Could an AI ‘digital twin’ help doctors choose the best treatment for ICU patients?

Researchers in the United States are developing artificial intelligence-powered “digital twins” of critically ill patients that could eventually allow doctors to test potential treatments virtually before administering them to the patient.

The University of Vermont is leading the ambitious project after receiving a research and development award worth up to US$38 million from the Advanced Research Projects Agency for Health (ARPA-H).

Known as ReSCUED — Reprogramming Severe Critical Illness Using Extensible Digital Twins — the project will focus on some of the most challenging patients treated in intensive care units, including people experiencing sepsis, severe trauma and burns.

The concept involves creating a continuously updated computational model — or digital twin — representing an individual patient’s immune and inflammatory response.

Researchers plan to collect blood samples from critically ill patients every six hours while also gathering physiological and clinical information. Those data will be fed into mathematical models designed to track how the patient’s condition is changing.

Artificial intelligence would then be used to analyze the enormous number of biological interactions occurring within the patient and explore how different treatment strategies could affect the course of the illness.

In effect, researchers hope clinicians will eventually be able to test potential interventions on the digital patient before deciding which approach may be most appropriate for the real patient.

The initial focus is on immune dysfunction.

In conditions such as sepsis, the body’s response to infection can become dangerously dysregulated, triggering widespread inflammation, tissue damage and organ failure. Modern intensive care can support failing organs with technologies such as ventilators and dialysis, but determining how to correct the underlying immune response remains extremely difficult.

The ReSCUED project aims to give clinicians a more detailed, individualized picture of that response.

The technology is not yet ready for routine clinical use.

During the first three years of the project, researchers plan to develop and validate the digital-twin technology and determine whether it can accurately predict patient outcomes. If those milestones are achieved, later phases would test the system further before progressing toward clinical trials involving critically ill patients.

Patient information for the project will be collected at three U.S. clinical sites, including Wake Forest University School of Medicine, the University of Alabama at Birmingham and Washington University School of Medicine.

The University of Vermont team will lead development of the computational models and digital-twin platform.

The project is part of ARPA-H’s broader Critical Illness Immunological Reprogramming and Control Point Learning Engine, or CIRCLE, program. The agency’s goal is to develop technologies capable of better measuring, modelling and ultimately controlling dangerous immune responses in critically ill patients.

One of the program’s ambitious targets is to reduce the length of ICU stays for critically ill patients by at least 25 per cent.

Whether digital twins can achieve that goal remains to be demonstrated. But if the technology proves successful, it could represent a significant shift toward highly personalized critical care — where treatment decisions are informed not only by a patient’s diagnosis, but by a continuously evolving computer model of that individual patient’s biology.

Source: University of Vermont; Advanced Research Projects Agency for Health (ARPA-H)

The post Could an AI ‘digital twin’ help doctors choose the best treatment for ICU patients? appeared first on Hospital News.