Approaches to health data integration
Security concerns and vendor limitations are some of the top barriers to unlocking the full potential of health data in improving patient outcomes.
In the HIMSS24 APAC plenary session, ‘Power of Data in Modern Healthcare’, Cho Chi-Heum, professor and president of Keimyung University Dongsan Medical Center in South Korea, and Dr. Shih-Sheng Chang, director of the Artificial Intelligence Center of China Medical University Hospital (CMUH ) in Taiwan discussed challenges in data integration and interoperability.
To enable their physicians to share patient data securely, Keimyung Hospital has an ISM (Information Security Management) system, which is at the core of its data privacy and security strategy. “We also strictly follow the guidelines for the protection of personal information,” Prof. Cho added.
Common vendor agreements also hinder hospitals from generating valuable insights from comprehensive, integrated data. Some vendors allow data integrations from multiple sources, but this typically comes at a high cost due to proprietary data use rights, Dr. Chang noted.
“Sometimes they provide really nice software that has (all) the data, but we can’t pay for every single one of them.”
Dr. Chang and his team have learned to approach this problem by taking the next step: “If we want to set up a new machine or a new modality, we will ask the supplier, ‘Can you share the data with us? Otherwise, we have to share the data share with us.’ will consider another supplier.
Driving efficiency
The two speakers also discussed how their doctors have wasted time on tedious administrative tasks.
“Most doctors in Korea are always looking at their computers. They are so busy (with their computers) that they cannot (meet) their patients… I don’t want to see these kinds of hospitals,” Prof Cho said.
Nurses, meanwhile, spend 60% of their time on repetitive tasks, mainly checking patients’ vital signs. “This is very repetitive and (mechanical).”
According to Professor Cho Keimyung Hospital wants to implement wearable devices as part of remote patient monitoring so nurses can save time.
Meanwhile, as part of its data quality management – and by extension AI development, CMUH is also considering large language models (LLM) to build a system for structured data. Through LLMs, Dr. Chang said, physicians do not have to prepare structured clinical reports, which can take up to 30 minutes each.
However, Prof Cho warns against over-reliance on computer software, especially software that runs on AI. “We are humans. We are not AI; we just use AI.” Bee Keimyung Hospital, he envisions an “emotional hospital” that prioritizes meeting patients in person.
“AI cannot replace humans. But it can save time by letting us have (that) eye contact with our patients,” Dr. Chang said, echoing Prof. Cho’s statement.