TO PREDICT A HIDDEN KILLER: Wilmington startup strives to improve global health standards

Recently, the New York Times published an article about a young college student named Sam Terblanche. On Saturday Sept. 16, 2023, he was on his way to a soccer game with some friends, when he told them that he was not feeling well. The next day he went to the emergency room with a headache and chills, and again on Monday. Both times, the ER discharged him—telling him to get some rest and take some Advil.

On Thursday, Sam's parents were notified by his university that he had died from sepsis.

This story is uncomfortably close to that of Rory Staunton, who in 2012 received a scrape on his arm during a basketball game. Later his parents took him first to his pediatrician, then his local ER, who discharged him. By the time he returned to the ER, it was too late. Sepsis had set in.

Less than three days after Rory scraped his arm, his parents were shopping for a coffin. This incident is also the catalyst for Rory's Regulations in New York.

The most horrifying part of these stories? There are tens of millions more, just like them. And that's what Predicate AI Labs of Wilmington, NC aims to curb—everywhere. 

What is Predicate AI Labs? 

Predicate AI Labs is the health technology company behind OpenDX, a biomarker prediction platform that specializes in calculating a probability score for the presence of sepsis using machine learning and artificial intelligence. This new tool is intended to integrate with standard procedures in order to avoid increasing burden or burnout to medical practitioners, and most importantly, assist in ensuring patients an accurate and timely diagnosis.

And it hasn't gone unnoticed—the most highly regarded hospital in the world, the Mayo Clinic, has partnered with Predicate to bring the technology all over the globe, to every socioeconomic class. It was also named a "Startup to Know" by Hypopotamus in September, one of 15 top Healthcare Discovery companies and startups in October 2025 by F65, and just received a Core Innovation award from the North Carolina Tech Association. Additionally, the Greater Wilmington Business Journal named Predicate's founder and CEO, Dr. Morris Nguyen, the 2024 Coastal Entrepreneur of the Year in May.

Nguyen says that Predicate's goal is to increase the standard of health worldwide. "We need to rebalance the system."

And he has a good start. Predicate is currently tested on five continents and in seven countries, with hubs in Dublin, Ireland, and Santiago, Chile. 

But why sepsis?

The Hidden Killer

Predicate began as a consultancy group with the intention of fusing its data with technology. Nguyen, a seasoned critical care doctor, teamed up with other physicians to improve standard processes to fix America's broken healthcare system. They went into medical facilities and asked them what their biggest pain points were in relation to quality benchmarks—which are usually measured by mortality, readmission, and cost. They all had the same answer: Sepsis. 

On average, 1 in 3 patients who die in an American hospital had sepsis during their hospitalization. Globally, there are around 50 million reported cases of sepsis per year and annually kills more people than all cancers combined. In addition, it is regularly the leading cause of death worldwide, and affects the global south, particularly sub-Saharan Africa, the worst. In the US, that death toll is approximately 350,000 per year.

"The reason why sepsis is such a problem is because it's a hidden killer. It's one of the hardest conditions to catch because the warning signs can be very subtle. Things like fatigue, fever, or just feeling 'off.' It may not look urgent until it's too late." Nguyen explains. 

"There's also the risk of medical gaslighting, which disproportionately affects women and minorities. They're told they're fine, until they deteriorate hours later." 

According to Nguyen, sometimes the symptoms just don't match the textbook examples, either, which makes it easier for misdiagnosis. On top of that, alternative sepsis surveillance products are designed more as a billing system than a diagnosis engine, and often are plagued with fragmented data, tedious workflows, and distracting pop-ups.

Another factor is that America's emergency rooms are increasingly under pressure, stretching medical teams to their limits. According to CDC data from 2018 to 2024 (exempting 2020 due to lack of data), the average number of Emergency Room visits in the US per year is approximately 146 million, a number that is expected to rise as the current US administration plans to further cut healthcare and defund the Affordable Care Act. Further, when a doctor or nurse is on the 20th or 24th hour of their shift, they are more likely to miss important details, or default to an implicit bias based on the major predispositions that any patient may have. 

"The funny thing is, when you hear hoof beats, you think it's horses—not zebras," Nguyen says, "But with sepsis, it's always zebras."

Find the Zebras

To find these sepsis "zebras," Predicate's OpenDX operates through a fairly simple workflow. There are two primary sources of input: biosensor monitoring technology, and text and vocal speech patterns (which we will get to shortly).

PREDICTABLE PATH: A high-level overview of OpenDX's workflow. Photo courtesy of predicatelabs.ai

Predicate's software scrubs a patient's personal smart wearable, like a Fitbit, or a hospital designated device to monitor their vitals and biomarkers. OpenDX continuously synchronizes this information with a timestamp and iterates (how machines learn and develop patterns in data) it in the patient's record. It monitors the patient's biomarkers for any obvious or subtle indications of early onset sepsis, and then calculates a probability score from 1-100 of the presence of sepsis for the physician to review. OpenDX then creates a detailed diagnostic of what inputs determined the score and how it calculated it. 

