Skip to main content

Royal Philips, US DoD Developing AI-Based Tool to Predict Infections

NEW YORK – Royal Philips and the US Department of Defense said on Tuesday that they are collaborating on a project to develop an early warning algorithm to detect infection before people show signs or symptoms.

Amsterdam-based Philips is collaborating with the Defense Threat Reduction Agency (DTRA) and the Defense Innovation Unit of the DoD in a large-scale empirical exploration to predict pre-symptomatic infection in humans. The 18-month project, called Rapid Analysis of Threat Exposure (RATE), is a component of efforts to improve military personnel readiness but is broadly applicable in healthcare settings, Philips said.

The firm noted that an early warning system that facilitates faster diagnosis and treatment of infection can reduce a person's downtime and aid in quickly containing the spread of communicable diseases by enabling quicker isolation of people exposed to infections.

A prototype developed during the project has shown that use of artificial intelligence to evaluate combinations of vital signs and other biomarkers may predict infection up to 48 hours in advance of clinical suspicion, including observable symptoms, Philips said.

The approach applies large-scale data machine learning and trade-space analyses — evaluation of alternative ways of achieving outcomes — across 165 different biomarkers. The investigators uncovered disease markers from a Philips dataset consisting of more than 41,000 hospital-acquired infection cases extracted from a repository with data associated with more than 7 million hospital-based patient encounters.

Project investigators used HAI cases as a surrogate dataset for infection in military personnel and as a basis to develop a predictive algorithm of disease.

The algorithm predicted infection 48 hours before clinical suspicion with an area-under-the-curve of 0.853, a level of performance between that achieved by blood-based breast and prostate cancer screening tests and an enzyme immunoassay used as a first-tier Lyme disease test, Philips said.

Joseph Frassica, chief medical officer and head of research for Philips North America, said in a statement that the firm believes the project data derived from acute-care settings are adaptable to evaluating active duty personnel. "These results can be extended in future work to also apply to other healthcare settings, and to include scenarios where vital signs and biomarkers fluctuate as a function of factors such as physical exertion and heat stress," he said.

Future research is being planned to leverage the project to develop an algorithm and integrate it into a wearable device to enable non-invasive monitoring of a soldier’s health and allow early alerts of potential infection. Further, the technology may be applied to help monitor hospital patients for infection prior to clinical symptoms, the firm said.