Yindalon Aphinyanaphongs

Yuxuan Hu, Albert Lui, Mark Goldstein et al.

Myocardial infarction and heart failure are major cardiovascular diseases that affect millions of people in the US. The morbidity and mortality are highest among patients who develop cardiogenic shock. Early recognition of cardiogenic shock is critical. Prompt implementation of treatment measures can prevent the deleterious spiral of ischemia, low blood pressure, and reduced cardiac output due to cardiogenic shock. However, early identification of cardiogenic shock has been challenging due to human providers' inability to process the enormous amount of data in the cardiac intensive care unit (ICU) and lack of an effective risk stratification tool. We developed a deep learning-based risk stratification tool, called CShock, for patients admitted into the cardiac ICU with acute decompensated heart failure and/or myocardial infarction to predict onset of cardiogenic shock. To develop and validate CShock, we annotated cardiac ICU datasets with physician adjudicated outcomes. CShock achieved an area under the receiver operator characteristic curve (AUROC) of 0.820, which substantially outperformed CardShock (AUROC 0.519), a well-established risk score for cardiogenic shock prognosis. CShock was externally validated in an independent patient cohort and achieved an AUROC of 0.800, demonstrating its generalizability in other cardiac ICUs.

Hao Zhang, Neil Jethani, Aahlad Puli et al.

Diabetes has a long asymptomatic period which can often remain undiagnosed for multiple years. In this study, we trained a deep learning model to detect new-onset diabetes using 12-lead ECG and readily available demographic information. To do so, we used retrospective data where patients have both a hemoglobin A1c and ECG measured. However, such patients may not be representative of the complete patient population. As part of the study, we proposed a methodology to evaluate our model in the target population by estimating the probability of receiving an A1c test and reweight the retrospective population to represent the general population. We also adapted an efficient algorithm to generate Shapley values for both ECG signals and demographic features at the same time for model interpretation. The model offers an automated, more accurate method for early diabetes detection compared to current screening efforts. Their potential use in wearable devices can facilitate large-scale, community-wide screening, improving healthcare outcomes.

Krithik Vishwanath, Mrigayu Ghosh, Anton Alyakin et al.

Specialized clinical AI assistants are rapidly entering medical practice, often framed as safer or more reliable than general-purpose large language models (LLMs). Yet, unlike frontier models, these clinical tools are rarely subjected to independent, quantitative evaluation, creating a critical evidence gap despite their growing influence on diagnosis, triage, and guideline interpretation. We assessed two widely deployed clinical AI systems (OpenEvidence and UpToDate Expert AI) against three state-of-the-art generalist LLMs (GPT-5, Gemini 3 Pro, and Claude Sonnet 4.5) using a 1,000-item mini-benchmark combining MedQA (medical knowledge) and HealthBench (clinician-alignment) tasks. Generalist models consistently outperformed clinical tools, with GPT-5 achieving the highest scores, while OpenEvidence and UpToDate demonstrated deficits in completeness, communication quality, context awareness, and systems-based safety reasoning. These findings reveal that tools marketed for clinical decision support may often lag behind frontier LLMs, underscoring the urgent need for transparent, independent evaluation before deployment in patient-facing workflows.

Salman Rahman, Lavender Yao Jiang, Saadia Gabriel et al.

Advances in large language models (LLMs) provide new opportunities in healthcare for improved patient care, clinical decision-making, and enhancement of physician and administrator workflows. However, the potential of these models importantly depends on their ability to generalize effectively across clinical environments and populations, a challenge often underestimated in early development. To better understand reasons for these challenges and inform mitigation approaches, we evaluated ClinicLLM, an LLM trained on [HOSPITAL]'s clinical notes, analyzing its performance on 30-day all-cause readmission prediction focusing on variability across hospitals and patient characteristics. We found poorer generalization particularly in hospitals with fewer samples, among patients with government and unspecified insurance, the elderly, and those with high comorbidities. To understand reasons for lack of generalization, we investigated sample sizes for fine-tuning, note content (number of words per note), patient characteristics (comorbidity level, age, insurance type, borough), and health system aspects (hospital, all-cause 30-day readmission, and mortality rates). We used descriptive statistics and supervised classification to identify features. We found that, along with sample size, patient age, number of comorbidities, and the number of words in notes are all important factors related to generalization. Finally, we compared local fine-tuning (hospital specific), instance-based augmented fine-tuning and cluster-based fine-tuning for improving generalization. Among these, local fine-tuning proved most effective, increasing AUC by 0.25% to 11.74% (most helpful in settings with limited data). Overall, this study provides new insights for enhancing the deployment of large language models in the societally important domain of healthcare, and improving their performance for broader populations.

Lavender Y. Jiang, Angelica Chen, Xu Han et al.

