A federated learning system and method between medical institutions and a disease prognosis prediction system including the same

• Main Authors: LEE JAE DONG, LEE YE JI, CHA HYO SOUNG, PARK HYUN WOO, BACK KYOUNG YEON, KIM YU MIN
• Format: Patent
• Language: English; Korean
• Published: 07.05.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC COMMUNICATION TECHNIQUE; ELECTRICITY; HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATIONTECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING ORPROCESSING OF MEDICAL OR HEALTHCARE DATA; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; PHYSICS; TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHICCOMMUNICATION

본 발명은 의료기관 간의 연합 학습 시스템 및 방법, 이를 포함하는 질환 예후 예측시스템을 제안한다. 본 발명의 연합 학습 시스템 및 방법은, 의료 데이터를 이용한 연합 학습 시에 계층적 군집화 학습 방법을 적용하고, 또 기계 학습 결과에 따라 생성된 가중치 정보를 양자 암호 및 타임 스탬프 코드 암호화 방법을 적용하여 전송하도록 구성된다. 따라서 의료기관 간의 데이터 이질성을 해소할 수 있어 학습 모델의 성능 향상을 기대되고, 의료 데이터의 개인정보 보호를 보호할 수 있어 안정성이 보장되는 이점이 있다. The present invention relates to a system and method for federated learning among medical institutions, and disease prognosis system including same. According to the system and method for federated learning of the present invention, a learning method of hierarchical clustering is applied to conduct federated learning in which medical data is used, and weights information generated according to results of machine learning is transmitted by applying methods of time stamp code encryption and quantum cryptography. Therefore, the present invention is expected to improve the performance of a learning model by solving the issue of data heterogeneity between medical institutions and has the advantage of ensuring stability by protecting personal information in medical data.