One in 50 people has an unreported brain aneurysm, which can be life-threatening if it ruptures. Early detection can save lives, yet brain imaging scans remain costly. As a result, 99% of patients are unable to access appropriate tests and continue living with undetected aneurysms.
TALOS Corp., a Korean AI company, has developed ANRISK, a non-generative AI model using 8.8 million person-years worth of data that is both safer and more affordable than other commonly used examinations. Following the company’s participation at GITEX Digi Health & Biotech in Bangkok, CEO and leading neurosurgeon Tackeun Kim, M.D., explains how his team of neurosurgery experts is helping to prevent the preventable.
Can you introduce ANRISK, and why you chose to focus on brain aneurysms over other neurological conditions?
TK: ANRISK is an AI-driven platform that predicts the risk of developing a brain aneurysm using only routine health checkup data—such as blood tests, anthropometric measurements, and lifestyle questionnaires—before any imaging is performed. This allows for low-cost, large-scale screening and enables clinicians to identify high-risk individuals early, preventing life-threatening ruptures through timely intervention.
We chose to focus on brain aneurysms because the disease is often silent until rupture, at which point mortality and disability rates are extremely high. Yet, when detected early, it is highly preventable through appropriate treatment. The challenge is that diagnosis today still relies heavily on expensive imaging modalities such as CT or MRI, making broad population screening impractical.
ANRISK was designed to bridge this critical gap. My personal experience also played a decisive role. I worked for over ten years as a neurosurgeon, specialising in cerebrovascular surgery. The most painful moments in my clinical career were when patients with subarachnoid hemorrhage arrived in the emergency room, and despite the best surgical and medical efforts, we still lost them. Many of these patients could have been saved if their aneurysms had been detected earlier, before rupture.
A neurosurgeon, even by sacrificing nights and weekends over a lifetime, might treat ten thousand patients at most. But my dream was to extend that impact to millions. I came to believe that the key was not individual practice, but building a system. ANRISK represents the first step toward realising that vision.
How accurate is ANRISK compared to standard brain imaging scans?
TK: That’s an important distinction. Brain imaging, such as CT or MRI, is a diagnostic test, and in that context, diagnostic accuracy is paramount. Imaging typically achieves an AUROC of about 0.90, which is appropriate since the goal is definitive diagnosis. ANRISK, however, is not a diagnostic tool—it operates in the screening domain, where the key question is whether the test provides clinically meaningful stratification of risk.
In this context, an AUROC of 0.77 is highly significant. For example, in our data the prevalence of aneurysms differs by nearly forty-fold between the lowest-risk and highest-risk groups defined by ANRISK. That means the tool effectively separates populations into groups where clinical decision-making can be very different.
By comparison, many widely used genetic-based cancer risk prediction tools have AUROCs in the 0.5–0.6 range, yet they are already considered valuable in guiding preventive strategies. Against that benchmark, ANRISK demonstrates a clear performance advantage as a practical and impactful screening solution.
Brain aneurysms often have no symptoms. What are the main risk factors, and who should consider sharing their health data for assessment? Are certain groups more vulnerable?
TK: Brain aneurysms are often silent until rupture, so understanding risk factors is critical. While studies across different regions show some variation, several factors are consistently reported. Women are about 1.5 to 2 times more likely than men to develop aneurysms. Smoking is another well-established and powerful risk factor.
Age also plays a role, but interestingly, women tend to develop aneurysms at a somewhat younger age compared to men. Taken together, we see middle-aged women, particularly those who smoke, as an especially important target group for screening and preventive assessment. Encouraging individuals in this demographic to share their health data can help identify those at elevated risk before any symptoms appear.
Would office workers, who are often under high stress, benefit from using ANRISK?
TK: There is limited direct evidence that stress itself increases the likelihood of developing or rupturing a brain aneurysm. However, from a broader medical perspective, stress is known to raise blood pressure variability, and this fluctuation places additional strain on the blood vessels.
Such hemodynamic stress is a recognised risk factor not only for aneurysms but for cardiovascular and cerebrovascular diseases in general. So while we cannot yet say that stress alone is a proven cause, individuals under chronic stress—such as many office workers—may still benefit from using ANRISK as part of a preventive health strategy, especially if they also have other established risk factors like smoking, hypertension, or family history.
What happens after someone receives their risk report? If flagged as high risk, are they referred for imaging scans?
TK: ANRISK is designed purely as a screening tool, not a diagnostic one. When someone receives a high-risk result, the next logical step is to consider imaging, such as MRI, for confirmation. To give perspective, among individuals classified in the highest-risk group, about 6.44% are expected to already have an aneurysm.
In practical terms, if 1,000 people in this group undergo MRI, roughly 64 will be diagnosed with a brain aneurysm. Given the severity of aneurysms and the high mortality risk if rupture occurs, this prevalence is more than sufficient to justify follow-up imaging.
For comparison, routine endoscopy for stomach cancer—one of the most common screening practices—has a cancer detection rate of only about 0.8%. By contrast, the prevalence in ANRISK’s highest-risk group is nearly eight times higher. Even the broader high-risk group shows a prevalence of around 4%, roughly double that of the general population, again supporting the case for further imaging in these individuals.
Can you share a success story or patient case where early detection made a real difference?
TK: There are many cases we could point to, but as you know, sharing individual medical records raises privacy concerns. What I can share is the real-world experience from one of the hospitals actively using ANRISK. In their practice, they strongly recommended MRA scans for individuals identified in the highest-risk group.
The outcomes were remarkable: over 20% of these patients were confirmed to have brain aneurysms, and when including other cerebrovascular conditions, the proportion with abnormal findings reached around 30%. This detection rate was substantially higher than what we originally anticipated when designing ANRISK.
