAI) can reproduce subjective human affect ratings of cannabis images. Four MGAI agents (model: gpt-4o-2024-11-20) were created to parallel the four human participant subgroups from Macatee et al. 2021, defined by primary method of cannabis administration (bong, bowl, joint/blunt, vaporizer). Using Macatee et al.’s participant instructions and standardized image set, each agent rated images of its primary method of administration on valence, arousal, and urge constructs. For each image-construct pair, n=100 ratings were generated in separate conversational threads using zero-shot prompting. Image-level MGAI mean ratings were compared with human mean ratings using Two One-Sided Tests of equivalence and Spearman correlations. Although formal statistical equivalence was rare (4% valence, 11% arousal, 3% urge), MGAI ratings approximated human ratings closely (Mean difference of mean ratings = – 0.31, SD = 1.23) and correlations between MGAI and human mean ratings were moderate to high: r s (valence) = 0.55, r s (arousal) = 0.34, r s (urge) = 0.56. MGAI also reproduced the parabolic relation between rating means and standard deviations observed in human data. These preliminary results indicate that MGAI can approximate human cannabis cue-reactivity patterns closely enough to justify continued refinement. MGAI could potentially be developed into a Cannabis Regulatory Science tool to aid regulatory oversight of online cannabis marketing.

  • Open Access

    Developing an online cannabis store paradigm for Cannabis Regulatory Science

    by J.T. Borodovsky, S. Bairaboina, C.A. Struble, S.M. Preum, J.D. Sargent, J.A. Emond, A.J. Budney

    Background The legal cannabis industry is using online point-of-sale marketing, but few tools exist for evaluating how this marketing influences consumer behavior. To address this gap, we developed the Platform for Online Evaluation of Marijuana Marketing and Sales (POEMMS), a realistic, customizable online cannabis store. This study tested the feasibility, usability, and ecological validity of POEMMS. Methods U.S. adults aged 18+ were recruited via social media advertisements. Participants engaged in a simulated shopping task using POEMMS and then completed a post-shopping survey. Participants were instructed to imagine having no cannabis and to buy products as they would in real life. Survey measures included typical real-world cannabis spending and 0–10 Likert ratings of the store’s usability and realism. Spearman correlations tested associations between in-store and self-reported spending. Results N=678 completed the shopping task. Approximately 58% used cannabis daily. Median total spending was $100. Shopping task spending and self-reported spending for flower, open concentrates, cartridges, and edibles were strongly correlated (rs = 0.74, 0.74, 0.67, 0.62, respectively; all pConclusions POEMMS elicited behaviorally and perceptually valid responses in a large national sample of cannabis users. Findings demonstrate the platform’s utility for Cannabis Regulatory Science by allowing controlled evaluation of online marketing effects. Future applications of POEMMS could generate evidence to inform cannabis marketing policy and regulation.

  • Open Access

    Information Needs for Opioid Use Disorder Treatment Using Buprenorphine Product: Qualitative Analysis of Suboxone-Focused Reddit Data

    by Madhusudan Basak, Omar Sharif, Sarah E Lord, Jacob T Borodovsky, Lisa A Marsch, Sandra A Springer, Edward V Nunes, Charles D Brackett, Luke J Archibald, Sarah M Preum

    Background: Buprenorphine is a Food and Drug Administration–approved medication for opioid use disorder. However, individuals with opioid use disorder often report information needs regarding buprenorphine treatment on social media platforms such as Reddit. The field lacks a systematic approach to organizing these data and characterizing treatment information needs that may be unique and unavailable elsewhere. Objective: In this study, we curated and analyzed large-scale data from social media to characterize self-reported buprenorphine treatment information needs using a qualitative analysis. Methods: We collected 15,253 Reddit posts from the subreddit r/Suboxone. Following a standard protocol and guidance from clinical experts, we first identified 5 main themes from the data and then manually coded 6000 posts based on these themes. Finally, we determined the most frequently appearing topics within each theme by analyzing samples from each group. Results: Among the 6000 posts, 2417 (40.3%) contained a single theme, 2160 (36%) contained two themes, and 834 (13.9%) contained three themes. The most frequent topics found across these themes included reports of psychological and physical effects during recovery, complexities in accessing buprenorphine, and significant information gaps regarding medication administration, tapering, and substance use at different stages of recovery. Moreover, self-treatment strategies and peer-driven advice revealed potential rumors and misinformation. Conclusions: Our findings can guide the design of interventions to improve patient education and communication. They also help address knowledge gaps and misinformation related to treatment. Additionally, the results can support hypothesis generation for future clinical research on medication for opioid use disorder treatment.


  • Open Access

    Socially Constructed Treatment Plans: Analyzing Online Peer Interactions to Understand How Patients Navigate Complex Medical Conditions

    by Madhusudan Basak, Omar Sharif, Jessica Hulsey, Elizabeth C. Saunders, Daisy J. Goodman, Luke J. Archibald, Sarah M. Preum

    When faced with complex and uncertain medical conditions (e.g., cancer, mental health conditions, recovery from substance dependency), millions of patients seek online peer support. In this study, we leverage content analysis of online discourse and ethnographic studies with clinicians and patient representatives to characterize how treatment plans for complex conditions are "socially constructed." Specifically, we ground online conversation on medication-assisted recovery treatment to medication guidelines and subsequently surface when and why people deviate from the clinical guidelines. We characterize the implications and effectiveness of socially constructed treatment plans through in-depth interviews with clinical experts. Finally, given the enthusiasm around AI-powered solutions for patient communication, we investigate whether and how socially constructed treatment-related knowledge is reflected in a state-of-the-art large language model (LLM). Leveraging a novel mixed-method approach, this study highlights critical research directions for patient-centered communication in online health communities.

