Dr. William Hsu Received a One-year Grant from UCLA Institute for Digital Research and Education (IDRE) and Institute for Informatics (I2)
Dr. William Hsu, a Visiting Assistant Professor affiliated with the Medical Imaging Informatics Group, has been awarded the grant to implement a research data repository for Radiology, which would provide users with a dedicated infrastructure to access de-identified patient data for retrospective studies, imaging biomarker validation, and education. He will be working with colleagues from the Office of Research Affairs and Medical Imaging Informatics to provide computational tools that analyze patient records, allowing users to search for patients with specific diseases, demographics, image findings, and interventions with higher precision.
In today's increasingly data-intensive environment, a significant challenge is matching the ability to generate data with a comparable ability to understand, analyze, and derive new insights from this data. The intent of this funding is to foster the practical application of technologies, services, and policies to the real data management challenges that researchers and educators face as the nature of scholarly work becomes increasingly interdisciplinary.
Medical data, in particular, is associated with unique challenges related to the large quantity of data being generated on a routine basis, the unstructured nature of this data, and patient privacy considerations.
While the UCLA Clinical and Translational Science Institute (CTSI) has started to create a research data repository for the medical enterprise as a whole, its efforts are not targeted to the specific information needs of a radiologist.
Funding from this grant will support a pilot implementation of a data repository that incorporates novel informatics tools to create and maintain a collection of clinical text and imaging data for radiology researchers and educators in a secure manner.
Privacy is an important issue when working with patient data. The repository will act as an "honest broker" by providing users with a means to search and obtain de-identified, descriptive information about the patient population and establishing a process by which researchers can obtain additional data after obtaining prerequisite institutional approvals. In addition, this project addresses several practical challenges:
how to store and maintain data being collected from different clinical sources (hospital information system, picture archive and communication system);
how to transform unstructured data (e.g., free-text radiology reports) into a form that is amenable to computational reasoning and retrieval;
how to unify the representation of data elements with common standards to enable sharing and aggregation across multiple institutions.
This effort will enable secondary use of clinical data to support the educational mission of the Department and facilitate the validation of imaging biomarkers for screening, diagnosis, and treatment assessment of diseases.