Korean tech companies are developing AI-powered tools for the medical field. These tools aim to analyze patient data and recommend treatments. Some experts believe AI could potentially reduce the workload of healthcare professionals.
On February 20, IT insiders revealed that Naver Cloud is reportedly developing a specialized large language model (LLM) tailored for the medical field. This innovative tool aims to assist in diagnosing and making treatment decisions, although its launch date remains undisclosed.
The model in development by Naver Cloud is believed to be trained on medical data, enhancing the capabilities of the existing Hyper Clova X platform. It resembles Google’s Med-PaLM, another large language model designed around medical data to support healthcare decisions.
Google’s Med-PaLM passed the U.S. Medical Licensing Examination (USMLE) with an 80% accuracy rate, surpassing the average passing score of U.S. medical students and interns (60%).
Before unveiling its medical specialized model, Naver Cloud revealed a medical service incorporating Hyper Clova X.
Medical institutions can use Clova Studio, an AI development tool based on Hyper Clova X, to create desired AI services.
For example, the institutions can create a service that provides medical consultation or suggests appropriate treatment. It can also extract the key information needed from medical papers.
Kakao Brain, an AI subsidiary of Kakao (035720), also recently unveiled KARA-CXR, a chest X-ray interpretation auxiliary.
Researchers at Inha University Hospital analyzed 2000 chest images using KARA-CXR. They worked with a 68-70% accuracy rate, outperforming OpenAI’s GPT-4 by 21-30 percentage points.
Following the KARA-CXR launch, Kakao Brain released a Labeler Project for clinical image annotation on GitHub. This open-source tool aims to expand the medical AI ecosystem by facilitating collaboration on medical image interpretation tasks.
The format of medical reports varies slightly from doctor to doctor. Developing AI technology using unstructured data is challenging.
Kakao Brain’s Labeler Project can be used to interpret medical reports written in different formats accurately.
The company believes that using this technology to develop AI technology for X-ray interpretation will enable various performance verification studies.
There are also expectations that domestic and international AI technology companies will quickly enter the medical field and be able to replace some medical personnel.
The Bank of Korea’s “AI and Labor Market Change” report flags potential disruption for general and Eastern medicine doctors, placing them among the 1% most exposed to AI in the workforce. This highlights the possibility of AI automating specific tasks these doctors currently perform, necessitating future adjustments.
An IT expert predicts extensive AI adoption in the medical field, stating, “Given the reliance on objective judgment and historical data, the healthcare industry presents a prime opportunity for AI implementation.” They foresee an influx of AI companies entering the medical space, expecting a significant impact.
eom@news1.kr