Browse by author
Lookup NU author(s): Dr Kheng-Lim GohORCiD, Dr Anurag Sharma
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
Key messages:• Overall principles: In biomedical research studies involving AI algorithms, it is crucial to ensure informedconsent, anonymity, diversity, feedback control, collabora:on with experts and stakeholders, regulatory compliance, responsible data sharing, transparency, and addressing systema:c biases and influencing factors in AI algorithms.• Patient/participant safety: Biomedical researchers and AI developers have an ethical responsibility to prioritize patient safety by rigorously evalua:ng and valida:ng AI models, following best practices, addressing biases, engaging in collabora:ve research, and con:nuously monitoring and improving the models.• AI algorithm development stage: In developing AI algorithms for biomedical research, collabora:on between biomedical researchers, clinicians and computer engineers is vital, along with diverse perspec:ves, inclusive data collec:on, and appropriate preprocessing techniques during AI training.• Data de-risking: After obtaining informed consent, prior to executing the AI algorithms, biomedical researchers should perform data de-risking through anonymization methods while adhering to data governance laws, ethical guidelines, and privacy protection, while also considering individuals' rights to access and update their data.• Data custodians have the responsibility to safeguard data for use with AI algorithms, comply with regulations, maintain governance and documentation, ensure ethical alignment, transparency, and accountability through communication, assessments, and audits.• Biomedical data management for use with AI algorithms should adhere to data privacy, security, ethical laws, data governance frameworks, standardiza:on approaches, and data sharing agreements. Also, adopting existing big data practices, such as data cataloging, metadata management, integra:on, API management, data security, monitoring, and disaster recovery-when employing AI algorithms to run the data, is recommended for effective data utilization and protec:on.• AI deployment: Data privacy, security, accessibility, and good prac:ces should be followed when employing AI algorithms; good prac:ces are data minimization, training researchers, regular backups, designated data protection roles, and security audits.• Sharing the benefits: When sharing the benefits of biomedical research with par:cipants whose data is used, clear and timely communication of results, analysis, and insights, co-creation of knowledge, and participant engagement are crucial aspects to consider.
Author(s): Goh KL, Cheong VS, Tan KCK, Sharma A
Series Editor(s): The Bioethics Advisory Committee (BAC)
Publication type: Report
Publication status: Published
Series Title: REACH Bytes E-Newsletter
Type: Feedback Paper to Public Consultation Paper
Year: 2023
Print publication date: 30/06/2023
Online publication date: 30/06/2023
Acceptance date: 30/06/2023
Institution: Newcastle University in Singapore
Place Published: Singapore
ePrints DOI: 10.57711/pgd8-ma17
Notes: Report was submitted to BAC Singapore in response to the call on PUBLIC CONSULTATION ON ETHICAL, LEGAL AND SOCIAL ISSUES ARISING FROM BIG DATA AND ARTIFICIAL INTELLIGENCE USE IN HUMAN BIOMEDICAL RESEARCH, see https://www.reach.gov.sg/Participate/Public-Consultation/Ministry-of-Health/Bioethics-Advisory-Committee/public-consultation-on-big-data-and-artificial-intelligence-in-human-biomedical-research Report is made externally visible