Hamilton Health Sciences Home
HHS surgeon Dr. Pablo E. Serrano, sits at a patient's bedside, holding an innovative AI tool surgeons are trialling.
HHS surgeons are testing an innovative AI tool that can catch signs of dangerous complications early in patients recovering from certain surgeries. Dr. Pablo E. Serrano, a surgeon at our Juravinski Hospital and Cancer Centre, is leading the trial.
January 15, 2025

Harnessing the power of AI to prevent deadly post-surgical infections

Hamilton Health Sciences (HHS) surgeons are testing an artificial intelligence (AI) tool that can catch signs of life-threatening complications early in patients recovering from certain colorectal, liver and pancreatic surgeries. This innovative AI tool, created by Waterloo medical technology company FluidAI Medical, could also save hospitals hundreds of thousands of dollars by reducing the need for certain medical tests, emergency department visits and readmissions.

Head and shoulders photo of Dr. Ted Scott

Dr. Ted Scott, HHS vice president of innovation and partnerships

“Through partnerships like this, we’re bring leading-edge innovations to HHS that are having a profound impact on patient care by saving and prolonging lives,” says Dr. Ted Scott, vice president of innovation and partnerships for HHS.

An early warning system

FluidAI is testing its Stream Platform AI system, including its proprietary Origin device, in an international trial with HHS, the primary hospital leading this innovation. Other hospitals in the study are partnering with HHS.

“This study has the potential to completely transform how we manage post-surgical complications.” — Dr. Pablo E. Serrano

This tool specializes in the early detection of potentially deadly leaks of bowel contents, like fecal matter or bile, into the abdominal cavity and blood stream in the days following surgery. Such post-surgical leaks sometimes happen when sections of bowel or bile duct have been cut and reattached, and sutures accidentally come loose.

Close up photo of the Fluid AI Origin sensor

There are two parts to the FluidAI Stream Platform – the Origin sensor platform to analyze fluid in real time, and a Delta monitor at the patient’s bedside that interprets and shows continuous data as it’s collected.

This complication, called an anastomotic leak, happens in about 20 per cent of pancreatic resection surgeries, says Dr. Pablo E. Serrano, an HHS surgeon who’s leading the trial. “Of that 20 per cent, five per cent of patients will die from complications from the leak, such as infections or sepsis, so it’s a very serious problem.”

Fifteen per cent of liver resections will have a bile leak after surgery, and five per cent of bowel resections will leak. It happens more often in pancreatic and liver resection surgeries because those organs are soft, making it more difficult for sutures to hold. The bowel is denser and muscular, making it easier to keep stitches in place.

How it works

There are two parts to the FluidAI Stream Platform – the Origin sensor platform to analyze fluid in real time, and a Delta monitor at the patient’s bedside that interprets and shows continuous data as it’s collected.

“With this technology, surgeons can rely on chemical analysis of leaked fluid to predict an infection, and intervene early, rather than waiting for symptoms to develop,” — Dr. Pablo E. Serrano

During an operation, the surgeon places a drainage system, which includes the sensor platform, close to where the bowel or bile duct has been reattached. This platform uses pH (acidity) and electrical conductivity sensors to analyze post-surgical fluid in real time, to determine if there’s a higher risk of leakage.

The system also uses Epic, HHS’ state-of-the-art electronic medical records system, to help evaluate risk by incorporating parameters from the patient’s history, surgical details, and routine post-operative information such as age, blood loss during surgery and number of blood transfusions, into its analysis.

“If a high risk for leakage is determined, the surgeon is immediately notified through the Epic system and can intervene early to prevent or manage infections,” says Serrano.

Early detection

The current standard of practice is to diagnose a post-surgical infection based on symptoms, such as a fever and abdominal pain. But symptoms often take several days to appear, often showing up after patients have been discharged from hospital and are recovering at home.

Dr. Pablo E. Serrano demonstrates using the FluidAI Origin device, which is attached to a staff member acting as a patient for this demonstration.

Dr. Pablo E. Serrano demonstrates using the FluidAI device.

According to FluidAI’s research, it takes 8.8 days on average from the onset of symptoms to diagnosis with the current standard of practice. During that time, a patient may undergo tests including blood work, cultures, x-rays and CT scans, and may also be readmitted to hospital. Such costs could be significantly reduced or avoided with early detection and intervention.

From theory to practice

Phase 1 of the trial was a fact-finding mission to determine how well the technology worked. It took place from 2020 to 2022 at hospitals including HHS’ Juravinski Hospital and Cancer Centre (JHCC), and found that the AI system could identify risk up to 80 per cent earlier than existing diagnostic methods and before the onset of symptoms.

“With this technology, surgeons can rely on chemical analysis of leaked fluid to predict an infection, and intervene early, rather than waiting for symptoms to develop,” says Serrano.

Phase 1 included 200 JHCC patients, of which 50 per cent had liver and pancreas resections, 30 per cent had colorectal resections and the remaining 20 per cent were a mix of small bowel and stomach resections.

Phase 2, which launched at the beginning of 2024 and will also last for two years, will involve another 200 HHS patient volunteers from JHCC and Hamilton General Hospital. This time, surgeons will make treatment decisions based on what this technology tells them about a patient’s risk of leakage. So far, more than 20 patients have volunteered for Phase 2.

“It’s exciting to now be using this technology to help patients,” says Serrano, adding, “This study has the potential to completely transform how we manage post-surgical complications. And as well as saving lives, early detection is also expected to save hospitals hundreds of thousands of dollars because fewer tests will be needed to diagnose and treat patients, and fewer patients will need to be readmitted to hospital.”