  • Open Access

    Follow-up Question Generation For Enhanced Patient-Provider Conversations

    by Joseph Gatto, Parker Seegmiller, Timothy Burdick, Inas S. Khayal, Sarah DeLozier, Sarah M. Preum

    Follow-up question generation is an essential feature of dialogue systems as it can reduce conversational ambiguity and enhance modeling complex interactions. Conversational contexts often pose core NLP challenges such as (i) extracting relevant information buried in fragmented data sources, and (ii) modeling parallel thought processes. These two challenges occur frequently in medical dialogue as a doctor asks questions based not only on patient utterances but also their prior EHR data and current diagnostic hypotheses. Asking medical questions in asynchronous conversations compounds these issues as doctors can only rely on static EHR information to motivate follow-up questions. To address these challenges, we introduce FollowupQ, a novel framework for enhancing asynchronous medical conversation. FollowupQ is a multi-agent framework that processes patient messages and EHR data to generate personalized follow-up questions, clarifying patient-reported medical conditions. FollowupQ reduces requisite provider follow-up communications by 34%. It also improves performance by 17% and 5% on real and synthetic data, respectively. We also release the first public dataset of asynchronous medical messages with linked EHR data alongside 2,300 follow-up questions written by clinical experts for the wider NLP research community.

  • Open Access

    Information Needs for Opioid Use Disorder Treatment Using Buprenorphine Product: Qualitative Analysis of Suboxone-Focused Reddit Data (Preprint)

    by Madhusudan Basak, Omar Sharif, Sarah E Lord, Jacob T Borodovsky, Lisa A Marsch, Sandra A Springer, Edward V Nunes, Charles D Brackett, Luke J Archibald, Sarah M Preum

    BACKGROUND Buprenorphine is a Food and Drug Administration–approved medication for opioid use disorder. However, individuals with opioid use disorder often report information needs regarding buprenorphine treatment on social media platforms such as Reddit. The field lacks a systematic approach to organizing these data and characterizing treatment information needs that may be unique and unavailable elsewhere. OBJECTIVE In this study, we curated and analyzed large-scale data from social media to characterize self-reported buprenorphine treatment information needs using a qualitative analysis. METHODS We collected 15,253 Reddit posts from the subreddit r/Suboxone. Following a standard protocol and guidance from clinical experts, we first identified 5 main themes from the data and then manually coded 6000 posts based on these themes. Finally, we determined the most frequently appearing topics within each theme by analyzing samples from each group. RESULTS Among the 6000 posts, 2417 (40.3%) contained a single theme, 2160 (36%) contained two themes, and 834 (13.9%) contained three themes. The most frequent topics found across these themes included reports of psychological and physical effects during recovery, complexities in accessing buprenorphine, and significant information gaps regarding medication administration, tapering, and substance use at different stages of recovery. Moreover, self-treatment strategies and peer-driven advice revealed potential rumors and misinformation. CONCLUSIONS Our findings can guide the design of interventions to improve patient education and communication. They also help address knowledge gaps and misinformation related to treatment. Additionally, the results can support hypothesis generation for future clinical research on medication for opioid use disorder treatment. CLINICALTRIAL

  • Open Access

    Subreddit to Symptomatology: A Lexicon-based Approach to Extract Symptoms of Complex Conditions from Online Discourse (Preprint)

    by Bushra Hossain, Sarah M Preum, Fazle Rabbi, Rifat Ara, Mohammed Eunus Ali

    Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Millions of people affected with complex medical conditions with diverse symptoms, often turn to online discourse to share their experiences. While some studies have explored natural language processing (NLP) methods and medical information extraction tools, these typically focus on generic symptoms in clinical notes and struggle to identify patient-reported, disease-specific, subtle symptoms from online health discourse. Objective: We aim to extract patient-reported, disease-specific symptoms shared on social media reflecting lived experiences of thousands of affected individuals and explore the characteristics, prevalence, and occurrence patterns of the symptoms. Methods: We propose a lexicon-based symptom extraction (LSE) method to identify a diverse list of disease-specific, patient-reported symptoms. We initially use a large language model (LLM) to accelerate the extraction of symptom-related keyphrases that form the lexicon. We evaluated the effectiveness of lexicon extraction against human annotation using a Jaccard Index score. Then, we leverage BERT-base, BioBERT, Phrase-BERT based embeddings to learn representations of these symptom-related keyphrases and cluster similar symptoms using k-means and HDBSCAN clustering. Among the different options explored in our experiments, the BioBERT-based k-means clustering was found to be the most effective. Finally, we apply symptom normalization to eliminate duplicate and redundant entries in the comprehensive symptom list. Results: In a real-world Polycystic Ovary Syndrome (PCOS) subreddit dataset, we find that LSE significantly outperforms state-of-the-art baselines, achieving at least 41% and 20% higher F1 scores (avg. 86.10) than automatic medical extraction tools and LLMs, respectively. Notably, the comprehensive list of 64 PCOS symptoms generated by LSE ensures extensive coverage of symptoms reported in 7 reputed e-health forums. Analyzing PCOS symptomatology reveals 28 potentially emerging symptoms and 8 self-reported comorbidities concurring with PCOS. Conclusions: The comprehensive patient-reported disease-specific symptom list potentially helps patients and health practitioners resolve the uncertainties surrounding the disease, eliminating the variability of PCOS symptoms prevailing in the community. Analyzing PCOS symptomatology across varied dimensions provides valuable insights for public health research.