Wei-Hung Weng
Research Areas
Authored Publications
Sort By
Towards conversational artificial intelligence for disease management
Khaled Saab
David Stutz
Yong Cheng
Kavita Kulkarni
Sara Mahdavi
Joelle Barral
James Manyika
Ryutaro Tanno
Adam Rodman
Nature (2026)
Preview abstract
Although large language models have shown promise in diagnostic dialogue, their capabilities for effective management reasoning, including disease progression, therapeutic response and safe medication prescription, have remained underexplored. We have advanced the previously demonstrated diagnostic capabilities of the Articulate Medical Intelligence Explorer (AMIE) using a new large-language-model-based agentic system optimized for multivisit clinical management and dialogue. To ground the reasoning of AMIE in authoritative clinical knowledge, we leveraged the long-context capabilities of Gemini, combining in-context retrieval with structured reasoning to align its output with up-to-date clinical practice guidelines and drug formularies. In a randomized, blinded virtual Objective Structured Clinical Examination study, AMIE was compared to 21 primary care physicians (PCPs) across 100 multivisit case scenarios designed to reflect the guidance of the UK National Institute for Health and Care Excellence and BMJ Best Practice guidelines. AMIE was non-inferior to PCPs in management reasoning, as assessed by specialists, and scored better both with respect to preciseness of treatment and investigation, and in terms of its alignment with and grounding in clinical guidelines. To benchmark medication reasoning, we developed RxQA, a multiple-choice question benchmark that was derived from two national drug formularies (from the USA and UK) and validated by board-certified pharmacists. Although AMIE and PCPs both benefited from the ability to access external drug information, AMIE outperformed PCPs on higher-difficulty questions. Although further research will be needed before real-world translation of AMIE, its strong performance across evaluations marks a significant step towards use of conversational artificial intelligence as a tool in disease management.
View details
A prospective clinical feasibility study of a conversational diagnostic AI in an ambulatory primary care clinic
Peter Brodeur
Jacob M. Koshy
Khaled Saab
Ava Homiar
Roma Ruparel
Charles Wu
Ryutaro Tanno
Joseph Xu
Amy Wang
David Stutz
Hannah M. Ferrera
David Barrett
Lindsey Crowley
Jihyeon Lee
Spencer E. Rittner
Selena K. Zhang
Elahe Vedadi
Christine G. Kohn
Kavita Kulkarni
Vinay Kadiyala
Sara Mahdavi
Wendy Du
David Feinbloom
Renee Wong
Petar Sirkovic
Alessio Orlandi
Juro Gottweis
Joelle Barral
Kat Chou
James Manyika
Rob Fields
Jonathan X. Li
Marc L. Cohen
Adam Rodman
arXiv (2026)
Preview abstract
Large language model (LLM)-based AI systems have shown promise for patient-facing diagnostic and management conversations in simulated settings. Translating these systems into clinical practice requires assessment in real-world workflows with rigorous safety oversight. We report a prospective, single-arm feasibility study of an LLM-based conversational AI, the Articulate Medical Intelligence Explorer (AMIE), conducting clinical history taking and presentation of potential diagnoses for patients to discuss with their provider at urgent care appointments at a leading academic medical center. 100 adult patients completed an AMIE text-chat interaction up to 5 days before their appointment. We sought to assess the conversational safety and quality, patient and clinician experience, and clinical reasoning capabilities compared to primary care providers (PCPs). Human safety supervisors monitored all patient-AMIE interactions in real time and did not need to intervene to stop any consultations based on pre-defined criteria. Patients reported high satisfaction and their attitudes towards AI improved after interacting with AMIE (p < 0.001). PCPs found AMIE's output useful with a positive impact on preparedness. AMIE's differential diagnosis (DDx) included the final diagnosis, per chart review 8 weeks post-encounter, in 90% of cases, with 75% top-3 accuracy. Blinded assessment of AMIE and PCP DDx and management (Mx) plans suggested similar overall DDx and Mx plan quality, without significant differences for DDx (p = 0.6) and appropriateness and safety of Mx (p = 0.1 and 1.0, respectively). PCPs outperformed AMIE in the practicality (p = 0.003) and cost effectiveness (p = 0.004) of Mx. While further research is needed, this study demonstrates the initial feasibility, safety, and user acceptance of conversational AI in a real-world setting, representing crucial steps towards clinical translation.