Hospitals and healthcare systems rely on operational decisions that determine patient flow, cost, and quality of care. Despite strong performance on medical knowledge and conversational benchmarks, foundation models trained on general text may lack the specialized knowledge required for these operational decisions. We introduce Lang1, a family of models (100M-7B parameters) pretrained on a specialized corpus blending 80B clinical tokens from NYU Langone Health's EHRs and 627B tokens from the internet. To rigorously evaluate Lang1 in real-world settings, we developed the REalistic Medical Evaluation (ReMedE), a benchmark derived from 668,331 EHR notes that evaluates five critical tasks: 30-day readmission prediction, 30-day mortality prediction, length of stay, comorbidity coding, and predicting insurance claims denial. In zero-shot settings, both general-purpose and specialized models underperform on four of five tasks (36.6%-71.7% AUROC), with mortality prediction being an exception. After finetuning, Lang1-1B outperforms finetuned generalist models up to 70x larger and zero-shot models up to 671x larger, improving AUROC by 3.64%-6.75% and 1.66%-23.66% respectively. We also observed cross-task scaling with joint finetuning on multiple tasks leading to improvement on other tasks. Lang1-1B effectively transfers to out-of-distribution settings, including other clinical tasks and an external health system. Our findings suggest that predictive capabilities for hospital operations require explicit supervised finetuning, and that this finetuning process is made more efficient by in-domain pretraining on EHR. Our findings support the emerging view that specialized LLMs can compete with generalist models in specialized tasks, and show that effective healthcare systems AI requires the combination of in-domain pretraining, supervised finetuning, and real-world evaluation beyond proxy benchmarks.

Anton Alyakin, Jaden Stryker, Daniel Alexander Alber et al.

General-purpose vision-language models (VLMs) demonstrate impressive capabilities, but their opaque training on uncurated internet data posse critical limitations for high-stakes decision-making, such as in neurosurgery. We present CNS-Obsidian, a neurosurgical VLM trained on peer-reviewed neurosurgical literature, and demonstrate its clinical utility compared with GPT-4o in a real-world setting. We compiled 23,984 articles from Neurosurgery Publications journals, yielding 78,853 figures and captions. Using GPT-4o and Claude Sonnet-3.5, we converted these image-text pairs into 263,064 training samples across three formats: instruction fine-tuning, multiple-choice questions, and differential diagnosis. We trained CNS-Obsidian, a fine-tune of the 34-billion parameter LLaVA-Next model. In a blinded, randomized deployment trial at NYU Langone Health (Aug 30-Nov 30, 2024), neurosurgeons were assigned to use either CNS-Obsidian or GPT-4o as a diagnostic co-pilot after patient consultations. Primary outcomes were diagnostic helpfulness and accuracy. CNS-Obsidian matched GPT-4o on synthetic questions (76.13% vs 77.54%, p=0.235), but only achieved 46.81% accuracy on human-generated questions versus GPT-4o's 65.70% (p<10-15). In the randomized trial, 70 consultations were evaluated (32 CNS-Obsidian, 38 GPT-4o) from 959 total consults. CNS-Obsidian received positive ratings in 40.62% of cases versus 57.89% for GPT-4o (p=0.230). Both models included correct diagnosis in approximately 60% of cases (59.38% vs 65.79%, p=0.626). Domain-specific VLMs trained on curated scientific literature can approach frontier model performance in specialized medical domains despite being orders of magnitude smaller and less expensive to train. However, low clinical utilization suggests chatbot interfaces may not align with specialist workflows, indicating need for alternative AI integration strategies.

Neil Jethani, Mukund Sudarshan, Yindalon Aphinyanaphongs et al.

While the need for interpretable machine learning has been established, many common approaches are slow, lack fidelity, or hard to evaluate. Amortized explanation methods reduce the cost of providing interpretations by learning a global selector model that returns feature importances for a single instance of data. The selector model is trained to optimize the fidelity of the interpretations, as evaluated by a predictor model for the target. Popular methods learn the selector and predictor model in concert, which we show allows predictions to be encoded within interpretations. We introduce EVAL-X as a method to quantitatively evaluate interpretations and REAL-X as an amortized explanation method, which learn a predictor model that approximates the true data generating distribution given any subset of the input. We show EVAL-X can detect when predictions are encoded in interpretations and show the advantages of REAL-X through quantitative and radiologist evaluation.

Anuroop Sriram, Matthew Muckley, Koustuv Sinha et al.

The rapid spread of COVID-19 cases in recent months has strained hospital resources, making rapid and accurate triage of patients presenting to emergency departments a necessity. Machine learning techniques using clinical data such as chest X-rays have been used to predict which patients are most at risk of deterioration. We consider the task of predicting two types of patient deterioration based on chest X-rays: adverse event deterioration (i.e., transfer to the intensive care unit, intubation, or mortality) and increased oxygen requirements beyond 6 L per day. Due to the relative scarcity of COVID-19 patient data, existing solutions leverage supervised pretraining on related non-COVID images, but this is limited by the differences between the pretraining data and the target COVID-19 patient data. In this paper, we use self-supervised learning based on the momentum contrast (MoCo) method in the pretraining phase to learn more general image representations to use for downstream tasks. We present three results. The first is deterioration prediction from a single image, where our model achieves an area under receiver operating characteristic curve (AUC) of 0.742 for predicting an adverse event within 96 hours (compared to 0.703 with supervised pretraining) and an AUC of 0.765 for predicting oxygen requirements greater than 6 L a day at 24 hours (compared to 0.749 with supervised pretraining). We then propose a new transformer-based architecture that can process sequences of multiple images for prediction and show that this model can achieve an improved AUC of 0.786 for predicting an adverse event at 96 hours and an AUC of 0.848 for predicting mortalities at 96 hours. A small pilot clinical study suggested that the prediction accuracy of our model is comparable to that of experienced radiologists analyzing the same information.

