arXiv subjects

Lucila Ohno-Machado

Publications and source records attributed to Lucila Ohno-Machado.

18 recordsLinked to original sources

arXiv2026

Toward Federated Large Language Models in Medicine: A Parameter-Efficient Framework for Privacy-Preserving, Multi-Institutional Adaptation

Large language models (LLMs) are increasingly adapted for medical applications, but most are trained using data from a single institution because privacy and governance constraints prevent multi-institutional data sharing. As a result, these models often generalize poorly across heterogeneous healthcare systems. We address this gap by introducing Fed-MedLoRA and Fed-MedLoRA+, a parameter-efficient federated framework for collaborative LLM adaptation across healthcare institutions. Fed-MedLoRA transmits only low-rank adapters rather than full model weights, reducing communication overhead. We also evaluate a privacy-preserving variant that applies Gaussian perturbation to transmitted adapter updates. Fed-MedLoRA+ further incorporates adaptive aggregation to better address cross-site heterogeneity in patient populations, annotation practices, and disease distributions. We evaluate the framework on clinical information extraction across five independent patient cohorts totaling 42,198 entities and 41,570 relations, and compare it with zero-shot and fine-tuned LLMs, domain-specific BERT models, and federated baselines. Across all settings, the proposed methods consistently improve extraction performance and generalize better to heterogeneous cohorts. In a real-world case study using clinical notes from the Yale New Haven Health System, the framework demonstrates strong performance under low-resource new-site deployment. These results suggest that federated, parameter-efficient LLM adaptation is feasible, scalable, and effective for multi-institutional clinical deployment.

cs.CL↗

arXiv2026

Memorization in Large Language Models in Medicine: Prevalence, Characteristics, and Implications

Large Language Models (LLMs) have demonstrated significant potential in medicine, with many studies adapting them through continued pre-training or fine-tuning on medical data to enhance domain-specific accuracy and safety. However, a key open question remains: to what extent do LLMs memorize medical training data. Memorization can be beneficial when it enables LLMs to retain valuable medical knowledge during domain adaptation. Yet, it also raises concerns. LLMs may inadvertently reproduce sensitive clinical content (e.g., patient-specific details), and excessive memorization may reduce model generalizability, increasing risks of misdiagnosis and making unwarranted recommendations. These risks are further amplified by the generative nature of LLMs, which can not only surface memorized content but also produce overconfident, misleading outputs that may hinder clinical adoption. In this work, we present a study on memorization of LLMs in medicine, assessing its prevalence (how frequently it occurs), characteristics (what is memorized), volume (how much content is memorized), and potential downstream impacts (how memorization may affect medical applications). We systematically analyze common adaptation scenarios: (1) continued pretraining on medical corpora, (2) fine-tuning on standard medical benchmarks, and (3) fine-tuning on real-world clinical data, including over 13,000 unique inpatient records from Yale New Haven Health System. The results demonstrate that memorization is prevalent across all adaptation scenarios and significantly higher than that reported in the general domain. Moreover, memorization has distinct characteristics during continued pre-training and fine-tuning, and it is persistent: up to 87% of content memorized during continued pre-training remains after fine-tuning on new medical tasks.

cs.CL↗

arXiv2026

A safety-oriented hypothetico-deductive framework for AI-assisted differential diagnosis

