Search for: All records

Abstract Applying machine learning to clinical outcome prediction is challenging due to imbalanced datasets and sensitive tasks that contain rare yet critical outcomes and where equitable treatment across diverse patient groups is essential. Despite attempts, biases in predictions persist, driven by disparities in representation and exacerbated by the scarcity of positive labels, perpetuating health inequities. This paper introduces , a synthetic data generation approach leveraging large language models, to address these issues. enhances algorithmic performance and reduces bias by creating realistic, anonymous synthetic patient data that improves representation and augments dataset patterns while preserving privacy. Through experiments on multiple datasets, we demonstrate that boosts mortality prediction performance across diverse subgroups, achieving up to a 21% improvement in F1 Score without requiring additional data or altering downstream training pipelines. Furthermore, consistently reduces subgroup performance gaps, as shown by universal improvements in performance and fairness metrics across four experimental setups. 

Abstract This study reports a comprehensive environmental scan of the generative AI (GenAI) infrastructure in the national network for clinical and translational science across 36 institutions supported by the CTSA Program led by the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH) at the United States. Key findings indicate a diverse range of institutional strategies, with most organizations in the experimental phase of GenAI deployment. The results underscore the need for a more coordinated approach to GenAI governance, emphasizing collaboration among senior leaders, clinicians, information technology staff, and researchers. Our analysis reveals that 53% of institutions identified data security as a primary concern, followed by lack of clinician trust (50%) and AI bias (44%), which must be addressed to ensure the ethical and effective implementation of GenAI technologies. 

Government agencies collect and manage a wide range of ever-growing datasets. While such data has the potential to support research and evidence-based policy making, there are concerns that the dissemination of such data could infringe upon the privacy of the individuals (or organizations) from whom such data was collected. To appraise the current state of data sharing, as well as learn about opportunities for stimulating such sharing at a faster pace, a virtual workshop was held on May 21st and 26th, 2021, sponsored by the National Science Foundation and National Institute of Standards and Technologies, where a multinational collection of researchers and practitioners were brought together to discuss their experiences and learn about recently developed technologies for managing privacy while sharing data. The workshop specifically focused on challenges and successes in government data sharing at various levels. The first day focused on successful examples of new technology applied to sharing of public data, including formal privacy techniques, synthetic data, and cryptographic approaches. Day two emphasized brainstorming sessions on some of the challenges and directions to address them. 

Recent studies suggest that genomic data can be matched to images of human faces, raising the concern that genomic data can be re-identified with relative ease. However, such investigations assume access to well-curated images, which are rarely available in practice and challenging to derive from photos not generated in a controlled laboratory setting. In this study, we reconsider re-identification risk and find that, for most individuals, the actual risk posed by linkage attacks to typical face images is substantially smaller than claimed in prior investigations. Moreover, we show that only a small amount of well-calibrated noise, imperceptible to humans, can be added to images to markedly reduce such risk. The results of this investigation create an opportunity to create image filters that enable individuals to have better control over re-identification risk based on linkage. 

The COVID-19 pandemic highlights the need for broad dissemination of case surveillance data. Local and global public health agencies have initiated efforts to do so, but there remains limited data available, due in part to concerns over privacy. As a result, current COVID-19 case surveillance data sharing policies are based on strong adversarial assumptions, such as the expectation that an attacker can readily re-identify individuals based on their distinguishability in a dataset. There are various re-identification risk measures to account for adversarial capabilities; however, the current array insufficiently accounts for real world data challenges - particularly issues of missing records in resources of identifiable records that adversaries may rely upon to execute attacks (e.g., 10 50-year-old male in the de-identified dataset vs. 5 50-year-old male in the identified dataset). In this paper, we introduce several approaches to amend such risk measures and assess re-identification risk in light of how an attacker's capabilities relate to missing records. We demonstrate the potential for these measures through a record linkage attack using COVID-19 case surveillance data and voter registration records in the state of Florida. Our findings demonstrate that adversarial assumptions, as realized in a risk measure, can dramatically affect re-identification risk estimation. Notably, we show that the re-identification risk is likely to be substantially smaller than the typical risk thresholds, which suggests that more detailed data could be shared publicly than is currently the case. 

Abstract Objective Supporting public health research and the public’s situational awareness during a pandemic requires continuous dissemination of infectious disease surveillance data. Legislation, such as the Health Insurance Portability and Accountability Act of 1996 and recent state-level regulations, permits sharing deidentified person-level data; however, current deidentification approaches are limited. Namely, they are inefficient, relying on retrospective disclosure risk assessments, and do not flex with changes in infection rates or population demographics over time. In this paper, we introduce a framework to dynamically adapt deidentification for near-real time sharing of person-level surveillance data. Materials and Methods The framework leverages a simulation mechanism, capable of application at any geographic level, to forecast the reidentification risk of sharing the data under a wide range of generalization policies. The estimates inform weekly, prospective policy selection to maintain the proportion of records corresponding to a group size less than 11 (PK11) at or below 0.1. Fixing the policy at the start of each week facilitates timely dataset updates and supports sharing granular date information. We use August 2020 through October 2021 case data from Johns Hopkins University and the Centers for Disease Control and Prevention to demonstrate the framework’s effectiveness in maintaining the PK11 threshold of 0.01. Results When sharing COVID-19 county-level case data across all US counties, the framework’s approach meets the threshold for 96.2% of daily data releases, while a policy based on current deidentification techniques meets the threshold for 32.3%. Conclusion Periodically adapting the data publication policies preserves privacy while enhancing public health utility through timely updates and sharing epidemiologically critical features.