View details
A large language model for complex cardiology care
Jack W O’Sullivan
Khaled Saab
Daniel K. Amponsah
Evaline Cheng
Yong Cheng
Emily Chu
Yaanik Desai
Aly Elezaby
Muhammad Fazal
Tasmeen Hussain
Sneha S. Jain
Daniel Seung Kim
Roy Lan
Jiwen Li
Wilson Tang
Natalie Tapaskar
Victoria Parikh
Ryan Sandoval
Gabriela Spencer-Bonilla
Bryan Wu
Kavita Kulkarni
Philip Mansfield
Juro Gottweis
Joelle Barral
Ryutaro Tanno
Sara Mahdavi
Euan Ashley
Nature Medicine (2026)
Preview abstract
The scarcity of subspecialist medical expertise poses a considerable challenge for healthcare delivery. This issue is particularly acute in cardiology, where timely, accurate management determines outcomes. We explored the potential of Articulate Medical Intelligence Explorer (AMIE), a large language model-based experimental medical artificial intelligence system, to augment clinical decision-making in this challenging context. We conducted a randomized controlled trial comparing large language model-assisted care with the usual care of complex patients suspected of having a genetic cardiomyopathy, and we curated a real-world dataset of complex cases from a subspecialist cardiology practice. Nine participating general cardiologists were provided with access to both clinical text reports and raw diagnostic data—including electrocardiograms, echocardiograms, cardiac magnetic resonance imaging scans and cardiopulmonary exercise testing—and were randomized to manage these cases, either with or without assistance from AMIE. We developed a ten-domain evaluation rubric used by three blinded subspecialists to evaluate the quality of triage, diagnosis and management. In our randomized controlled trial with retrospective patient data, subspecialists favored large language model-assisted responses overall, and for the management plan and diagnostic testing domains, with the remaining domains considered a tie. Overall, subspecialists preferred AMIE-assisted cardiology assessments 46.7% of the time, compared with 32.7% for cardiologists alone (P = 0.02), with 20.6% rated as a tie. Subspecialists also quantified errors, extra and missing content, reasoning and potential bias. Cardiologists alone had more clinically significant errors (24.3% versus 13.1%, P = 0.033) and more missing content (37.4% versus 17.8%, P = 0.0021) than cardiologists assisted by AMIE. Lastly, cardiologists who used AMIE reported that AMIE helped their assessment more than half the time (57.0%) and saved time in 50.5% of cases.
View details
AI mirrors experimental science to uncover a novel mechanism of gene transfer crucial to bacterial evolution
Juro Gottweis
Jose R Penades
Alexander Daryin
Artiom Myaskovsky
Tiago R D Costa
Cell (2025)
Preview abstract
Note this is a re-submission of a previously approved ITP. The previous approval was conditional for a journal pre-sub enquiry only and we are submitting a new ITP for the preprint of the paper.
AI models have been proposed for hypothesis generation, but testing their ability to drive
high-impact research is challenging, since an AI-generated hypothesis can take decades to
validate. Here, we challenge the ability of a recently developed LLM-based platform to
generate high-level hypotheses by posing a question that took years to resolve
experimentally but remained unpublished: How could capsid-forming phage-inducible
chromosomal islands (cf-PICIs) spread across bacterial species? Remarkably, the AI’s top-
ranked hypothesis matched our experimentally confirmed mechanism: cf-PICIs hijack
diverse phage tails to expand their host range. We critically assess the AI’s five highest-
ranked hypotheses, showing that some opened new research avenues in our laboratories.
We benchmark its performance against other LLMs and outline best practices for integrating
AI into scientific discovery. Our findings suggest that AI can act not just as a computational
tool, but as a creative engine, accelerating discovery and reshaping how we generate and
test scientific hypotheses.