Diagnostic error is a major threat to patient safety, yet current large language model (LLM) systems often treat diagnosis as a one-shot prediction task, lacking safeguards against missed high-risk alternatives or rigorous verification of their reasoning. Here, we present AegisDx, a safety-oriented framework for hypothetico-deductive clinical reasoning. AegisDx coordinates specialized LLM components through role-specific contracts, structured intermediate outputs, evidence-retrieval interfaces, and verification gates to generate broad differential diagnoses, enforce explicit screening for dangerous "must-not-miss" conditions, verify reasoning against grounded medical evidence, and structure actionable next steps. We evaluated AegisDx across three layers. On literature-derived case reports from NEJM and JAMA, with GPT-oss-120B as the shared backbone, Top-3 diagnostic accuracy was 59.9% versus 52.1% for the standalone LLM on JAMA cases and 62.7% versus 51.4% on NEJM cases. On cases from Annals of Emergency Medicine, Top-3 accuracy was 85.7% versus 68.6%; against physician-consensus must-not-miss diagnosis sets, AegisDx captured at least one such condition among its top three diagnoses in 78.0% of cases versus 52.0%. In a blinded physician evaluation of 43 real-world emergency department notes from the Yale New Haven Health System compared against GPT-5, AegisDx improved the physician-rated composite safety score from 4.31 to 4.55 on a 5-point scale (adjusted p = 2.1x10^-4), with qualitative gains in must-not-miss identification and reasoning safety. Our findings suggest that engineering diagnostic AI as a safety-oriented reasoning framework, rather than optimizing raw predictive accuracy alone, can provide a safer, more transparent, and clinically meaningful layer of bedside decision support for acute care workflows.

cs.AI↗

arXiv2026

Foundation Models to Unlock Real-World Evidence from Nationwide Medical Claims

Evidence derived from large-scale real-world data (RWD) is increasingly informing regulatory evaluation and healthcare decision-making. Administrative claims provide population-scale, longitudinal records of healthcare utilization, expenditure, and detailed coding of diagnoses, procedures, and medications, yet their potential as a substrate for healthcare foundation models remains largely unexplored. Here we present ReClaim, a generative transformer trained from scratch on 43.8 billion medical events from more than 200 million enrollees in the MarketScan claims data spanning 2008-2022. ReClaim models longitudinal trajectories across diagnoses, procedures, medications, and expenditure, and was scaled to 140 million, 700 million, and 1.7 billion parameters. Across over 1,000 disease-onset prediction tasks, ReClaim achieved a mean AUC of 75.6%, substantially outperforming disease-specific LightGBM (66.3%) and the transformer-based Delphi model (69.4%), with the largest gains for rare diseases. These advantages held across retrospective and prospective evaluations and in external validation on two independent datasets. Performance improved monotonically with scale, and post-training added 13.8 percentage points over pre-training alone. Beyond disease prediction, ReClaim captured financial outcomes and improved real-world evidence (RWE) analyses: for healthcare expenditure forecasting it increased explained variance from 0.28 to 0.37 relative to LightGBM, and in a target trial emulation it reduced systematic bias by 72% on average relative to Delphi. Together, these results establish administrative claims as a scalable substrate for healthcare foundation models and show that learned representations generalize across time periods and data sources, supporting disease surveillance, expenditure forecasting, and RWE generation.

cs.AI↗

arXiv2026

We Need Granular Sharing of De-Identified Data-But Will Patients Engage? Investigating Health System Leaders' and Patients' Perspectives on A Patient-Controlled Data-Sharing Platform

Patient-controlled data-sharing systems are increasingly promoted as a way to empower patients with greater autonomy over their health data. Yet it remains unclear how different stakeholders, especially patients and health system leaders, perceive the benefits and challenges of enabling granular control over the sharing of de-identified medical data for research. To address this gap, we developed a high-fidelity prototype of a patient-controlled, web-based consent platform and conducted a two-phase mixed-methods study:semi-structured interviews with 16 health system leaders and a survey with 523 patient participants. While both groups appreciated the potential of such a platform to enhance transparency and autonomy, their views diverged in meaningful ways. Leaders viewed transparency and granular control through the lens of informed consent and institutional ethics, whereas patients interpreted these factors as safeguards against potential risks and uncertainties. Our findings underscore critical tensions such as individual control and research integrity. We offer design implications for building trustworthy, context-aware systems that support flexible granularity, provide ongoing benefit-centered transparency, and adapt to diverse literacy and privacy needs.

cs.HC↗

arXiv2025

Near-Lossless Model Compression Enables Longer Context Inference in DNA Large Language Models

