Harmonizing AI governance regulations and neuroinformatics: perspectives on privacy and data sharing
In the rapidly evolving field of neuroinformatics, the intersection of artificial intelligence (AI) and neuroscience presents both unprecedented opportunities and formidable ethical challenges (Ienca and Ignatiadis 2020;Dubois et al. 2023;Parellada et al. 2023;Scheinost et al. 2023). As AI technolog...
Uloženo v:
| Vydáno v: | Frontiers in neuroinformatics Ročník 18; s. 1472653 |
|---|---|
| Hlavní autoři: | , , , , , , |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
Switzerland
Frontiers Research Foundation
17.12.2024
Frontiers Media S.A |
| Témata: | |
| ISSN: | 1662-5196, 1662-5196 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Shrnutí: | In the rapidly evolving field of neuroinformatics, the intersection of artificial intelligence (AI) and neuroscience presents both unprecedented opportunities and formidable ethical challenges (Ienca and Ignatiadis 2020;Dubois et al. 2023;Parellada et al. 2023;Scheinost et al. 2023). As AI technologies increasingly underpin neuroscientific research, it is crucial to establish robust governance frameworks that not only match the ambitious scope of this research but also adhere to stringent requirements for privacy and data sharing (Eke et al. 2022;Jwa and Martinez-Martin 2024;Yuste 2023;UK Government 2018). This paper explores the urgent need to harmonize AI governance regulations with neuroinformatics practices, with a specific focus on the domains of data sharing and privacy.This opinion paper is grounded in a comprehensive analysis of over 4000 research articles and AI regulation documents, supplemented by referencing over 100 pivotal articles and documents. It offers a critical examination of current AI governance frameworks and the existing challenges at the intersection of AI and neuroinformaticsfoot_0 . Through this analysis, we systematically explore the stateof-the-art in neuroinformatics (Section 2), its challenges (Section 3), and the evaluation of AI governance (Section 4), identifying key alignments and gaps (Section 5). We conclude with strategic recommendations for better integration of these fields, aimed at enhancing research outcomes while ensuring privacy and fostering ethical practices (Section 6).By integrating these diverse perspectives, the paper aims to spark a constructive dialogue among policymakers, researchers, and practitioners. The objective is to develop a cohesive framework that not only supports innovation in neuroinformatics but also operates under the umbrella of conscientious and effective AI governance, ensuring that neuroinformatics can continue its rapid advancement in a responsible and ethically sound manner.Neuroinformatics has experienced transformative advancements through enhanced data sharing frameworks and technological innovations (Daidone et al. 2024;Weiner, Veitch, Aisen, Beckett, Cairns, Cedarbaum, Donohue, et al. 2015;MacGillivray et al. 2018;Cao et al. 2023). These developments have significantly improved research efficiency and fostered innovation, particularly in complex areas such as autism (Parellada et al. 2023;Zucchini et al. 2023;Saponaro et al. 2022) and Alzheimer's disease (Yao et al. 2023;Y. Zhang et al. 2022;Dubois et al. 2023).One of the most notable advancements in neuroinformatics is the standardization of data sharing practices (J. Wang et al. 2023;Alzheimer Europe 2021). Initiatives such as the Alzheimer's Disease Neuroimaging Initiative (ADNI) (Weiner, Veitch, Aisen, Beckett, Cairns, Cedarbaum, Donohue, et al. 2015;Weiner, Veitch, Aisen, Beckett, Cairns, Cedarbaum, Green, et al. 2015) and the Common Data Element (CDE) Project in epilepsy research (Loring et al. 2011) exemplify how standardized practices, including shared ontologies, common data elements, and standardized data formats, facilitate robust validation of results across diverse studies and enable large-scale, multi-centre studies (L. Wang et al. 2023;MacGillivray et al. 2018;Yaseen et al. 2023). These elements are fundamental for integrating data from various sources, evident in the success of these projects (Ojo et al. 2020;Viejo et al. 2023). This integration is vital for the scalability and reproducibility of neuroinformatics research, leading to more reliable outcomes and faster scientific progress (Gurari et al. 2015;Baker et al. 2015;Sarwate et al. 2014).Technological enhancements such as electronic health records and sophisticated data repositories have revolutionized how data is collected, managed, and shared within the field (Gentili et al. 2021;Leoratto et al. 2023). These technologies are crucial for supporting longitudinal studies and comprehensive data analyses necessary for understanding long-term outcomes of neurological conditions including traumatic brain injury (Vallmuur et al. 2023;Yaseen et al. 2023). Moreover, the role of international collaborations cannot be overstated. Initiatives such as the Dominantly Inherited Alzheimer Network (DIAN) (Bateman et al. 2012) and global epilepsy research consortia (Galanopoulou et al. 2021;Mishra et al. 2022) highlight the importance of pooling resources and expertise to tackle complex scientific questions, significantly enhancing the scope and impact of research efforts (Chou et al. 2022). Privacy-preserving technologies including differential privacy, encryption, anonymization, and blockchain have become integral to maintaining data confidentiality, while enabling expansive research and clinical applications (Z. Zhang, Xu, and Xiao 2023;Yuste 2023;C. Yang, Yuan, and Feng 2023;Patel, Provenzano, and Loew 2023). Notably, federated learning and edge computing have gained attention for their role in supporting decentralized research models while ensuring privacy (Zou et al. 2023;Y. Yang et al. 2024;Mitrovska et al. 2024). These technologies enable researchers to collaborate without compromising the security of sensitive data, crucial in neuroinformatics where privacy concerns are paramount (Gong et al. 2022;Selfridge et al. 2023;Cali et al. 2023).The landscape of neuroinformatics is fraught with complex challenges that stem from the integration of advanced data sharing, privacy, and security considerations (White, Blok, and Calhoun 2022;Sarwate et al. 2014). These challenges are crucial to address as they directly impact the efficacy and ethical alignment of neuroinformatics research (Ienca and Ignatiadis 2020).Resistance to data sharing remains a primary obstacle, often fuelled by concerns over data ownership and the potential for misuse (Tudosiu et al. 2022). This resistance necessitates clear policies that balance intellectual property rights with the need for open access to data (Redolfi et al. 2023).Additionally, the traditional academic reward system, which prioritizes individual achievements over collaborative efforts, further discourages open data sharing (Versalovic et al. 2023). Technical challenges such as managing and standardizing large, complex datasets add another layer of difficulty. Data heterogeneity, varying formats, and the necessity for robust metadata standards complicate data integration and utilization across various research platforms, making it challenging to achieve consistent and reliable research outcomes (L. Advancing neuroinformatics also requires substantial resources and infrastructure, including secure data repositories, high-performance computing facilities, and efficient data-sharing platforms, which support large-scale initiatives and sophisticated data analysis (Zhu et al. 2023;Yu et al. 2023;Viejo et al. 2023). These resources enable not only cutting-edge research but also the implementation of technologies including blockchain and federated learning, which demand considerable computational power (Xia et al. 2023;Tozzi et al. 2023;Ay, Ekinci, and Garip 2024;C. Yang, Yuan, and Feng 2023). The significant investment and logistical challenges associated with these technologies often limit their widespread adoption, impacting the field's ability to ensure data privacy and manage large datasets effectively (Li et al. 2020). (Roberts et al. 2021). Organizations such as OECD (OECD 2024b; 2024a) and UNESCO (UNESCO 2023) set global standards for ethical AI practices, advocating for human rights, transparency, and international cooperation, which aim to bridge regional differences and foster a unified approach to AI governance.The integration of neuroinformatics within global AI governance frameworks reveals a robust alignment, especially in privacy and data protection (J. Wang et al. 2023;Tozzi et al. 2023). Initiatives such as the ADNI (Weiner, Veitch, Aisen, Beckett, Cairns, Cedarbaum, Donohue, et al. 2015;Weiner, Veitch, Aisen, Beckett, Cairns, Cedarbaum, Green, et al. 2015) and the CDE Project in epilepsy research (Loring et al. 2011) demonstrate compliance with international privacy regulations such as the GDPR (European Union 2016; Alzheimer Europe 2021; White, Blok, and Calhoun 2022;Muchagata et al. 2020). These efforts underscore a commitment to safeguarding sensitive health data and adhering to high ethical standards (Alzheimer Europe 2021). Ethical considerations in neuroinformatics strongly resonate with the principles outlined in frameworks such as the EU's AI Act (Stahl and Leach 2023). Neuroinformatics practices, particularly in handling data related to genetic research and brain-computer interfaces (BCIs), strive to align with these governance frameworks, ensuring informed consent (Bannier et al. 2021) and cognitive liberty (Schiliro et al. 2023) as central to their operations (Kulynych 2002;Ligthart and Meynen 2023;de Hemptinne and Posthuma 2023).Despite these alignments, significant gaps persist, particularly in data standardization and interoperability (Daidone et al. 2024;J. Wang et al. 2023). The lack of unified data formats and protocols across international borders complicates efforts in global neuroinformatics collaborations (Zuk et al. 2020;Mulugeta et al. 2018). For instance, the variability in data management practices hinders the ability to maintain consistent transparency and accountability, making it challenging to comply fully with AI governance regulations across jurisdictions (Cheung et al. 2023;Yi et al. 2020) (Kharat et al. 2014;Higuchi 2013). However, neuroinformatics often struggles with the practical implementation of these technologies due to inconsistent regulatory support and the nascent state of these technologies in practical, researchfocused environments (Zhu et al. 2023;Yu et al. 2023). The need for interdisciplinary collaboration is highlighted by the complex ethical, legal, and technical challenges in neuroinformatics (Farah 2005;Blinowska and Durka 20 |
|---|---|
| Bibliografie: | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 content type line 23 Edited by: Mike Hawrylycz, Allen Institute for Brain Science, United States Reviewed by: Alexander Leichtle, University Hospital of Bern, Switzerland |
| ISSN: | 1662-5196 1662-5196 |
| DOI: | 10.3389/fninf.2024.1472653 |