OpenELIS

Digital Initiatives Group (DIGI)

OpenELIS

OpenELIS: An open-source laboratory information system using next-gen AI to elevate clinical lab management, data insight, and decision-making.

Seattle, WA, USA

United States

Other, including part of a larger organization (please explain below)

DIGI is an informatics resource center based at the University of Washington

—

Public health laboratories are increasingly turning to open source solutions, with OpenELIS spanning both US and global health settings (~10 countries, 300+ public health laboratories, millions of laboratory tests). Laboratory data is critical to optimizing laboratory testing, clinical care, surveillance and outbreak control, and healthcare system management. In both US and global health settings, public health faces ever growing increased need for accessing and using laboratory data to care for the population. In addition, public health needs are evolving faster than ever before, bringing new demands for access to data in different ways. But often health professionals using the laboratory software systems are highly constrained by pre-defined and pre-built reports based on historical understanding, that limit the ability to slice and dice the data in different and new ways as the needs evolve (such as in rapidly shifting public health outbreaks like COVID-19). These health professionals are challenged in rapidly accessing the data in meaningful and dynamic ways to inform their decision making to improve laboratory management practices and clinical care as their needs evolve.

problem you are trying to solve. If relevant, share a link to a video of a product demo. Our solution will aim to empower laboratory system users (lab technicians, management, clinicians with portal access) the ability to use natural language prompts within the OpenELIS laboratory software user interface to rapidly access, collate, and summarize data in a variety of formats about lab testing, performance, and activities. This allows users to “build reports” dynamically about their laboratory data, as their needs evolve in decision making. The solution will be built to work within high connectivity and low connectivity settings, with the ability to utilize either cloud hosted or locally hosted LLM. Our approach leverages RAG to help the LLM model understand our data model and concept dictionary. We will use RAG to index and retrieve the SQL schema and dictionary. The table schemas will be stored, and convert the dictionary into a structured format like JSON. We envision a two part process for the model: First, taking the user’s question and generating a relevant SQL query; then, extracting that query’s results and turning that into information displayed to the clinician.

OpenELIS AI serves to target laboratory staff (technicians, management) in low-to-middle income countries (LMIC), clinicians that utilize lab data through portal access, and ultimately the patients that are cared for by the public health system that OpenELIS supports. This population suffers constraints in numerous ways that prevent more advanced access and novel summarization of data to improve patient outcomes; including, limited human resources for ongoing software development for creating new data reports and extracts, limited training for healthcare professionals in data analysis and use for decision making, and limited exchanges of data between systems for secondary data use outside of the data source. Our solution will reduce the need for additional ongoing investment into new reports and data extracts, as well as more advanced data training, as the user will be able to dynamically prompt for data using natural language at the time it is needed and in formats that are readily accessible. Lastly, users will be able to access summarized and analyzed data through a user interface they are already using every day, rather than accessing a different system requiring additional training.