Trained on massive cross-species DNA corpora, DNA large language models (LLMs) learn the fundamental "grammar" and evolutionary patterns of genomic sequences. This makes them powerful priors for DNA sequence modeling, particularly over long ranges. However, two major constraints hinder their use in practice: the quadratic computational cost of self-attention and the growing memory required for key-value (KV) caches during autoregressive decoding. These constraints force the use of heuristics such as fixed-window truncation or sliding windows, which compromise fidelity on ultra-long sequences by discarding distant information. We introduce FOCUS (Feature-Oriented Compression for Ultra-long Self-attention), a progressive context-compression module that can be plugged into pretrained DNA LLMs. FOCUS combines the established k-mer representation in genomics with learnable hierarchical compression: it inserts summary tokens at k-mer granularity and progressively compresses attention key and value activations across multiple Transformer layers, retaining only the summary KV states across windows while discarding ordinary-token KV. A shared-boundary windowing scheme yields a stationary cross-window interface that propagates long-range information with minimal loss. We validate FOCUS on an Evo-2-based DNA LLM fine-tuned on GRCh38 chromosome 1 with self-supervised training and randomized compression schedules to promote robustness across compression ratios. On held-out human chromosomes, FOCUS achieves near-lossless fidelity: compressing a 1 kb context into only 10 summary tokens (about 100x) shifts the average per-nucleotide probability by only about 0.0004. Compared to a baseline without compression, FOCUS reduces KV-cache memory and converts effective inference scaling from O(N^2) to near-linear O(N), enabling about 100x longer inference windows on commodity GPUs with near-lossless fidelity.

q-bio.GN↗

arXiv2025

LEME: Open Large Language Models for Ophthalmology with Advanced Reasoning and Clinical Validation

The rising prevalence of eye diseases poses a growing public health burden. Large language models (LLMs) offer a promising path to reduce documentation workload and support clinical decision-making. However, few have been tailored for ophthalmology, and most evaluations focus mainly on knowledge-based QA without clinically relevant benchmarks or real-world validation. Here, we present LEME, a suite of open-weight LLMs developed through a two-stage process: (1) instruction tuning on 200,000 samples from clinical guidelines, textbooks, and case reports to enhance reasoning and task-following, and (2) reinforcement learning with ~30,000 preference labels to enhance accuracy and informativeness. LEME was evaluated on five curated zero-shot benchmarks spanning tasks such as patient QA, consultation, and treatment planning. It outperformed all seven baselines (all p < 0.004), exceeding GPT-4o by 3.32% (absolute ROUGE-L gain). It was further evaluated on three downstream tasks using deidentified patient data, reviewed by clinicians. In patient QA, LEME received the highest ratings from attending clinicians in 3 out of 4 criteria, with scores of 4.67 for factuality, 4.77 for specificity, 4.79 for completeness, and 4.88 for safety (1-5 scale). Its completeness score surpassed that of expert-written answers (4.79 vs. 4.56; p = 0.015). In visual acuity extraction, LEME achieved the highest F1, outperforming LLaMA-3 by 14.1% and Eye-LLaMA by 59.0%. In a pilot evaluation on assessment and treatment planning for diabetic retinopathy, AMD, and glaucoma, LEME received scores of 4.36 for factuality, 4.55 for specificity, 4.42 for completeness, and 4.36 for safety, approaching attending-level performance. All models, data, and code will be released to support further development and clinical translation, laying the groundwork for improved efficiency and patient care

cs.CL↗

arXiv2025

Supporting Patients in Managing Electronic Health Records and Biospecimens Consent for Research: Insights from a Mixed-Methods Usability Evaluation of the iAGREE Portal

De-identified health data are frequently used in research. As AI advances heighten the risk of re-identification, it is important to respond to concerns about transparency, data privacy, and patient preferences. However, few practical and user-friendly solutions exist. We developed iAGREE, a patient-centered electronic consent management portal that allows patients to set granular preferences for sharing electronic health records and biospecimens with researchers. To refine the iAGREE portal, we conducted a mixed-methods usability evaluation with 40 participants from three U.S. health systems. Our results show that the portal received highly positive usability feedback. Moreover, participants identified areas for improvement, suggested actionable enhancements, and proposed additional features to better support informed granular consent while reducing patient burden. Insights from this study may inform further improvements to iAGREE and provide practical guidance for designing patient-centered consent management tools.

