[论文解读] Securing AI-based Healthcare Systems using Blockchain Technology: A State-of-the-Art Systematic Literature Review and Future Research Directions

Rucha Shinde, Shruti Patil+4 more|arXiv (Cornell University)|May 30, 2022

Blockchain Technology Applications and Security被引用 10

一句话总结

本文提出了一种集成区块链的框架，以增强基于人工智能的医疗系统（特别是在自然语言处理、计算机视觉和声学人工智能应用中）的安全性、隐私性和可信度。通过利用区块链实现数据溯源、可验证的模型训练以及抵御对抗性攻击，该研究识别出若干关键优势，如减少单点故障并提高透明度，尽管也指出当前研究仍处于早期阶段，面临诸多挑战。

ABSTRACT

Healthcare systems are increasingly incorporating Artificial Intelligence into their systems, but it is not a solution for all difficulties. AI's extraordinary potential is being held back by challenges such as a lack of medical datasets for training AI models, adversarial attacks, and a lack of trust due to its black box working style. We explored how blockchain technology can improve the reliability and trustworthiness of AI-based healthcare. This paper has conducted a Systematic Literature Review to explore the state-of-the-art research studies conducted in healthcare applications developed with different AI techniques and Blockchain Technology. This systematic literature review proceeds with three different paths as natural language processing-based healthcare systems, computer vision-based healthcare systems and acoustic AI-based healthcare systems. We found that 1) Defence techniques for adversarial attacks on AI are available for specific kind of attacks and even adversarial training is AI based technique which in further prone to different attacks. 2) Blockchain can address security and privacy issues in healthcare fraternity. 3) Medical data verification and user provenance can be enabled with Blockchain. 4) Blockchain can protect distributed learning on heterogeneous medical data. 5) The issues like single point of failure, non-transparency in healthcare systems can be resolved with Blockchain. Nevertheless, it has been identified that research is at the initial stage. As a result, we have synthesized a conceptual framework using Blockchain Technology for AI-based healthcare applications that considers the needs of each NLP, Computer Vision, and Acoustic AI application. A global solution for all sort of adversarial attacks on AI based healthcare. However, this technique has significant limits and challenges that need to be addressed in future studies.

研究动机与目标

• 调查区块链技术如何解决基于人工智能的医疗系统中的安全与隐私挑战。
• 分析现有将区块链与人工智能在三大关键医疗人工智能领域（自然语言处理、计算机视觉和声学人工智能）集成的研究。
• 识别当前在对抗性攻击防御和人工智能驱动医疗系统中数据完整性方面的方法局限性与研究空白。
• 提出一个面向自然语言处理、计算机视觉和声学人工智能应用独特需求的统一区块链解决方案的概念框架。
• 概述利用区块链保障医疗领域人工智能安全的未来研究方向，重点关注可扩展性、互操作性和鲁棒性。

提出的方法

• 在三大人工智能应用路径（基于自然语言处理、基于计算机视觉和基于声学人工智能）的医疗系统中，开展系统性文献综述。
• 评估现有对抗性攻击防御机制，包括对抗性训练，并评估其对进一步攻击的脆弱性。
• 分析区块链在通过不可篡改账本实现医疗数据溯源、访问控制和可审计性方面的作用。
• 评估区块链在保障异构医疗数据源之间联邦学习与分布式学习方面的潜力。
• 综合构建一个将区块链与医疗人工智能集成的概念框架，针对自然语言处理、计算机视觉和声学人工智能工作负载的独特需求进行定制。
• 识别技术与实际挑战，如性能开销、标准化差距以及对去中心化系统的信任问题。

实验结果

研究问题

• RQ1 区块链技术如何增强基于人工智能的医疗系统的安全性和可信度？
• RQ2 当前人工智能驱动医疗系统中对抗性攻击防御机制存在哪些局限性？区块链能否缓解这些问题？
• RQ3 区块链在医疗人工智能应用中如何支持数据溯源、完整性验证和访问控制？
• RQ4 区块链如何实现跨异构医疗数据集的安全透明分布式学习？
• RQ5 将区块链与人工智能在医疗领域集成的关键挑战和未来研究方向是什么？

主要发现

• 区块链技术通过实现不可篡改的审计日志和去中心化的访问控制，有效缓解了医疗领域中的安全与隐私问题。
• 基于区块链的账本可可靠地建立医疗数据溯源和用户身份验证，从而增强对人工智能输出结果的信任。
• 区块链支持在多个医疗机构之间实现安全、隐私保护的分布式学习，且数据来源异构。
• 区块链的集成减少了基于人工智能的医疗系统中的单点故障，并提升了系统透明度。
• 尽管具有上述优势，当前关于区块链与人工智能在医疗领域集成的研究仍处于早期阶段，面临显著的可扩展性、性能和标准化挑战。
• 对抗性防御技术仍易受新型攻击变体影响，目前尚无适用于所有类型对抗性攻击的通用解决方案。