AMIA Joint Summits on Translational Science proceedings. AMIA Joint Summits on Translational Science最新文献

Logistic regression is a widely used model in machine learning, particularly as a baseline for binary classification tasks due to its simplicity, effectiveness, and interpretability. It is especially powerful when dealing with categorical features. Despite its advantages, standard logistic regression fails to capture the distributional and geometric structure of data, especially when features are derived from structured spaces like brain imaging. For instance, in Voxel-Based Morphometry (VBM), measurements from distinct brain regions follow a clear spatial organization, which standard logistic regression cannot fully leverage. In this paper, we propose Sinkhorn Logistic Regression (SLR), a variant of logistic regression that incorporates the Sinkhorn divergence as a loss function. This adaptation enables the model to leverage geometric information about the data distribution, enhancing its performance on structured datasets.

This paper introduces a user-friendly platform developed by the University of Kentucky Center for Applied AI, designed to make customized large language models (LLMs) more accessible. By capitalizing on recent advancements in multi-LoRA inference, the system efficiently accommodates custom adapters for a diverse range of users and projects. The paper outlines the system's architecture and key features, encompassing dataset curation, model training, secure inference, and text-based feature extraction. We illustrate the establishment of a tenant-aware computational network using agent-based methods, securely utilizing islands of isolated resources as a unified system. The platform strives to deliver secure, affordable LLM services, emphasizing process and data isolation, end-to-end encryption, and role-based resource authentication. This contribution aligns with the overarching goal of enabling simplified access to cutting-edge AI models and technology in support of scientific discovery and the development of biomedical informatics.

The progress in natural language processing (NLP) using large language models (LLMs) has greatly improved patient information extraction from clinical narratives. However, most methods based on the fine-tuning strategy have limited transfer learning ability for cross-domain applications. This study proposed a novel approach that employs a soft prompt-based learning architecture, which introduces trainable prompts to guide LLMs toward desired outputs. We examined two types of LLM architectures, including encoder-only GatorTron and decoder-only GatorTronGPT, and evaluated their performance for the extraction of social determinants of health (SDoH) using a cross-institution dataset from the 2022 n2c2 challenge and a cross-disease dataset from the University of Florida (UF) Health. The results show that decoder-only LLMs with prompt tuning achieved better performance in cross-domain applications. GatorTronGPT achieved the best F1 scores for both datasets, outperforming traditional fine-tuned GatorTron by 8.9% and 21.8% in a cross-institution setting, and 5.5% and 14.5% in a cross-disease setting.

Federal- and state-level governance as well as local institutional oversight are changing rapidly to address the accelerated growth in the usage of digital health technologies (DHT) -such as apps, wearables, and websites-to enable clinical and translational research. While studies have described frameworks for assessing and/or implementing individual DHTs, to our knowledge there are none describing how to implement and support multiple DHTs at an academic medical center (AMC). A multi-disciplinary team including information technology, institutional review board, legal, and privacy professionals identified 33 items to evaluate as part of onboarding studies using DHTs. In a one-year period at one AMC, we applied the novel instrument to review 98 requests for research (93) and non-research (5) use of DHTs. The 33-item instrument may be valuable to researchers and practitioners in other settings seeking to scale institutional support for DHTs.

Automatic text summarization (ATS) is an emerging technology to assist clinicians in providing continuous and coordinated care. This study presents an approach to summarize doctor-patient dialogues using generative large language models (LLMs). We developed prompt-tuning algorithms to instruct generative LLMs to summarize clinical text. We examined the prompt-tuning strategies, the size of soft prompts, and the few-short learning ability of GatorTronGPT, a generative clinical LLM developed using 277 billion clinical and general English words with up to 20 billion parameters. We compared GatorTronGPT with a previous solution based on fine-tuning of a widely used T5 model, using a clinical benchmark dataset MTS-DIALOG. The experimental results show that the GatorTronGPT-20B model achieved the best performance on all evaluation metrics. The proposed solution has a low computing cost as the LLM parameters are not updated during prompt-tuning. This study demonstrates the efficiency of generative clinical LLMs for clinical ATS through prompt tuning.