View details
Exploring large language models for specialist-level oncology care
Vikram Dhillon
Polly Niravath
Preethi Prasad
Khaled Saab
Ryutaro Tanno
Yong Cheng
Hanh Mai
Ethan Burns
Zainub Ajmal
Kavita Kulkarni
Philip Mansfield
Joelle Barral
Juro Gottweis
Sara Mahdavi
NEJM AI (2025)
Preview abstract
Large language models have shown rapid progress in encoding clinical knowledge and demonstrating clinical reasoning. However, their capabilities in subspecialty or complex medical settings remain underexplored. In this work, we probe the performance of Articulate Medical Intelligence Explorer (AMIE), a conversational diagnostic AI system in the subspecialty of breast oncology care without specific fine-tuning to this challenging domain. To perform this evaluation, we curated a set of 60 synthetic breast cancer vignettes representing a range of treatment-naive, treatment-refractory, and rare histology cases encountered in a community-based breast oncology clinic. We developed a detailed clinical rubric for evaluating management plans, including axes such as the quality of case summarization, safety of the proposed care plan, and recommendations for treatment (i.e., chemotherapy, radiotherapy, surgery, and hormonal therapy). To improve performance, we enhanced AMIE with the inference-time ability to perform web search retrieval to gather relevant and up-to-date clinical knowledge and refine its responses with a multistage, self-critique pipeline. We compare the response quality of AMIE with that of internal medicine trainees, oncology fellows, and general oncology attendings under both automated and specialist clinician evaluations. Although our evaluations were limited to a few physicians, AMIE outperformed trainees and fellows, demonstrating the potential of the system in this important domain. However, AMIE’s performance was overall inferior to that of attending oncologists, suggesting that further prospective research is needed.
View details
Predicting Cardiovascular Disease Risk using Photoplethysmography and Deep Learning
Sebastien Baur
Mayank Daswani
Christina Chen
Sujay Kakarmath
Mariam Jabara
Babak Behsaz
Shravya Shetty
Goodarz Danaei
Diego Ardila
PLOS Global Public Health, 4(6) (2024), e0003204
Preview abstract
Cardiovascular diseases (CVDs) are responsible for a large proportion of premature deaths in low- and middle-income countries. Early CVD detection and intervention is critical in these populations, yet many existing CVD risk scores require a physical examination or lab measurements, which can be challenging in such health systems due to limited accessibility. We investigated the potential to use photoplethysmography (PPG), a sensing technology available on most smartphones that can potentially enable large-scale screening at low cost, for CVD risk prediction. We developed a deep learning PPG-based CVD risk score (DLS) to predict the probability of having major adverse cardiovascular events (MACE: non-fatal myocardial infarction, stroke, and cardiovascular death) within ten years, given only age, sex, smoking status and PPG as predictors. We compare the DLS with the office-based refit-WHO score, which adopts the shared predictors from WHO and Globorisk scores (age, sex, smoking status, height, weight and systolic blood pressure) but refitted on the UK Biobank (UKB) cohort. All models were trained on a development dataset (141,509 participants) and evaluated on a geographically separate test (54,856 participants) dataset, both from UKB. DLS’s C-statistic (71.1%, 95% CI 69.9–72.4) is non-inferior to office-based refit-WHO score (70.9%, 95% CI 69.7–72.2; non-inferiority margin of 2.5%, p<0.01) in the test dataset. The calibration of the DLS is satisfactory, with a 1.8% mean absolute calibration error. Adding DLS features to the office-based score increases the C-statistic by 1.0% (95% CI 0.6–1.4). DLS predicts ten-year MACE risk comparable with the office-based refit-WHO score. Interpretability analyses suggest that the DLS-extracted features are related to PPG waveform morphology and are independent of heart rate. Our study provides a proof-of-concept and suggests the potential of a PPG-based approach strategies for community-based primary prevention in resource-limited regions.
