The National University Health System has recently showcased the capability of its new AI platform, which among things can predict hospital bed availability up to two weeks ahead of time. 

Its ENDEAVOUR AI platform integrates live data from the next-generation EMR (NGEMR) system to compute multiple AI insights. 

It hosts multiple AI tools, one of which can predict the estimated length of stay of each patient admitted to three public hospitals under NUHS. The AI tool does this by reading patients' histories and doctor notes in real time, running up to 30 times per hour. It can also provide clinical insights into factors contributing to a patient's prolonged stay. 

According to a press statement, the AI's accuracy had been validated using NGEMR data over the past six months. It is said that it can predict hospital bed states up to two weeks in advance to optimise bed capacity and patient placement. 

WHY IT MATTERS

The predictive tool is NUHS's solution to rising bed occupancy rates and increasing bed wait times at EDs. It enables doctors to intervene early in anticipation of problems. For instance, it can flag patients who have been staying past two weeks in the hospital, allowing medical teams to either change their management or plan a patient's early transfer to a community hospital for rehabilitation.

With its ability to read notes, vital signs, and other test reports, the AI tool can also predict an admitted patient's risk of deterioration.

NUHS plans to further develop the AI's ability to recommend care plans "that [can] change the trajectory of a patient’s disease course."

Meanwhile, ENDEAVOUR AI also has the ability to automatically alert administrators of climbing ED wait times, allowing manpower resources to be activated early. Wait times can be cut down from 30 minutes to hours, depending on how resources are deployed.

THE LARGER TREND

NUHS also disclosed its plan to evaluate and later deploy an imaging AI model to enhance scoliosis x-ray assessment. Each year, about 7,000 x-rays are screened through Singapore's scoliosis screening programme. Doctors are still manually measuring the degree of spinal curvature, which can be time-consuming and prone to errors. Results can also take time to be communicated.

A new AI model is being designed to automatically measure the degree of scoliosis, improving doctors' ability to interpret scans. Initial trials found the AI model to reduce reporting time with fair accuracy, potentially helping raise clinicians' productivity and communicate results and specialist referrals and treatment earlier. 

ON THE RECORD

"We leverage AI to improve healthcare practices and outcomes, enabling clinical practitioners to make faster, more accurate diagnoses and precise treatments. Healthcare institutions today aggregate vast quantities of data but most of this data is analysed retrospectively. With the technology in ENDEAVOUR AI, we can now stream data in real time, feeding AI models that produce actionable insights on the fly, resulting in better patient outcomes," said Ngiam Kee Yuan, associate professor and CTO at NUHS.