Oded Nov, Yindalon Aphinyanaphongs, Yvonne W. Lui et al.

One of the dramatic trends at the intersection of computing and healthcare has been patients' increased access to medical information, ranging from self-tracked physiological data to genetic data, tests, and scans. Increasingly however, patients and clinicians have access to advanced machine learning-based tools for diagnosis, prediction, and recommendation based on large amounts of data, some of it patient-generated. Consequently, just as organizations have had to deal with a "Bring Your Own Device" (BYOD) reality in which employees use their personal devices (phones and tablets) for some aspects of their work, a similar reality of "Bring Your Own Algorithm" (BYOA) is emerging in healthcare with its own challenges and support demands. BYOA is changing patient-clinician interactions and the technologies, skills and workflows related to them. In this paper we argue that: (1) BYOA is changing the patient-clinician relationship and the nature of expert work in healthcare, and (2) better patient-clinician-information-interpretation relationships can be facilitated with solutions that integrate technological and organizational perspectives.

Farah E. Shamout, Yiqiu Shen, Nan Wu et al.

During the coronavirus disease 2019 (COVID-19) pandemic, rapid and accurate triage of patients at the emergency department is critical to inform decision-making. We propose a data-driven approach for automatic prediction of deterioration risk using a deep neural network that learns from chest X-ray images and a gradient boosting model that learns from routine clinical variables. Our AI prognosis system, trained using data from 3,661 patients, achieves an area under the receiver operating characteristic curve (AUC) of 0.786 (95% CI: 0.745-0.830) when predicting deterioration within 96 hours. The deep neural network extracts informative areas of chest X-ray images to assist clinicians in interpreting the predictions and performs comparably to two radiologists in a reader study. In order to verify performance in a real clinical setting, we silently deployed a preliminary version of the deep neural network at New York University Langone Health during the first wave of the pandemic, which produced accurate predictions in real-time. In summary, our findings demonstrate the potential of the proposed system for assisting front-line physicians in the triage of COVID-19 patients.

Benedict Guzman, MS, Isabel Metzger et al.

In November 27, 2018, Amazon Web Services (AWS) released Amazon Comprehend Medical (ACM), a deep learning based system that automatically extracts clinical concepts (which include anatomy, medical conditions, protected health information (PH)I, test names, treatment names, and medical procedures, and medications) from clinical text notes. Uptake and trust in any new data product relies on independent validation across benchmark datasets and tools to establish and confirm expected quality of results. This work focuses on the medication extraction task, and particularly, ACM was evaluated using the official test sets from the 2009 i2b2 Medication Extraction Challenge and 2018 n2c2 Track 2: Adverse Drug Events and Medication Extraction in EHRs. Overall, ACM achieved F-scores of 0.768 and 0.828. These scores ranked the lowest when compared to the three best systems in the respective challenges. To further establish the generalizability of its medication extraction performance, a set of random internal clinical text notes from NYU Langone Medical Center were also included in this work. And in this corpus, ACM garnered an F-score of 0.753.

Vincent Major, Alisa Surkis, Yindalon Aphinyanaphongs

Conventional text classification models make a bag-of-words assumption reducing text into word occurrence counts per document. Recent algorithms such as word2vec are capable of learning semantic meaning and similarity between words in an entirely unsupervised manner using a contextual window and doing so much faster than previous methods. Each word is projected into vector space such that similar meaning words such as "strong" and "powerful" are projected into the same general Euclidean space. Open questions about these embeddings include their utility across classification tasks and the optimal properties and source of documents to construct broadly functional embeddings. In this work, we demonstrate the usefulness of pre-trained embeddings for classification in our task and demonstrate that custom word embeddings, built in the domain and for the tasks, can improve performance over word embeddings learnt on more general data including news articles or Wikipedia.

Josua Krause, Aritra Dasgupta, Jordan Swartz et al.

Human-in-the-loop data analysis applications necessitate greater transparency in machine learning models for experts to understand and trust their decisions. To this end, we propose a visual analytics workflow to help data scientists and domain experts explore, diagnose, and understand the decisions made by a binary classifier. The approach leverages "instance-level explanations", measures of local feature relevance that explain single instances, and uses them to build a set of visual representations that guide the users in their investigation. The workflow is based on three main visual representations and steps: one based on aggregate statistics to see how data distributes across correct / incorrect decisions; one based on explanations to understand which features are used to make these decisions; and one based on raw data, to derive insights on potential root causes for the observed patterns. The workflow is derived from a long-term collaboration with a group of machine learning and healthcare professionals who used our method to make sense of machine learning models they developed. The case study from this collaboration demonstrates that the proposed workflow helps experts derive useful knowledge about the model and the phenomena it describes, thus experts can generate useful hypotheses on how a model can be improved.