Cloud-based hospital information system 'PHIS' and medical-specialized large language model (LLM) fusion

Aizen Sciences and Huniverse Global announced on the 23rd that they have signed a memorandum of understanding (MOU) for research cooperation to develop and distribute a large language model (LLM) for hospital work assistance. This agreement aims to provide advanced medical services by applying Aizen Sciences' medical-specialized LLM technology to Huniverse's cloud-based hospital information system (PHIS).

https://seenthis.kr/newspage/2338⁠⁠⁠⁠⁠⁠⁠

Huniverse's cloud-based hospital information system PHIS is the first system in Korea to be introduced to a general hospital. It started with the introduction of the cloud-based hospital information system to three general hospitals (An Am, Guro, and Ansan Hospitals) of the Korea University Medical Center, the first case of a cloud-based hospital information system introduced to a general hospital with more than 1,000 beds worldwide. PHIS has all the core functions of a large hospital computer system and can meet different requirements of each hospital with its modular structure.

PHIS is a system that enhances the efficiency of hospital operations and drives digital innovation based on cloud technology. It establishes a PHIS clinical terminology standard system that meets international standards, providing flexibility and scalability for application to various medical institutions.

This research collaboration is expected to significantly enhance the efficiency of patient care and medical administration within hospitals by applying Aizen Sciences' medical-specialized LLM to PHIS.

Aizen Sciences' 'AIGEN InSight' is a small-scale LLM that minimizes costs by significantly reducing the number of parameters and improves accuracy through fine-tuning. This model consists of 7 billion parameters, making it light enough to be operated with a single computer. It was the first to overcome the US Medical Licensing Examination, a barrier that other models of its class had not been able to overcome. AIGEN InSight achieved a high score of 74, comfortably passing the average passing score of 60, compared to MediTron-7B, the existing highest performing 7 billion parameter model, which received a score of 52.

AIGEN InSight has multi-step reasoning capabilities required to solve complex medical problems and has shown performance that is 13% higher on average than OpenAI's GPT-3.5 (175 billion parameters) model, which is significantly larger, in seven medical benchmark performance evaluations.

Through this collaboration, the fusion of PHIS and medical-specialized LLM is expected to create an innovative hospital work assistance AI that supports clinical decision-making, organizes non-standardized medical charts, and assists medical staff in areas such as patient counseling and scheduling, as well as medical administrative tasks.

"The integration of Huniverse's PHIS and Aizen's medical-specialized LLM technology is expected to significantly improve the efficiency and quality of medical services," said Jae-Woo Kang, CEO of Aizen Sciences. "Through this, hospitals can implement more accurate and efficient healthcare services as well as cost-reduction effects, while patients can be provided with faster and more courteous new user experiences."

"AIGEN InSight from Aizen Sciences is a small but highly performing language model that solves the problems of existing large language models, such as computing power and electricity consumption, and is expected to be used in repetitive tasks and information creation tasks at the hospital," said Sang-Heon Lee, CEO of Huniverse Global. "AIGEN InSight and PHIS working together will synergize to reduce the burden on medical staff and improve the quality of patient care through accurate diagnosis support, leading to healthcare innovation that is felt by the people."

Huniverse Global Introduction

Huniverse Global developed PHIS, a cloud-based precision medicine hospital information system targeting general hospitals, for the first time in the world through the national strategic project 'Precision Medicine Hospital Information System Development Project' in 2017. This has evolved into a big data platform for patient-tailored precision medicine. PHIS developed by Huniverse Global has been developed in 38 standard modules for all information produced at hospitals, including outpatient and inpatient treatment, administrative affairs, and electronic medical record system (EMR), and is applied to medical institutions of various sizes. PHIS was introduced to three general hospitals of the Korea University Medical Center (An Am, Guro, and Ansan Hospitals) in 2021, Asan/Cheonan Chungmu Hospital in 2022, Seoul Samik Hospital and Donggang Hospital in 2023, and is currently being expanded to Wonju Severance Christian Hospital in 2024. The PHIS clinical terminology/standard system, which complies with international standards, continuously generates multi-level, multi-institutional standardized medical big data, with the goal of enhancing the domestic medical big data ecosystem to a world-class level.

Aizen Sciences Introduction

Aizen Sciences is an artificial intelligence (AI)-based new drug development company established on April 13, 2021. CEO Jae-Woo Kang founded the company based on original technology developed over the past seven years while leading the DMIS research team at Korea University's Department of Computer Science. In this period, they won seven times in various international biomedical AI competitions, beating global leading AI research institutions such as Google, Stanford, MIT, Janssen, NVIDIA, and Siemens. Aizen Sciences is currently conducting 14 in-house drug development projects, including cancer and rare/refractory disease treatments. The company aims to grow by creating value through technology transfer (license-out) of self-developed drug candidate substances and securing revenue (profit sharing) through commercialization of co-development pipelines. This will play a crucial role in opening up the path for AI- driven business innovation throughout the pharmaceutical and medical industries.

Huniverse Global: https://huniverse.co.kr/

Website: https://www.aigensciences.com/

Contact Information
Aizen Sciences
Jae-Woo Kang, CEO
02-6952-3530