Workshop for Security, Privacy, and Ethics in Health and Biomedical Research
Wearables and Connected Devices [Group Lead: Andy Coravos, Elektra Labs]
The rapid pace of miniaturization and expanding capabilities has led to an explosion of health-related devices, some designed for use in a medical setting and others primarily for personal use. The speed and novelty of these technologies has created a number of security, privacy, and ethical knowledge gaps that require attention.
Considering the cyber-physical nature of the system, security and its impact on patient safety is a significant concern due to the potential vulnerability of connected medical devices. Privacy concerns, beginning with knowing who owns personal data, is a significant challenge, which is tightly coupled to the ethics of the need for transparent consent to privacy agreements and data authorizations. This becomes especially critical when these technologies are utilized for care delivery. Furthermore, challenges such as outdated firmware in legacy devices, standardization, and decommissioning of medical devices need to be examined.
This breakout group will be tasked with identifying the core issues underlying security, privacy, and ethics in the context Wearables and Connected devices in the health care and health research settings, and then laying out a vision for a roadmap for developing potential solutions.
Big Data Analytics and AI / ML [Group Lead: Ling Liu, Georgia Tech]
Artificial Intelligence (AI) and Machine Learning (ML) algorithms fueled by big data are taking on an ever-increasing role in healthcare in providing clinical decision support to even independent diagnostics. Furthermore, biomedical and healthcare research itself has become much more reliant on AI/ML tools and technologies.
However, the use of automated and potentially autonomous systems brings forth numerous issues related to transparency, replicability, ethics, and effectiveness. Additionally, it is important to ensure that such machine learning systems and frameworks are trustworthy and accountable. For example, adversarial examples can significantly reduce the robustness of such solutions, and the use of transfer learning and advanced machine learning techniques, makes them less explainable.
This breakout group will be tasked with identifying the core issues underlying security, privacy, and ethics in the context of AI/ML in Health and Biomedical Informatics in the health care and health research settings, and then laying out a vision for a roadmap for developing potential solutions.
Real-time Multi-modal Data [Group Lead: Santosh Kumar, University of Memphis]
Multi-modal health-related data, including medical care data (clinical trials, X-Rays, MRI scans, videos, etc.) collected by a potentially large number of healthcare providers, daily life data (socioeconomic and employment status, environment, etc…) that inform social determinants of health (SDOH), and personal data through wearables or other monitoring devices can all potentially be utilized to identifying challenges to achieving and maintaining optimal wellness and help devise personalized medical treatment. Protecting and properly curating such data – over its life-time – for supporting real-time data-driven decision-making through high powered analytics/AI is critical. This can generate significant security, privacy and ethical issues that need to be understood within various healthcare as well as social contexts (e.g., disease, age/gender, social stigma, employment issues, etc.). The multi-modality of data as well as the criticality of real-time intervention pose significant domain-specific challenges in this context. More importantly, there may be challenges due to the integration of the data, as opposed to looking at it in isolation, especially given data quality issues. Life-cycle issues need to be considered because of the persistence of users’ medical/health information. Sharing of multimedia information is critical for realizing the personal health and well-being vision and supporting public health research. Significant security and privacy challenges arise in allowing such sharing and it is not clear how existing techniques (e.g., Secure Multiparty Computation, Differential Privacy) can be effectively utilized and what alternative techniques would be more appropriate. Ensuring privacy, accountability, availability and information authenticity is critical in such contexts.
This breakout group will be tasked with identifying the core issues underlying security, privacy, and ethics in the context of multimodal data acquisition, integration, and sharing, and also laying out a vision for a roadmap for developing potential solutions.