cs.HC↗

arXiv2025

ByzSFL: Achieving Byzantine-Robust Secure Federated Learning with Zero-Knowledge Proofs

The advancement of AI models, especially those powered by deep learning, faces significant challenges in data-sensitive industries like healthcare and finance due to the distributed and private nature of data. Federated Learning (FL) and Secure Federated Learning (SFL) enable collaborative model training without data sharing, enhancing privacy by encrypting shared intermediate results. However, SFL currently lacks effective Byzantine robustness, a critical property that ensures model performance remains intact even when some participants act maliciously. Existing Byzantine-robust methods in FL are incompatible with SFL due to the inefficiency and limitations of encryption operations in handling complex aggregation calculations. This creates a significant gap in secure and robust model training. To address this gap, we propose ByzSFL, a novel SFL system that achieves Byzantine-robust secure aggregation with high efficiency. Our approach offloads aggregation weight calculations to individual parties and introduces a practical zero-knowledge proof (ZKP) protocol toolkit. This toolkit supports widely used operators for calculating aggregation weights, ensuring correct computations without compromising data privacy. Not only does this method maintain aggregation integrity, but it also significantly boosts computational efficiency, making ByzSFL approximately 100 times faster than existing solutions. Furthermore, our method aligns with open-source AI trends, enabling plaintext publication of the final model without additional information leakage, thereby enhancing the practicality and robustness of SFL in real-world applications.

cs.CR↗

arXiv2025

Enhancing Patient-Centric Communication: Leveraging LLMs to Simulate Patient Perspectives

Large Language Models (LLMs) have demonstrated impressive capabilities in role-playing scenarios, particularly in simulating domain-specific experts using tailored prompts. This ability enables LLMs to adopt the persona of individuals with specific backgrounds, offering a cost-effective and efficient alternative to traditional, resource-intensive user studies. By mimicking human behavior, LLMs can anticipate responses based on concrete demographic or professional profiles. In this paper, we evaluate the effectiveness of LLMs in simulating individuals with diverse backgrounds and analyze the consistency of these simulated behaviors compared to real-world outcomes. In particular, we explore the potential of LLMs to interpret and respond to discharge summaries provided to patients leaving the Intensive Care Unit (ICU). We evaluate and compare with human responses the comprehensibility of discharge summaries among individuals with varying educational backgrounds, using this analysis to assess the strengths and limitations of LLM-driven simulations. Notably, when LLMs are primed with educational background information, they deliver accurate and actionable medical guidance 88% of the time. However, when other information is provided, performance significantly drops, falling below random chance levels. This preliminary study shows the potential benefits and pitfalls of automatically generating patient-specific health information from diverse populations. While LLMs show promise in simulating health personas, our results highlight critical gaps that must be addressed before they can be reliably used in clinical settings. Our findings suggest that a straightforward query-response model could outperform a more tailored approach in delivering health information. This is a crucial first step in understanding how LLMs can be optimized for personalized health communication while maintaining accuracy.

