While studying medicine in Sydney in 2015 with a view to specializing in cardiology, Aengus Tran did a two-month clinical placement in his native Vietnam that upended his plan. At the Tam Duc Cardiology Hospital in Ho Chi Minh City, the doctor-to-be witnessed first-hand how the short-staffed hospital struggled to provide timely diagnoses to its patients. For Tran, it was a moment of clarity that medicine “needs a system-wide change, one based on technology that can read scans and make diagnoses instantly…and is run so effectively and efficiently that it is available to everyone,” he says.
With a mission to drive that change, he cofounded Harrison.ai, a clinical AI service provider, in 2018 in his final year of medical school with his older brother, Dimitry, then-head of innovation at Australian private hospital giant Ramsay Health Care. Using the country’s mature healthcare market as a testing ground, the pair developed AI software capable of analyzing chest X-rays and chest and brain CT scans, flagging critical cases and generating draft reports. The company says its software can cut diagnostic report turnaround times by up to 40% in time-critical cases and boost overall reporting speeds by 10%.
As of May, Harrison.ai’s software was being used by more than 3,400 clinicians in over 1,000 hospitals and clinics in more than 40 countries. Its two biggest markets are Australia, where it says it serves half the country’s roughly 3,200 registered radiologists, and the U.K., where it assesses more than 40% of all chest X-rays for NHS England, part of the U.K.’s National Health Service. It also has a presence across Asia and in Europe.
Now Tran, 32, is targeting expansion in what he believes could become Harrison.ai’s biggest market, the U.S. “We’ve already built a very comprehensive AI that makes 455 different diagnoses and has been battle-tested,” says the 30 Under 30 Asia alum from 2020. “The U.S. is our next corridor.”
The startup’s $112 million Series C funding round in February last year at an undisclosed valuation brought its total amount raised to more than $240 million and will be used to bankroll the expansion. Investors in that round included pension fund Aware Super and fund manager ECP Asset Management, both Australian, and Hong Kong’s Horizons Ventures, a VC firm cofounded by billionaire Solina Chau and backed by the territory’s richest man, Li Ka-shing.
In May, Harrison.ai received another big-name backer when Hong Kong-based conglomerate Jardine Matheson, controlled by the billionaire Keswick family, agreed to acquire I-Med Radiology Network, which holds a minority stake in Harrison.ai, in a deal that puts I-Med’s enterprise value at $2.4 billion. The Australian diagnostic imaging provider, the country’s largest by number of clinics, bought its Harrison.ai stake in 2021.
Harrison.ai generates revenue from the annual subscription fee paid by healthcare providers to access its platform, which is based on the expected volume of scans. When a subscribing radiologist opens an X-ray or CT scan using their usual software, Harrison.ai’s findings appear as an overlay that also prioritizes urgent cases.
*Excluding angel investors and grants
Source: Harrison.ai, Crunchbase
In the year ended June 30, 2025, revenue from customer contracts jumped 54% from a year earlier to A$7.3 million ($5.1 million), according to a filing made by Harrison.ai to the Australian Securities and Investments Commission. It swung to a net loss of A$118 million in the same period from a year-earlier net profit of A$4.4 million. The company says the filing only provides “a partial picture” of its operations and declines to comment on the reason for the loss.
Unfazed, Tran is forging ahead in what he believes is a burgeoning market. The shortage of radiologists he witnessed in Vietnam is a universal problem as aging populations put a strain on healthcare systems across the world. In the U.S., there are about 11 radiologists per 100,000 people, according to the American College of Radiology, while the Medical Board of Australia says the country has roughly 12 for every 100,000 people. In the U.K., meanwhile, a 31% shortfall of clinical radiologists results in long waiting times for patients, according to the Royal College of Radiologists.
The shortage is just one factor driving demand for clinical AI services. San Francisco-based Grand View Research cites the increasing need for early and accurate diagnosis amid the mounting volume of imaging data being generated. The firm expects the industry to notch up a compound annual growth rate (CAGR) of 35% between 2026 and 2033 to top $20 billion, led by the U.S., which accounts for almost half the market and where 600 million medical imaging procedures are conducted yearly, according to Dublin-based Research and Markets.
The country is already Harrison.ai’s third-largest revenue contributor, but Tran is aiming to make it the biggest by next year. Harrison.ai entered the U.S. in 2022, with Dimitry relocating to Cambridge, Massachusetts, to oversee the expansion as managing director, Americas. Since then, it has landed some key clients, including the Mass General Brigham healthcare system in Boston; Radiology Associates of North Texas, one of the nation’s largest private radiology practices based on number of radiologists; and Inland Imaging, a medical imaging provider headquartered in Washington state.
