HKU's CardiOmicScore: AI Blood Test Claims to Flag Six Cardiovascular Diseases Up to 15 Years Before Symptoms

How does HKU's CardiOmicScore AI blood test work, and how strong is its predictive claim?

According to a report by TechNews, HKU's Li Ka Shing Faculty of Medicine has built an AI system called CardiOmicScore that requires only a single blood draw to assess a person's risk of developing six major cardiovascular diseases (E1). The report quotes the study's core claim directly: "民眾只需透過一次抽血，就能評估未來罹患六種主要心血管疾病的風險，最快可在臨床症狀出現前 15 年發出警訊" — meaning the system can, at best, issue a warning as early as 15 years before clinical symptoms emerge (E1).

The underlying data comes from the UK Biobank, a large-scale population dataset. Per TechNews, the research team analyzed 2,920 circulating proteins and 168 metabolites found in blood samples from this cohort to build the predictive model (E3). The project is led by Chang Ching-peng (張清鵬), Associate Professor in HKU's Department of Pharmacology and Pharmacy, who is quoted as saying the team hopes the technology can shift health management "from passive disease treatment to proactive prediction and intervention" (E4).

What is the scientific evidence behind this claim, and where has it been published?

According to TechNews, the research underpinning CardiOmicScore has been formally published in the international journal Nature Communications (E2). This is the sole piece of evidence available on the study's publication status — the report does not provide additional detail such as sample size for validation cohorts, peer-review timeline, or specific accuracy/sensitivity metrics beyond the 15-year lead-time claim and the protein/metabolite counts already cited (E2, E3).

Where does CardiOmicScore stand in terms of development and commercial rollout?

Despite the headline claim, TechNews reports that the technology remains at the research stage, and the team has not announced a commercialization timeline (E5). The direct quote from the report states: "該技術目前仍處於研究階段，尚未公布明確的商業化時間表" (E5). This places CardiOmicScore in an early phase — there is no evidence in the report of clinical trials, regulatory filings, or a product launch date.

Why does early cardiovascular prediction matter, and how large is Taiwan's disease burden?

Separate reporting on cardiovascular disease in Taiwan, compiled by UDN Health, provides context for why early-warning tools are being pursued. Wei Zheng (魏崢), Director of the Cardiovascular Center at Cheng Hsin General Hospital, is quoted stating that 15% of patients with coronary artery blockage never experience chest pain, and recommends that people over 40 arrange for exercise electrocardiogram testing (E6).

Government mortality data cited in the same reporting shows the pace of cardiac-related deaths accelerating over time: according to Taiwan's Ministry of Health and Welfare, in 2010 (民國99年) one person died of heart disease every 33 minutes and 32 seconds; by the following year, that interval had shortened to 31 minutes and 50 seconds (E7).

How many people in Taiwan have hypertension, and how has the diagnostic threshold changed?

The same UDN Health reporting notes a regulatory shift with direct implications for patient counts. When the American Heart Association lowered its hypertension threshold in November (of the cited year) to 130/80 mmHg, Taiwan reportedly gained more than one million newly classified hypertension patients overnight (E8).

More broadly, cardiology sources at MacKay Memorial Hospital cited in the report state that Taiwan has more than 4 million hypertension patients in total, of whom nearly 500,000 are classified as having treatment-resistant hypertension (E9).

What this means

Taken together, the evidence describes two separate but thematically linked developments. HKU's CardiOmicScore, published in Nature Communications and built on UK Biobank analysis of 2,920 proteins and 168 metabolites, claims a 15-year predictive window for six cardiovascular diseases (E1, E2, E3) — but by the research team's own admission, it has no commercialization timeline and remains a research-stage tool (E5). Meanwhile, separate reporting on Taiwan's existing cardiovascular burden shows why such a tool could be relevant: 15% of coronary blockage patients show no symptoms (E6), the country's cardiac death clock has shortened from 33 minutes 32 seconds to 31 minutes 50 seconds year-over-year (E7), and lowering the hypertension threshold to 130/80 mmHg added over a million patients to Taiwan's rolls in a single policy change (E8), on top of an existing base of more than 4 million hypertension patients, nearly 500,000 of whom are treatment-resistant (E9). The gap between HKU's early-stage research claim and Taiwan's already-documented disease burden is not bridged by any evidence in this pack — no report here confirms CardiOmicScore has been tested on, or is intended for, the Taiwan population specifically.