cs.AI↗

arXiv2024

Me LLaMA: Foundation Large Language Models for Medical Applications

Recent advancements in large language models (LLMs) like ChatGPT and LLaMA show promise in medical applications, yet challenges remain in medical language comprehension. This study presents Me-LLaMA, a new medical LLM family based on open-source LLaMA models, optimized for medical text analysis and diagnosis by leveraging large-scale, domain-specific datasets. The Me-LLaMA family, including foundation models Me-LLaMA 13/70B and their chat-enhanced versions, was developed through continued pre-training and instruction tuning with 129B tokens and 214K samples from biomedical and clinical sources. Training the 70B models required over 100,000 A100 GPU hours. Me-LLaMA's performance was evaluated across six medical text analysis tasks using 12 benchmark datasets and complex clinical case diagnosis, with automatic and human evaluations. Results indicate Me-LLaMA outperforms LLaMA and other open-source medical LLMs in zero-shot and supervised settings. Task-specific tuning further boosts performance, surpassing ChatGPT on 7 of 8 datasets and GPT-4 on 5 of 8. For complex clinical cases, Me-LLaMA achieves performance comparable to ChatGPT and GPT-4. This work underscores the importance of domain-specific data in developing medical LLMs and addresses the high computational costs involved in training, highlighting a balance between pre-training and fine-tuning strategies. Me-LLaMA models are now accessible under user agreements, providing a valuable resource for advancing medical AI.

cs.CL↗

arXiv2018

The Impact of Automatic Pre-annotation in Clinical Note Data Element Extraction - the CLEAN Tool

Objective. Annotation is expensive but essential for clinical note review and clinical natural language processing (cNLP). However, the extent to which computer-generated pre-annotation is beneficial to human annotation is still an open question. Our study introduces CLEAN (CLinical note rEview and ANnotation), a pre-annotation-based cNLP annotation system to improve clinical note annotation of data elements, and comprehensively compares CLEAN with the widely-used annotation system Brat Rapid Annotation Tool (BRAT). Materials and Methods. CLEAN includes an ensemble pipeline (CLEAN-EP) with a newly developed annotation tool (CLEAN-AT). A domain expert and a novice user/annotator participated in a comparative usability test by tagging 87 data elements related to Congestive Heart Failure (CHF) and Kawasaki Disease (KD) cohorts in 84 public notes. Results. CLEAN achieved higher note-level F1-score (0.896) over BRAT (0.820), with significant difference in correctness (P-value < 0.001), and the mostly related factor being system/software (P-value < 0.001). No significant difference (P-value 0.188) in annotation time was observed between CLEAN (7.262 minutes/note) and BRAT (8.286 minutes/note). The difference was mostly associated with note length (P-value < 0.001) and system/software (P-value 0.013). The expert reported CLEAN to be useful/satisfactory, while the novice reported slight improvements. Discussion. CLEAN improves the correctness of annotation and increases usefulness/satisfaction with the same level of efficiency. Limitations include untested impact of pre-annotation correctness rate, small sample size, small user size, and restrictedly validated gold standard. Conclusion. CLEAN with pre-annotation can be beneficial for an expert to deal with complex annotation tasks involving numerous and diverse target data elements.

cs.CL↗

arXiv2018

Efficient Determination of Equivalence for Encrypted Data

Secure computation of equivalence has fundamental application in many different areas, including healthcare. We study this problem in the context of matching an individual identity to link medical records across systems. We develop an efficient solution for equivalence based on existing work that can evaluate the greater than relation. We implement the approach and demonstrate its effectiveness on data, as well as demonstrate how it meets regulatory criteria for risk.

cs.CR↗

arXiv2018

ModelChain: Decentralized Privacy-Preserving Healthcare Predictive Modeling Framework on Private Blockchain Networks

Cross-institutional healthcare predictive modeling can accelerate research and facilitate quality improvement initiatives, and thus is important for national healthcare delivery priorities. For example, a model that predicts risk of re-admission for a particular set of patients will be more generalizable if developed with data from multiple institutions. While privacy-protecting methods to build predictive models exist, most are based on a centralized architecture, which presents security and robustness vulnerabilities such as single-point-of-failure (and single-point-of-breach) and accidental or malicious modification of records. In this article, we describe a new framework, ModelChain, to adapt Blockchain technology for privacy-preserving machine learning. Each participating site contributes to model parameter estimation without revealing any patient health information (i.e., only model data, no observation-level data, are exchanged across institutions). We integrate privacy-preserving online machine learning with a private Blockchain network, apply transaction metadata to disseminate partial models, and design a new proof-of-information algorithm to determine the order of the online learning process. We also discuss the benefits and potential issues of applying Blockchain technology to solve the privacy-preserving healthcare predictive modeling task and to increase interoperability between institutions, to support the Nationwide Interoperability Roadmap and national healthcare delivery priorities such as Patient-Centered Outcomes Research (PCOR).

