A team of South Korean researchers has developed artificial intelligence (AI) technology to support the initial consultation stage of psychiatric care. The tool is expected to ease the workload of medical staff who must review each patient's extensive history and symptoms.
The Korea Advanced Institute of Science and Technology (KAIST) said Monday that a research team led by Professor Lee Ui-jin of the School of Computing and Professor Lee Tak-yeon of the Department of Industrial Design, together with a team led by Professor Kim Eun-joo of the Department of Psychiatry at Gangnam Severance Hospital, jointly developed the large language model (LLM)-based psychiatric initial consultation support technology.
The researchers designed the model so that patients can first converse with the AI before meeting a doctor, allowing them to structure their symptoms and condition. The AI analyzes the patient's responses in real time by cross-referencing them with specialized medical knowledge in psychiatry, and adjusts the flow of conversation by posing appropriate follow-up questions.
A key feature is that the AI does not simply conduct mechanical question-and-answer exchanges but applies "actual counseling techniques." "Through expressions of empathy, restatements that reorganize the patient's words, and clarifications that address ambiguous content, the AI helps patients talk about their condition more comfortably," the research team explained.
Based on the collected dialogue, the AI then generates a clinical dashboard that presents symptoms and potential disorders at a glance, providing it to medical staff. This allows doctors to understand a patient's condition more systematically before they enter the consultation room, enabling them to focus more on in-depth counseling during the actual session.
In an experiment involving 1,440 virtual patients, the AI demonstrated strong performance, effectively securing the core clinical information needed for treatment within 30 minutes in most cases.
However, the research team noted clear limitations in the AI's ability to detect subtle emotional changes or handle sensitive topics. "AI cannot replace doctors, and final judgments must be made by skilled specialist medical staff," the team emphasized. The AI is intended to serve as a "smart assistant" handling repetitive and structured information gathering, while doctors make the final diagnoses and prescriptions based on that input, forming an efficient collaborative system.
"If AI reduces the burden of the initial consultation stage, medical staff can focus more deeply on patient counseling," Professor Lee Ui-jin said. "It shows the potential to evolve into a new mode of care in which humans and AI collaborate in clinical settings."
