Claudia Tanner

By Claudia Tanner

0

Share

Newsweek is a Trust Project member

See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.


Developing a new drug can take years and cost billions of dollars. Yet even treatments that show promise in early research often fail in clinical trials.

One challenge is identifying the patients most likely to benefit, understanding how diseases may progress and predicting how people might respond to new treatments.

Labcorp believes artificial intelligence could help address these problems by creating digital models of human biology, sometimes known as ‘digital biological twins’.

Read More on Health Care Vertical

It's like a flight simulator for drug development: these virtual models could allow researchers to explore how diseases develop, test different treatment scenarios and identify potential problems before testing drugs in real people.

Bola Oyegunwa, Labcorp’s chief information and technology officer, said the company hopes to use its extensive diagnostic data to help develop the technology. “At Labcorp, we are a diagnostics company, and we perform over 750 million tests annually, and that is a significant amount of data that we have internally,” he said at Newsweek’s AI Health Summit in New York City in September.

Labcorp hopes to harness this vast database, including pathology data and medical images, to build detailed models of how diseases progress and affect the body.

“The strategy was: Are we able to convert all of this vast data into some sort of clinical digital signature that allows us to start correlating the expression of biomarkers to disease progression?” Oyegunwa said.

The longer-term ambition was to use these models to simulate clinical trials from beginning to end

While digital twins are being explored in several areas of medicine, established applications remain relatively narrow, often focusing on specific organs, such as the heart or specific diseases.

In an early clinical study, researchers at Johns Hopkins University created 3D digital replicas of patients' diseased hearts to identify the source of dangerous irregular heart rhythms. Doctors tested treatment procedures on the virtual models before carrying them out on the patients.

Computer-generated patients have also already been used in clinical research to help analyse trial results. Some models predict how individual participants might have fared under a placebo or standard treatment, potentially allowing researchers to reduce the number of people needed in control groups while still conducting trials with real patients.

The terms virtual patient and digital twin are sometimes used interchangeably, but there is a distinction. A virtual patient is generally a hypothetical person modelled using data from a wider population, while a digital twin aims to represent a specific real person, or part of their biology, using their own medical information.

However, creating a digital model of an entire person is still a long way off. One obstacle is that medical test results are not always recorded consistently across health care providers. Differences in terminology and reporting can make information difficult to combine and compare, potentially affecting the accuracy of AI systems trained on large medical datasets.

Researchers also need to establish how accurately digital models predict real patient outcomes before relying on them to guide drug development.

For now, digital twins are not a substitute for clinical trials. But if Labcorp's vision becomes reality, the technology could help researchers make better decisions before a trial begins, potentially saving time and resources, improving the chances of developing effective medicines and reducing the risk of exposing patients to treatments unlikely to benefit them.

View 0 Comments

Request Reprint & Licensing

Submit Correction

View Editorial & AI Guidelines