"I'm actually more excited about ruling out sepsis than finding it," Nguyen says.

Nguyen points out that OpenDX has no intention of trying to replace any health care worker, and it doesn't diagnose patients. It's designed as a tool to help an overburdened system. 

"Doctors need to trust and accept the fact that technology is here. AI is already being taught in medical schools today. When I give talks on AI, I describe AI as 'augmented intelligence,' not 'artificial intelligence.' It's a clinical decision support system designed to augment clinical decision making, not replace it."

Ever the empathetic physician, Nguyen continues, "I think it gives patients a peace of mind as well, accepting the fact that the doctors are human, they're not flawless, and that mistakes can happen. Our system has a built in AI safety layer. There’s always a human in the loop."

Giving the Afflicted a Voice

Then there's OpenDx's perhaps most novel feature. 

"We're building a movement to promote the idea that symptoms deserve a voice, and voice is a vital sign. It should be standard in every single ER and built into every telemedicine platform." Nguyen says. 

And Nguyen means that, literally. OpenDX uses a Natural Language Processor (NLP) to provide patients a text and speech interface to record their symptoms through their own app on their phone or on a diagnostic device at a medical facility. 

Patient self-advocacy is a notorious issue, as some people don't speak up at the doctor's office for a multitude of reasons. Nguyen offers his father as an example—the classic strong and stoic type that takes a lot of motivation to see a doctor. In other reports, people sometimes feel like they are challenging an authority, or don't feel confident in their own knowledge about their health to speak up, or don't feel like they have enough time to speak with their doctor, so they prioritize their most pressing issue. 

"There's some countries and cultures where it's stigmatized to report feeling sick, especially with women," Nguyen affirms. "So we're giving them a voice. We treat what patients are saying about their own symptoms as real data, not just background noise."

But how does OpenDX record vocal and written data as a biomarker? For text, it's just typing information into a textbox field, a place where the patient can confidently—and privately—type in their symptoms in their own words as they're happening, something they may not feel empowered to do in person, or even remember to do by the time their appointment comes around. 

Additionally, OpenDX analyzes people's speech patterns. Predicate collects this data and uses it to build their own unique biolingual library that time-synchronizes with its data inputs all over the world. As much as turning speech patterns into biomarkers sounds like science-fiction, it's already being used to detect Alzheimer's cases. 

"As that library grows, we want to give access to researchers to better understand critical health conditions and innovators to build technologies from it. Promoting health equity is a big part of what we're doing," Nguyen says.

In their initial tests, Predicate analyzed words patients used at the Hospital Escuela in Tegucigalpa, Honduras with a focus group of 270 patients. The results were encouraging, to say the least. 

They correctly predicted 98% true sepsis out of the 100 patients that developed the condition. Meaning that OpenDx was able to predict sepsis at nearly a 100% efficiency based solely on people's own words.

"And that was an early model," Nguyen points out. 

Nguyen has made it clear that OpenDx is not about replacing medical professionals, or adding to their already strained load. But how does Predicate plan to integrate OpenDX into the complex web of medical processes that change not only country-to-country, but hospital-to-hospital? 

According to Nguyen, it starts by making it accessible. OpenDX is sensor agnostic and can work with any sensor, regardless of the brand. So anyone who already has an Apple watch or Fitbit need only download the app and consent to the license agreements. The vocal processor is available on the app regardless of whether someone has a wearable device or not. For people who can't afford these wearables, it's compatible with devices $100 or less that a medical team or facility can supply. In essence, the first step in health equity is equal access. 

As far as integration, Nguyen sees a three-point approach, namely in Emergency Room and Hospital Triage, Remote Patient Monitoring kits, and Concierge Medicine.

Why Wilmington?

"Wilmington, for me, has the Goldilocks effect, not too big, not too small, just right. Wilmington has been ranked as the number one startup ecosystem in the world for cities under 300,000 as an emerging tech hub," says Nguyen.

"It's an ecosystem where we all have each other's back because it's just small enough. It's not cutthroat competitive like in Raleigh or in you know, the bigger hubs like Silicon Valley or Boston, and Austin."

Nguyen says that the tech startups in the area are tasked with recruiting talent to the area, to keep the students coming out of the local schools, and attract the career ML and AI engineers, the data scientists, the medical tech roles. For that, the startup ecosystem has to work together.

And it seems that this sense of noble responsibility is deeply ingrained in Dr. Nguyen. It's not a desire, but the bare minimum to help or solve a problem without causing an issue elsewhere. He has the ability to see any system as a working or deteriorated central nervous system, as a whole that needs treatment for its parts.

And it makes sense—Morris Nguyen began his medical journey with a passion for science, specifically physiology. But he also always wanted to fix things, especially when it came to them being sick. 

"It was personally meaningful when I took the [Hippocratic] Oath to 'First Do No Harm.' In fact, that same oath is what we still live through every single day at Predicate," Nguyen says.

Dr. Morris Nguyen will speak at the East coast Healthcare Summit at Hotel Ballast, 301 N Water St, Wilmington, NC 28401, on October 28.