Dr. Kai-Wei Chang is a new Assistant Professor in the Computer Science Department at UCLA. After completing training in computer science with a specialization in machine learning (ML) and natural language processing (NLP), he decided to build his career in translating cutting-edge ML and NLP techniques into the biomedical and health domains to enable real-world impact. Specifically, although electronic health records (EHRs) now capture detailed observations related to patient care, much of the information is locked in unstructured text (e.g., physician notes, admission/discharge reports), ultimately hindering the large-scale, deeper analyses needed to inform clinical decision support tools. In this context, Dr. Chang has set his overarching career goal to develop novel methods for extracting information from free-text, applying the techniques to collaborate with other clinical scientists' development and embedding of quantitative approaches in research and applications. The goal of this KL2 supplement award aligns perfectly with this objective: with his partnering clinician scientist, Dr. John Mafi, they will build a new generation of automated low-value care EHR-based metric by leveraging these computational approaches. These eMeasures will have profound impact by establishing a scalable, objective framework for conducting quality measurements, with the potential to become a model for broad national adoption. Supporting Dr. Chang's efforts in this project is an outstanding group of experienced mentors: my primary mentor, Dr. Alex Bui (Director, Medical & Imaging Informatics (MII) Group); Dr. Catherine Sarkisian (Director, UCLA Center for Value-based Care); and Dr. Wei Wang (Director, Scalable Analytics Institute). Under their mentorship he will acquire skills in translating ML/NLP into clinical/health domains and grow his understanding of implementation and dissemination science. To foster Dr. Chang's career development, he will work with his mentors to conduct direct reading, attend relevant seminars, and publish in top-tier conferences and journals. Dr. Chang proposes two specific aims for his research: 1) to extract information related to proton pump inhibitor (PPI) usage and its appropriate usage criteria from unstructured free-text notes in the EHR, adapting cutting-edge NLP and ML techniques; and 2) to use extracted information to inform and evaluate eMeasures around PPI usage, developing ML-based models that inform the reliability of the metrics. He will apply his knowledge of computational approaches and work with his clinician-scientist partner in the analyses described in the proposal and the implementation of eMeasures, with the goal of submitting our developments to the National Quality Forum (NQF) for endorsement and dissemination.
This project seeks to translate cutting-edge machine learning and natural language processing techniques to support automated metrics around low-value healthcare (?eMeasures?) by extracting information from electronic health records (EHRs) ? a timely and potentially transformative innovation in the quality of care field. The work develops data-driven information extraction techniques for processing clinical notes, which can also support other clinical decision tasks. The long-term goal of the effort is to leverage EHR data to formulate a new, efficient way of understanding and implementing quality measurements in the United States.
Showing the most recent 10 out of 36 publications
