Back to Our Technologies

ConsultBot: Generative AI-Powered Software to Improve Interpretation of Arterial Blood Gas Results

A generative AI-powered software for rapid, streamlined, and easy-to-use clinical interpretation of arterial blood gas results to lower the risk of missing or incorrectly identifying complex disorders during patient diagnosis and care.

Diagnostics

Project Overview

Arterial blood gas analysis tests for oxygenation levels and the acid/base balance in a person’s blood.
Current methods for ABG analysis vary with wide user interpretation variability.
Dana-Farber has developed ConsultBot, generative AI-powered software for rapid, streamlined, and easy-to-use interpretation of ABG results for reliable clinical decision-making.
ConsultBot is in the process of further testing and validation.
ConsultBot is available for licensing.

Technology

Arterial blood gas (ABG) analysis is an essential part of managing a patient’s blood oxygenation status and acid–base balance to aid in diagnosis and guide treatment. Disorders of acid–base balance can create complications in many disease states, with abnormalities occasionally being a life-threatening risk factor. However, its usefulness is dependent on correctly interpreting the results, which can vary from clinician to clinician. Current methods can be difficult to interpret, particularly for complex ABG profiles when multiple coexistent acid-base abnormalities exist.

To fill that unmet need, Praveen Meka, MD, an inpatient hospitalist at Dana-Farber, partnered with Amazon Web Services (AWS) Bedrock platform to build and scale a generative AI-powered tool to standardize and enhance the interpretation of ABGs, thereby providing clinicians with a reliable adjunct tool.

The result is ConsultBot, a software solution that rapidly and robustly analyzes and interprets ABG results to provide better-informed clinical management options for patients. Meka and his Amazon colleagues described their protocol in an AWS blog.

Arterial blood gas tests include measurements of a patient’s pH, oxygen, carbon dioxide, acid, and base levels – all factors used to diagnose and monitor patients and aid in selecting treatment options. Misinterpreting complex ABGs can lead to delays in diagnosis and treatment, which can have a profound effect on patient care. Therefore, solutions that are able to standardize the interpretation and serve as an adjunct to the clinician to help make decisions can lower the risk of missing or incorrectly identifying complex disorders and help improve patient care.

The model’s output provides a detailed analysis of the ABG, identifying potential acid-base disorders and offering differential diagnoses. The protocol reduces the interpretation time from one to ten minutes down to a few seconds. As importantly, ConsultBot serves as a solid starting point for clinicians when assessing a patient’s clinical status.

First developed as a proof-of-concept software tool, Meka and colleagues are currently improving the accuracy and plan to clinically validate ConsultBot with key end users such as frontline clinicians, hospitalists, and intensivists.

Team Members: Praveen Meka, MD, MBA, Christine Tsien Silvers, MD, PhD, Qing Liu, Bharath Gunapati

Applications

ConsultBot’s primary use is improving and correcting the interpretation of ABG results, which are critical for patient care. The goal is to integrate the software into the established install base of blood gas analyzers in clinical laboratories around the globe, allowing for seamless integration and interpretation with electronic medical records.

Opportunities

ConsultBot is available for licensing to companies interested in further developing and integrating the software tool into their diagnostic devices and solutions.

Get in touch