More importantly, it meant that these patients received the opportunity for early diagnosis and timely treatment, turning a potentially life-threatening silent condition into one that could be managed and prevented.
How is Talos supporting preventive healthcare? What practical steps can people take to lower their risk of aneurysms?
TK: At TALOS Corp., our mission is to support preventive healthcare by shifting the focus from late-stage treatment to early detection and risk stratification. With ANRISK, we provide hospitals, health centers, and eventually individuals with a scalable, cost-effective tool to identify those at higher risk before symptoms appear. By integrating routine health data into predictive algorithms, we help clinicians prioritise whoshould receive imaging or closer monitoring, ensuring that preventive care becomes actionable rather than reactive.
When it comes to lowering the risk of aneurysms, ANRISK goes beyond generic advice. Our platform provides personalised, data-driven recommendations tailored to each individual’s profile, rather than offering one-size-fits-all guidance. For example, the reports we generate highlight specific lifestyle adjustments—such as controlling blood pressure, quitting smoking, or modifying certain habits—based on the person’s actual health data.
If you look at some of the reports we’ve shared, you can clearly see how these recommendations are both practical and realistic for everyday life. In this way, TALOS Corp. delivers not just a risk score but a roadmap for prevention, empowering both clinicians and individuals to take meaningful action.

Could people potentially take their health measurements at home and send them securely by email or another channel, since the data required for analysis appears to be quite limited?
TK: Yes, that is exactly the kind of need we anticipated. To address it, we developed a dedicated application that allows individual users to try the analysis even with a minimal set of health data. The app provides a secure channel for entering information and receiving personalised results, making it accessible outside of the hospital setting.
You can explore it at: https://www.taloscorp.io/anrisk-app/.
Can you update us on your international partnerships, especially with academic institutions in the US and Japan?
TK: ANRISK has been making steady progress internationally. In Japan, we have already received product classification from the PMDA as a non-medical device, which enables us to accelerate commercialisation. We are also active in academic circles—for example, through the Japanese Society for Artificial Intelligence in Medicine—and we maintain close collaborations with leading professors at Nagoya University and Hiroshima University.
You can see more about our academic engagement here: LinkedIn update. In the United States, our Deputy CEO, Dr. Jindeok Joo, will begin a visiting professorship at Stanford University this October. This represents an important step in building our academic and clinical network in the US, and we expect it will open doors to further collaborations with both researchers and healthcare institutions.
What are your plans for the Middle East — are you looking at Saudi Arabia as well as the UAE and Qatar?
TK: We see the Middle East as a strategically important growth market, and Saudi Arabia, the UAE, and Qatar are all countries we are actively considering for expansion. While we have not yet launched a full-scale initiative there, our intention is clear: the region represents a tremendous opportunity for preventive healthcare solutions like ANRISK.
Earlier this summer, we had a promising meeting with LEAN Business Services, facilitated through my former institution, Seoul National University Bundang Hospital (SNUBH). We believe collaborations like this could become a strong entry point into the Middle East healthcare ecosystem, and we are prepared to move quickly should the right partnership align. Establishing a foothold in the region would not only accelerate our global expansion but also allow TALOS Corp. to contribute meaningfully to preventive healthcare in one of the world’s most dynamic markets.
Can you explain VesselRisk? Does it use the same patient data inputs as ANRISK?
TK: VesselRisk is best understood as a horizontal expansion of our core technology, built on the same large-scale dataset that powers ANRISK. While ANRISK focuses specifically on brain aneurysms, VesselRisk extends to other critical vascular conditions such as angina, acute myocardial infarction, other ischemic heart diseases, intracerebral hemorrhage, non-traumatic intracranial hemorrhage, cerebral infarction, and cerebral artery stenosis.
These are all conditions associated with sudden death and severe disability. However, there is a key difference in how we approach these diseases compared to aneurysms. For brain aneurysms, a high predicted risk can lead directly to definitive diagnostic imaging and, if confirmed, to curative treatment options such as surgery or endovascular procedures. This makes risk prediction highly actionable.
For many of the other vascular diseases, however, even if we can predict high risk with good accuracy, the pathway from prediction to action is less straightforward. For example, if someone has a high predicted risk of cerebral infarction, imaging will not reveal a lesion unless the stroke has already occurred, and preventive drug therapy—while beneficial—does not offer the same curative potential as aneurysm treatment.
In short, VesselRisk uses the same type of patient data inputs as ANRISK, but the clinical utility differs. ANRISK was our first focus precisely because it links prediction with clear, preventive, and often curative interventions. VesselRisk represents the next step, expanding our predictive capabilities to a broader set of vascular conditions while carefully considering their real-world clinical impact.
Do you have plans to expand into other digital health solutions beyond brain and cardiovascular conditions?
TK: For us, expanding into new areas of digital health is not simply about diversification. We believe expansion only makes sense when there is both a clear socio-economic impact to be made and sufficient scientific and clinical expertise to support it.
Personally, I do not see business as an end in itself—it is a means to realise the kind of world I envision. That is why my colleagues and I want to develop and deliver technologies that we can explain with solid evidence, and that truly help humanity. At this point, our focus remains firmly on cerebrovascular and cardiovascular health, where we already have deep expertise and where the potential to prevent catastrophic events is extremely high.
So, while we are not currently aiming to move beyond these domains, our commitment is to make sure that in brain and cardiovascular disease, we develop world-class, evidence-based technologies that can make a real difference on a global scale.
Discover more from HealthTechAsia
Subscribe to get the latest posts sent to your email.