“Other companies built AI that detects A, B or C while we have the A-to-Z detection system.”
But Harrison.ai is up against hefty competition, including Israel’s Aidoc and India’s Qure.ai. The former was set up in 2016 and has raised a total of $500 million from blue-chip American VC firms such as Nvidia’s NVentures and General Catalyst. It specializes in the diagnosis of X-rays and CT scans for life-threatening conditions such as strokes and hemorrhages and its software is also geared to provide clinicians with services such as mobile alerts about critical cases and a chat function between hospital departments.
Qure.ai, also founded in 2016, has raised a total of $126 million from investors including U.S. pharma giant Merck’s venture fund, and Novo Holdings, the controlling shareholder of Danish drugmaker Novo Nordisk. The company is best-known for assessing X-rays for tuberculosis and lung cancer detection, and its software notifies clinicians when a patient needs a follow-up scan.
Asked about his competitors, Tran says Harrison.ai’s edge is the larger number of findings it is capable of analyzing. Its chest X-ray tool can identify up to 124 findings within 20 seconds, and that has led to a 45% increase in diagnostic accuracy, according to the company. Its CT brain software can detect as many as 130 in less than two minutes while its CT chest software can spot over 200. By comparison, Aidoc is trained to detect about 30 acute findings while Qure.ai is capable of roughly 100.
“If you think of AI as a spell checker for radiologists, preventing them from missing critical things, other companies built AI that detects A, B or C while we have the A-to-Z detection system,” Tran says. Rhidian Bramley, a consultant radiologist at the Christie NHS Foundation Trust in Manchester, says Harrison.ai’s chest X-ray software has boosted its lung cancer detection rates and accelerated the process, with 90% of chest X-rays of suspected cases being reported the same day, up from 20% earlier. “Harrison is the leader in terms of the number of findings in its chest X-ray solution,” he says.
Sources: Harrison.ai, Aidoc, Qure.ai
That said, the U.S. market is a challenge as it is among the most stringent for clinical AI companies, which are required to secure approval from the U.S. Food & Drug Administration for each discrete finding their software is capable of identifying. As Prashant Warier, cofounder and CEO of Qure.ai says, “Having a lot of findings that you can detect on a scan is an algorithmic capability, but to have a product that is FDA cleared, it takes a lot of time.”
Qure.ai and Aidoc each have roughly twice as many findings approved by the FDA as Harrison.ai. However, Warier says it’s not a winner-takes-all market. “If you look at AI in radiology, there are probably thousands, if not hundreds of thousands, of problems, and everybody is solving their own piece,” he says.
To better compete, Harrison.ai in November started to include third-party tools on its platform such as those of Singapore’s Us2.ai, which specializes in cardiac ultrasounds, and South Korea’s Lunit for cancer diagnostics. Harrison.ai doesn’t charge a markup or commission for these third-party offerings because, as Tran explains, “To us, the benefit is we have a wider portfolio of offerings for customers.”
The company is also moving beyond “decision support and detection into radiologist reasoning,” says Tran, with new foundational models that parse X-rays like a human radiologist, taking into account the patient’s clinical history. Called Harrison.rad and using a chatbot interface, the model is still in the testing phase and presently available only to researchers and select customers. “We don’t want to just release it into the wild. We want to do it in a responsible and thoughtful way,” Tran notes.
Aidoc also has third parties on its platform and offers a reasoning engine that provides multiple findings simultaneously and prioritizes critical cases. Its chief business officer, Tom Valent, says Aidoc sets itself apart by acting like an operating system that brings “imaging AI, clinical workflows, and enterprise analytics into a single coherent platform.”
But Chris Liu, investor at Horizons Ventures, insists Harrison.ai’s extensive offering is a key differentiator. Companies developing AI tools that only flag specific diseases limit workflow automation because physicians still have to review every scan to rule out other conditions, Liu says. Harrison.ai, he adds, is positioned to become “the standard of care for early detection diagnostics” in the next five years.
“No one else offers anything near the level of comprehensiveness that Harrison.ai provides,” concurs Samantha Wong, a partner at Australian VC firm Blackbird Ventures, an early investor. “The Tran brothers, one a medical doctor and AI engineer, the other a healthcare executive…had the right mix of skills to do exactly what was needed.”
