April 30, 2024

• 

From left: Richard Clarke, Chief Analytics Officer at Highmark; Rob Birdsong, Vice President of Google Cloud Consulting West-Central Region; Hwang Hee, Kakao Healthcare CEO; and Shweta Maniar, Global Director for Google Cloud, pose for a photo at Google Cloud Next 2024, Las Vegas, Tuesday. (Kakao Healthcare)

Kakao Healthcare, the digital health care unit under platform giant Kakao, announced Thursday that the company would boost its collaboration with Google to improve its cloud-based data platform.

The announcement came after Kakao Healthcare CEO Hwang Hee shared the company's recent progress with Healthcare Data Research Suite (HRS), the company’s cloud-based data platform service, during Google's Cloud Next conference in Las Vegas this week.

According to Hwang, Korea has a significant amount of medical data available for research, but the absence of tools that allow combining and using the data is preventing hospitals and medical institutions from making the best out of the data. HRS has been introduced as a solution to solve this problem.

Kakao Healthcare said the company is currently in the process of expanding the distribution of HRS to local medical institutions that plan on collaborative research, including the members of R-Alliance, an alliance of Korea’s major medical institutions and hospitals to collaborate on AI-based research.

R-Alliance members include University Medical Center, Yonsei University Health Systems, Ewha Womans University Medical Center, Samsung Medical Center, Chonnam National University Hospital and Chosun University Hospital.

Kakao Healthcare said Korea University Anam Hospital, Keimyung University Dongsan Hospital and Ewha Womans University Seoul Hospital and Mokdong Hospital newly joined the alliance and conducted a federated learning test.

During his presentation at Cloud Next 2024, Hwang shared that the company has successfully completed a federated learning test using HRS with R-Alliance members.

Federated learning in AI refers to the practice of training AI models in multiple independent and decentralized training regimes. Kakao Healthcare explained that the company confirmed that HRS is fully capable of federated learning, meaning that it does not require a centralized dataset and a unified system of training, allowing better security of patients’ medical data.

Kakao Healthcare also noted that the company has recently started a project that will improve HRS to become able to recognize and store more handwritten medical records. For the project, Kakao Healthcare will use Google’s latest technologies, including LLM-based named entity recognition tech -- a natural language processing method that extracts information from text.

“Kakao Healthcare’s HRS is already utilizing Google’s language learning models, as well as the latter’s most recent technologies such as MedLM, Gemini, Gemma, Vertex AI. The company plans to increase its collaboration in the future to improve HRS,” an official from Kakao Healthcare said.

By Shim Woo-hyun (ws@heraldcorp.com)
The Korea Herald