A recent study in South Korea explored using homomorphic encryption (HE) to protect patient data in training AI models. The study gathered EMR data from three hospitals to predict mortality rates after surgery. The findings suggest HE can be applied to combine data from various sources while maintaining privacy and security. This encryption method allows small hospitals to develop AI models using data from larger hospitals. HE has applications in finance, IT, and healthcare to enhance data privacy. The Korean government supports AI research in medical institutions by facilitating collaboration and access to datasets, aiming to improve healthcare outcomes.
Source link
New way to protect patient data for training AI models