Harrison.ai’s chest X-ray tool shows findings as an overlay.
Harrison.ai
Tran and his brother were born in Vietnam to two high school math teachers. Growing up with a programming-enthusiast father, Tran learned coding early. He moved to Australia at age 16 for high school, joining Dimitry, who was already working in Sydney. It was a massive culture shock, he recalls, and although he had learned English at school in Vietnam he found the Australian accent challenging. Pushing himself to make friends beyond the Vietnamese community helped him to assimilate, he says.
While Trans says he is “a programmer at heart,” when it came to his career he chose to study medicine at the University of New South Wales. Following his placement in Vietnam, he started looking for ways to combine his two interests. While attending a talk by an executive from Virtus Health, one of Australia’s biggest assisted reproductive service providers, he learnt how difficult it was to select the most viable fertilized egg for implantation.
That sparked a business idea and he created Ivy, an AI tool in collaboration with Virtus that analyzes videos of embryos in test tubes to identify the one most likely to result in pregnancy. Ivy has increased the rate of successful implantation to 93% from 65% when human embryologists make the selection, according to the Australian Alliance for Artificial Intelligence in Healthcare, a network of specialists. “That was the very first time I glimpsed the vision of what would become Harrison.ai,” recalls Tran who sees his entrepreneurial efforts as “a different way of practicing medicine.”
In 2019, Virtus sold the rights to Ivy’s technology to Swedish IVF firm Vitrolife for an initial payment of $6 million with an additional $2 million to be paid based on product-development targets over a three-year period. The same year, Tran raised $20 million for Harrison.ai in a Series A funding round that included Horizons Ventures and Ramsay Health Care, Dimitry’s former employer.
“The Tran brothers, one a medical doctor and AI engineer, the other a healthcare executive…had the right mix of skills to do exactly what was needed.”
At first, the brothers focused on creating just a few diagnostic tools, but soon realized it would be more lucrative to offer a broader array. To speed things along, they formed a joint venture in late 2019 with I-Med to set up the first software platform, Annalise.ai. This helped address one of the biggest challenges facing AI diagnostic startups: access to a large set of scans required to train AI models.
Tran also bought data from third-party providers, whom he declines to name, and went on to gain access to a vast multiethnic dataset in 2023 when NHS England signed on as a client. An Asian dataset became available the same year when Harrison.ai was incorporated in all emergency rooms at Hong Kong’s public hospitals.
But the founders soon encountered an issue that looms over AI diagnostic firms hungry for data to train their models: patient data protection. The scans provided by I-Med to Annalise.ai had been stripped of all obvious personal identifiers, but in 2024 a local media report accused I-Med of providing images without patient consent to Annalise.ai between 2020 and 2022. The Office of the Australian Information Commissioner began inquiries with a view to opening an investigation but last August said it was satisfied the scans had been adequately de-identified and I-Med had not breached Australia’s Privacy Act.
Harrison.ai bought the shares it didn’t already own in Annalise.ai from I-Med in 2024 and has been gradually rebranding the platform Harrison.ai Radiology. Asked about the controversy, Tran says: “We deeply believe in doing the right thing by the patient, including treating the datasets that we use with utmost care and respect, and meeting all the laws and regulations.”
Elsewhere, Tran says one of his biggest challenges has been figuring out how to lead a rapidly growing startup that went from just himself as the sole programmer and Dimitry managing business operations to about 200 employees now. To develop Harrison.ai into a global healthcare system, he knew he had to build “a diverse, multidisciplinary team” to tackle everything from data engineering to distribution and regulatory clearances. His experience of adjusting to life in Australia as a teenager has helped him manage Harrison.ai’s global staff, he discloses.
“Sometimes people ask if I gave up on medicine,” Tran says. “I did not. All the things I learned in medicine and all the experience I had, I apply every day.”
Check out our full Forbes 30 Under 30 Asia 2026 list here.
ForbesMeet The Forbes 30 Under 30 Asia Class Of 2026Forbes30 Under 30 Asia 2026: Meet The Entrepreneurs Building AI For Individuals And EnterprisesBy Yue WangForbes30 Under 30 Asia: The Healthtech Founders And Researchers Pushing Boundaries Of Scientific DiscoveryBy Yue WangForbesForbes 30 Under 30 Asia: The Social Entrepreneurs Tackling Issues From Food Waste To InequalityBy Zinnia LeeForbesMeet The Forbes 30 Under 30 Asia Class Of 2025By Rana Wehbe Watson
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。