Researchers from National Taiwan University Hospital (NTUH), Academia Sinica and the Lithuanian University of Health Sciences have developed an artificial intelligence (AI) platform for the early detection of pancreatic cancer.
In a paper published this month in Nature Communications, the researchers detailed how they developed a machine learning tool to identify pancreatic ductal adenocarcinoma (PDAC) biomarkers using “metabolomics, the study of metabolites and lipids within biological systems.”
The researchers’ tabular foundation model-based algorithm, which they call PanMETAI, “offers a rapid, accurate and noninvasive tool for early PDAC detection,” the paper says.
Photo: Hou Chia-yu, Taipei Times
The model achieved an area under the curve value of 0.99 in the Taiwanese training and validation cohort and 0.93 in the external Lithuanian validation cohort, demonstrating high accuracy, reproducibility and applicability across different ethnicities.
Area under the curve is a metric of diagnostic accuracy in medical research.
Most people are already in the late stage of pancreatic cancer by the time it is diagnosed, the study said, adding that PDAC has a five-year survival rate of 13 percent.
NTUH doctor Chang Yu-ting (張毓廷), a contributing author to the paper, said that pancreatic cancers are called “the king of cancers” because they are deep inside the body, are often asymptomatic in the early stages and are also difficult to diagnose through scans.
The researchers achieved a stable framework for the early detection of pancreatic cancer, Chang said.
PanMETAI has great applicability and could be used for diagnosing other cancers, assessing treatment efficacy and analyzing prognosis, becoming foundational for the early diagnosis of other diseases, the study said.
In other news, NTUH and Taipei Veterans General Hospital (TVGH) were featured in Newsweek magazine’s list of the world’s 250 best hospitals, released on Wednesday.
TVGH was ranked 174th, up 34 places from 208th last year, while NTUH returned to the rankings at No. 249 after a three-year absence, the list showed.
The World’s Best Hospitals rankings, first published by Newsweek and Statista in March 2019, includes data on more than 2,500 hospitals across 32 countries, including 35 in Taiwan.
The rankings were based on hospital quality metrics, making up 40 percent of the score; hospital recommendations from surveyed medical experts (30 percent); patient experience surveys (18.5 percent); and surveys on how hospitals implemented and used patient experience surveys (6.5 percent), the article said.
TVGH works with institutions in the US, Japan and South Korea, which help raise its international visibility and possibly boost its world ranking, hospital superintendent Chen Wei-ming (陳威明) said.
Newsweek’s rankings tend to favor hospitals in English-speaking countries, but “many more” Taiwanese hospitals would have made the list in a more neutral evaluation, Chen said.
The survey also provides rankings for hospitals in each country and region. The top five in Taiwan were TVGH, NTUH, Taichung Veterans General Hospital, Linkou Chang Gung Memorial Hospital and Kaohsiung Chang Gung Memorial Hospital.
The top 10 was rounded out by Taichung’s China Medical University Hospital, National Cheng Kung University Hospital, Kaohsiung Medical University Chung-Ho Memorial Hospital, Cathay General Hospital and Hualien Tzu Chi Hospital.
South Korea has adjusted its electronic arrival card system to no longer list Taiwan as a part of China, a move that the Ministry of Foreign Affairs said would help facilitate exchanges between the two sides. South Korea previously listed “Taiwan” as “Taiwan (China)” in the drop-down menus of its online arrival card system, where people had to fill out where they came from and their next destination. The ministry had requested South Korea make a revision and said it would change South Korea’s name on Taiwan’s online immigration system from “Republic of Korea” to “Korea (South),” should the issue not be
Tainan, Taipei and New Taipei City recorded the highest fines nationwide for illegal accommodations in the first quarter of this year, with fines issued in the three cities each exceeding NT$7 million (US$220,639), Tourism Administration data showed. Among them, Taipei had the highest number of illegal short-term rental units, with 410. There were 3,280 legally registered hotels nationwide in the first quarter, down by 14 properties, or 0.43 percent, from a year earlier, likely indicating operators exiting the market, the agency said. However, the number of unregistered properties rose to 1,174, including 314 illegal hotels and 860 illegal short-term rental
ECONOMIC COERCION: Such actions are often inconsistently applied, sometimes resumed, and sometimes just halted, the Presidential Office spokeswoman said The government backs healthy and orderly cross-strait exchanges, but such arrangements should not be made with political conditions attached and never be used as leverage for political maneuvering or partisan agendas, Presidential Office spokeswoman Karen Kuo (郭雅慧) said yesterday. Kuo made the remarks after China earlier in the day announced 10 new “incentive measures” for Taiwan, following a landmark meeting between Chinese President Xi Jinping (習近平) and Chinese Nationalist Party (KMT) Chairwoman Cheng Li-wun (鄭麗文) in Beijing on Friday. The measures, unveiled by China’s Xinhua news agency, include plans to resume individual travel by residents of Shanghai and China’s Fujian
Chinese President Xi Jinping (習近平) and Chinese Nationalist Party (KMT) Chairwoman Cheng Li-wun (鄭麗文) met in Beijing yesterday, where they vowed to bring people on both sides of the Taiwan Strait closer to facilitate the “great rejuvenation of the Chinese nation.” The meeting was held in the East Hall of the Great Hall of the People, a venue typically reserved for meetings between Xi and foreign heads of state. In public remarks prior to a closed-door meeting, Xi, in his role as head of the Chinese Communist Party (CCP), said that Taiwan is historically part of China, and remains an “inalienable” and
RSS