Falls among the elderly and especially those with NeuroDegenerative Disorders (NDD) reduces life expectancy. The purpose of this study is to explore the role of Machine Learning on Electronic Health Records (EHR) data for time-to-event survival analysis prediction of injuries, and role of sensitive attributes, e.g., Race, Ethnicity, Sex, in these models. We used multiple survival analysis methods on a cohort of 29,045 patients 65 years and older treated at PennMedicine for either NDD, Mild Cognitive Impairment (MCI), or another disease. We compare the algorithms and explore the role of multiple modalities on improving prediction of injuries among NDD patients, specifically medications and laboratory tests. Overall, we found that medication features resulted in either increased Hazard Ratios (HR) or reduced HR depending on the NDD type. We found that being of Black race significantly increased the risk offall/injury in the models that included only medication and sensitive attribute features. The combined model that used both modalities (medications and laboratory information) removed this relationship between being of Black race and increases in fall/injury. Therefore, we found that combining modalities in these survival models in the prediction offall/injury risk among NDD and MCI individuals results in findings that are robust to different Racial and Ethnic groups with no biases apparent in our final combined modality results. Furthermore, combining modalities (both medications and laboratory values) improved the survival analysis performance across multiple survival analysis methods, when compared using the C-index.

Diagnosing Alzheimer's Disease (AD) early and cost-effectively is crucial. Recent advancements in Large Language Models (LLMs) like ChatGPT have made accurate, affordable AD detection feasible. Yet, HIPAA compliance and the challenge of integrating these models into hospital systems limit their use. Addressing these constraints, we introduce ADetectoLocum, an open-source LLM equipped model designed for AD risk detection within hospital environments. This model evaluates AD risk through spontaneous patient speech, enhancing diagnostic processes without external data exchange. Our approach secures local deployment and significantly surpasses previous models in predictive accuracy for AD detection, especially in early-stage identification. ADetectoLocum therefore offers a reliable solution for AD diagnostics in healthcare institutions.

Explainable AI (XAI) techniques are necessary to help clinicians make sense of AI predictions and integrate predictions into their decision-making workflow. In this work, we conduct a survey study to understand clinician preference among different XAI techniques when they are used to interpret model predictions over text-based EHR data. We implement four XAI techniques (LIME, Attention-based span highlights, exemplar patient retrieval, and free-text rationales generated by LLMs) on an outcome prediction model that uses ICU admission notes to predict a patient's likelihood of experiencing in-hospital mortality. Using these XAI implementations, we design and conduct a survey study of 32 practicing clinicians, collecting their feedback and preferences on the four techniques. We synthesize our findings into a set of recommendations describing when each of the XAI techniques may be more appropriate, their potential limitations, as well as recommendations for improvement.

We investigated the association between systole and diastole, and outpatient portal use during pregnancy. We used electronic and administrative data from our academic medical center. We categorized patients into two groups: (<140 mm Hg; <90 mm Hg), and out-of-range (≥140 mm Hg, ≥ 90 mm Hg). Random effects linear regression models examined the association between mean trimester blood pressure (BP) levels and portal use, adjusting for covariates. As portal use increased, both systole and diastole levels decreased for the out-of-range group. These differences were statistically significant for patients who were initially out-of-range. For the in-range group, systole and diastole levels were stable as portal use increased. Results provide evidence to support a relationship between outpatient portal use and BP outcomes during pregnancy. More research is needed to expand on our findings, especially those focused on the implementation and design of outpatient portals for pregnancy.

As genomic research continues to advance, sharing of genomic data and research outcomes has become increasingly important for fostering collaboration and accelerating scientific discovery. However, such data sharing must be balanced with the need to protect the privacy of individuals whose genetic information is being utilized. This paper presents a bidirectional framework for evaluating privacy risks associated with data shared (both in terms of summary statistics and research datasets) in genomic research papers, particularly focusing on re-identification risks such as membership inference attacks (MIA). The framework consists of a structured workflow that begins with a questionnaire designed to capture researchers' (authors') self-reported data sharing practices and privacy protection measures. Responses are used to calculate the risk of re-identification for their study (paper) when compared with the National Institutes of Health (NIH) genomic data sharing policy. Any gaps in compliance help us to identify potential vulnerabilities and encourage the researchers to enhance their privacy measures before submitting their research for publication. The paper also demonstrates the application of this framework, using published genomic research as case study scenarios to emphasize the importance of implementing bidirectional frameworks to support trustworthy open science and genomic data sharing practices.