::: 前往中央內容區塊
:::

News

Research institute tackles Taiwan’s healthcare problems with AI and exoskeletons 2025/02/15

TAIPEI (Taiwan News) — Industrial Technology Research Institute is harnessing AI and technology to lessen the burdens of Taiwan’s graying population.
 

According to a 2019 UN report, the percentage of those aged 65 in Taiwan will increase from 15% in 2019 to 35% in 2025. Furthermore, the National Development Council said Taiwan is set to become a super-aged society this year.

Political economist Chiang Min-hua said Taiwan’s low birthrate partially stems from government policies following World War II, which encouraged families to have only two children. Additionally, the higher cost of living and stagnant wages have discouraged Taiwanese from having children.

As Taiwan becomes a super-aged society, healthcare becomes more important. Taiwan News sat down with ITRI on Friday to discuss its latest technologies — featured at CES in Las Vegas last month — and how they bolster Taiwan’s wellness.

MedBobi

Free Bionic's HipBO. (Taiwan News, Michael Nakhiengchanh photo)
Free Bionic's HipBO. (Taiwan News, Michael Nakhiengchanh photo)

Edward Hsu (許凱程), Chief Medical Officer of Biomedical Technology and Device Research Laboratories at ITRI, said that language barriers can make patient-doctor discussions difficult and slow, taking up to 30 minutes per appointment.

Enter MedBobi, ITRI’s translation and diagnosis system. Powered by AI, it cuts down appointments to just five minutes.

Hsu explained that MedBobi is two systems. The first is voice recognition, which transcribes the user’s speech in real-time and creates a recording.

“In Taiwan, doctors often use Chinese mixed with English names for diseases, which proves challenging for traditional speech recognition software,” Hsu said. MedBobi accurately detects these nuances with a 6% error rate.

The second system then translates the patient-doctor dialogue into the target language. The AI-powered system uses large language models from Google and Meta. Large language models (LLMs) are deep learning models trained on massive amounts of data.

MedBobi was trained on data from Far Eastern Memorial Hospital, Kaohsiung Veterans General Hospital, and National Taiwan University Hospital, where it has been deployed. The platform is cloud-powered, meaning the system can be accessed on a phone or tablet with an internet connection.

Aside from Chinese and English, MedBobi can understand over 90 languages including Japanese and Indonesian. The system also scans dialogue for keywords including symptoms and diseases and automatically fills out medical forms, further speeding up appointments.

Hsu said he hopes that the technology can improve care quality for patients while reducing workload for healthcare workers.

KneeBO and HipBO

Healthcare is not just about finding symptoms and treating patients. Those with mobility disabilities can now find ways to move again with KneeBO and HipBO.

Free Bionics Taiwan founder and CEO Wu Cheng-hua (巫震華), who worked for ITRI from 2002 to 2016, said the two devices combine robotics, AI, and healthcare. He now runs the robotics spin-off Free Bionics Taiwan.

By using an algorithm, the exoskeletons offer real-time feedback, adjusting users’ movements more naturally. They do not use external sensors and instead work by reading movements from the user’s motion

Source