View details
An intentional approach to managing bias in embedding models
Atilla P. Kiraly
Jungyeon Park
Rory Pilgrim
Charles Lau
Heather Cole-Lewis
Shravya Shetty
Krish Eswaran
Leo Anthony Celi
The Lancet Digital Health, 6 (2024), E126-E130
Preview abstract
Advances in machine learning for health care have brought concerns about bias from the research community; specifically, the introduction, perpetuation, or exacerbation of care disparities. Reinforcing these concerns is the finding that medical images often reveal signals about sensitive attributes in ways that are hard to pinpoint by both algorithms and people. This finding raises a question about how to best design general purpose pretrained embeddings (GPPEs, defined as embeddings meant to support a broad array of use cases) for building downstream models that are free from particular types of bias. The downstream model should be carefully evaluated for bias, and audited and improved as appropriate. However, in our view, well intentioned attempts to prevent the upstream components—GPPEs—from learning sensitive attributes can have unintended consequences on the downstream models. Despite producing a veneer of technical neutrality, the resultant end-to-end system might still be biased or poorly performing. We present reasons, by building on previously published data, to support the reasoning that GPPEs should ideally contain as much information as the original data contain, and highlight the perils of trying to remove sensitive attributes from a GPPE. We also emphasise that downstream prediction models trained for specific tasks and settings, whether developed using GPPEs or not, should be carefully designed and evaluated to avoid bias that makes models vulnerable to issues such as distributional shift. These evaluations should be done by a diverse team, including social scientists, on a diverse cohort representing the full breadth of the patient population for which the final model is intended.
View details
Optimizing Audio Augmentations for Contrastive Learning of Health-Related Acoustic Signals
Louis Blankemeier
Sebastien Baur
Diego Ardila
Zaid Nabulsi
arXiv (2023)
ELIXR: Towards a general purpose X-ray artificial intelligence system through alignment of large language models and radiology vision encoders
Shawn Xu
Lin Yang
Christopher Kelly
Timo Kohlberger
Martin Ma
Atilla Kiraly
Sahar Kazemzadeh
Zakkai Melamed
Jungyeon Park
Patricia MacWilliams
Chuck Lau
Christina Chen
Mozziyar Etemadi
Sreenivasa Raju Kalidindi
Kat Chou
Shravya Shetty
Daniel Golden
Rory Pilgrim
Krish Eswaran
arxiv (2023)
Preview abstract
Our approach, which we call Embeddings for Language/Image-aligned X-Rays, or ELIXR, leverages a language-aligned image encoder combined or grafted onto a fixed LLM, PaLM 2, to perform a broad range of tasks. We train this lightweight adapter architecture using images paired with corresponding free-text radiology reports from the MIMIC-CXR dataset. ELIXR achieved state-of-the-art performance on zero-shot chest X-ray (CXR) classification (mean AUC of 0.850 across 13 findings), data-efficient CXR classification (mean AUCs of 0.893 and 0.898 across five findings (atelectasis, cardiomegaly, consolidation, pleural effusion, and pulmonary edema) for 1% (~2,200 images) and 10% (~22,000 images) training data), and semantic search (0.76 normalized discounted cumulative gain (NDCG) across nineteen queries, including perfect retrieval on twelve of them). Compared to existing data-efficient methods including supervised contrastive learning (SupCon), ELIXR required two orders of magnitude less data to reach similar performance. ELIXR also showed promise on CXR vision-language tasks, demonstrating overall accuracies of 58.7% and 62.5% on visual question answering and report quality assurance tasks, respectively. These results suggest that ELIXR is a robust and versatile approach to CXR AI.
View details
Preview abstract
Class imbalance is a common problem in medical diagnosis, causing a standard classifier to be biased towards majority classes and ignore the importance of the rest. This is especially true for dermatology, a specialty with thousands of skin conditions but many of which rarely occur in the wild. Buoyed by recent advances, we explore meta-learning based few-shot learning approaches in skin condition recognition problem and propose an evaluation setup to fairly assess the real-world impact of such approaches. When compared to conventional class imbalance techniques, we find that the state-of-the-art few-shot learning methods are not as performant, but combining the two approaches using a novel ensemble leads to improvement in all-way classification, especially the rare classes. We conclude that the ensemble can be useful to address the class imbalance problem, yet progress here can further be accelerated by the use of real-world evaluation setups for benchmarking new methods.
View details