cs.CY↗

arXiv2016

Prerequisites for International Exchanges of Health Information: Comparison of Australian, Austrian, Finnish, Swiss, and US Privacy Policies

Capabilities to exchange health information are critical to accelerate discovery and its diffusion to healthcare practice. However, the same ethical and legal policies that protect privacy hinder these data exchanges, and the issues accumulate if moving data across geographical or organizational borders. This can be seen as one of the reasons why many health technologies and research findings are limited to very narrow domains. In this paper, we compare how using and disclosing personal data for research purposes is addressed in Australian, Austrian, Finnish, Swiss, and US policies with a focus on text data analytics. Our goal is to identify approaches and issues that enable or hinder international health information exchanges. As expected, the policies within each country are not as diverse as across countries. Most policies apply the principles of accountability and/or adequacy and are thereby fundamentally similar. Their following requirements create complications with re-using and re-disclosing data and even secondary data: 1) informing data subjects about the purposes of data collection and use, before the dataset is collected; 2) assurance that the subjects are no longer identifiable; and 3) destruction of data when the research activities are finished. Using storage and compute cloud services as well as other exchange technologies on the Internet without proper permissions is technically not allowed if the data are stored in another country. Both legislation and technologies are available as vehicles for overcoming these barriers. The resulting richness in information variety will contribute to the development and evaluation of new clinical hypotheses and technologies.

cs.CY↗

arXiv2014

Natural Language Processing in Biomedicine: A Unified System Architecture Overview

In modern electronic medical records (EMR) much of the clinically important data - signs and symptoms, symptom severity, disease status, etc. - are not provided in structured data fields, but rather are encoded in clinician generated narrative text. Natural language processing (NLP) provides a means of "unlocking" this important data source for applications in clinical decision support, quality assurance, and public health. This chapter provides an overview of representative NLP systems in biomedicine based on a unified architectural view. A general architecture in an NLP system consists of two main components: background knowledge that includes biomedical knowledge resources and a framework that integrates NLP tools to process text. Systems differ in both components, which we will review briefly. Additionally, challenges facing current research efforts in biomedical NLP include the paucity of large, publicly available annotated corpora, although initiatives that facilitate data sharing, system evaluation, and collaborative work between researchers in clinical NLP are starting to emerge.

cs.CL↗

arXiv2012

Enhancing Twitter Data Analysis with Simple Semantic Filtering: Example in Tracking Influenza-Like Illnesses

Systems that exploit publicly available user generated content such as Twitter messages have been successful in tracking seasonal influenza. We developed a novel filtering method for Influenza-Like-Illnesses (ILI)-related messages using 587 million messages from Twitter micro-blogs. We first filtered messages based on syndrome keywords from the BioCaster Ontology, an extant knowledge model of laymen's terms. We then filtered the messages according to semantic features such as negation, hashtags, emoticons, humor and geography. The data covered 36 weeks for the US 2009 influenza season from 30th August 2009 to 8th May 2010. Results showed that our system achieved the highest Pearson correlation coefficient of 98.46% (p-value<2.2e-16), an improvement of 3.98% over the previous state-of-the-art method. The results indicate that simple NLP-based enhancements to existing approaches to mine Twitter data can increase the value of this inexpensive resource.

cs.SI↗

arXiv2012

Predicting accurate probabilities with a ranking loss

In many real-world applications of machine learning classifiers, it is essential to predict the probability of an example belonging to a particular class. This paper proposes a simple technique for predicting probabilities based on optimizing a ranking loss, followed by isotonic regression. This semi-parametric technique offers both good ranking and regression performance, and models a richer set of probability distributions than statistical workhorses such as logistic regression. We provide experimental results that show the effectiveness of this technique on real-world applications of probability prediction.

cs.LG↗