HCPP: A Data-Oriented Framework to Preserve Privacy during Interactions with Healthcare Chatbot

Abstract

Healthcare chatbots are becoming increasingly popular, but with their use comes the issues of excessive personal information collection and privacy leakage. To address this issue, we propose a Healthcare Chatbot-based Privacy Preserving (HCPP) Framework that adopts a data-oriented approach to reduce the excessive disclosure of personal information. HCPP consists of two main components: the Healthcare Chatbot-based Minimized Personal Information (HCMPI) method and the Healthcare Chatbot-based Zero Knowledge Proof (HCZKP) method. HCMPI leverages large language models (LLMs) to minimize the acquisition of unnecessary personal health information without significantly affecting healthcare service. HCZKP further encrypts a part of the minimized information, making the data available but invisible. The experimental evaluation results demonstrate the effectiveness and feasibility of our approach.