The UK’s House of Lords covid-19 committee is looking into how the rapidly increasing reliance on digital technology—accelerated by the pandemic—may have a long-term impact on our social and economic wellbeing. [1] Social media use increased substantially during the covid-19 pandemic. [2] During this period of social isolation/distancing, social media platforms have enabled vulnerable and isolated people to come together for mutual support. [3,4] Nevertheless, the stress of the pandemic and the urgency to seek help, advice, and support, could potentially breach social norms, where what is revealed online is far more than would be revealed pre-pandemic, or in face-to-face discussions. This phenomenon is called the online disinhibition effect whereby some people self-disclose or act out more frequently or intensely than they would in person. [5]

The proliferation of social media use during the pandemic offers a way to observe information about the pandemic directly from the population and could give researchers real time health data on populations. The information on social media platforms has become a potential dataset. [6] Automated technologies such as AI (Artificial Intelligence) based web scraping or screen scraping bots can be used to data-mine or extract data from social media platforms without user or platform consent. This information could be analysed by AI technologies using Natural Language Processing (NLP) technologies to process the text, and convert it to data which can be researched/modelled. [7] There is growing concern that this technological innovation is being used to circumvent the requirement to obtain informed consent, especially in healthcare research where information is collected from social media. [8] A further consideration is that sensitive personal information may be identified either through a single piece of information or through multiple bits of information gathered through different sources. 

This in turn raises considerable ethical considerations in this area. Support groups and networks may feel inhibited to offer help and support to each other if they realise that they may become unconsented research subjects, thereby handicapping a very useful support forum. Equally, not seeking informed consent for research in social media platforms, could be seen as an objectification and exploitation of vulnerable participants—commodification in this way could be regarded as disrespectful for patients. [9] Looking back at the history of medical ethics, such as the Tuskergee syphilis experiment, where study participants were not consented or made aware that they were participants in an experiment, is a grim reminder of the need to always seek informed consent from human participants in medical research. [10] Researchers should not be dazzled by the irresistible draw of using new technologies without ethical reflection. 

We acknowledge that there is a distinction in the UK (but not necessarily elsewhere in the world), between unconsented social media and internet-mediated research with technology, versus the audit of national electronic records for quality improvements. The latter takes place under principles of transparency, accountability, due process, and procedural fairness. The NHS Health Research Authority “Is my study research” tool can be helpful in determining whether a study is research or audit. [11] Obtaining informed consent in social media medical research should not be circumvented. The ease of availability of AI technology; the lack of adequate regulation in this area; the lack of laws which penalise these types of infringements; and the difficulty in identifying infringements, have contributed to the potentially unethical practice. [12] However, case law such as Patel v Facebook [2019], Copland v UK [2007] and 77m Ltd v Ordnance Survey Ltd [2019] shows increased judicial involvement in the area of consent and personal information on the internet. [13]

The development rate of AI based technologies and computer power has enabled new ways in which technology may interact with humans, and will likely develop into areas we have yet not considered. While that makes the task of regulation difficult, the absence of regulations and guidance on the use of the technology gives rise to misuse by infringing human rights as well as intellectual property law. Proponents of AI based unconsented social media research technologies formulate the issues around this debate as to do with risks vs. benefits, such as the potential absence of humans in handling sensitive information [14,15] However, AI is a fallible process which can be prone to initial programming bias from human involvement and can also require humans to improve accuracy, therefore can never be devoid of human actors in research. [16] The global nature of this form of research means that governments and regulators internationally, should collectively develop policies to combat unconsented use of social media data for healthcare research, and also bring in new laws to prohibit unethical data mining of social media platforms. 

S. Dayal C. Pathberiya, lawyer, Acculegal Solicitors, London, UK.

Amali U. Lokugamage, consultant obstetrician & Gynaecologist, Honorary Associate Professor, Whittington NHS Trust & University College London

Disclosure of Interests: AUL is on the Board of Directors of the International MotherBaby Childbirth Organisation. Both AUL and SDCP are company directors of Docamali Ltd, a small-scale publishing company.

References:

1. COVID19 Committees.  Living online: the long-term impact on wellbeing. UK Parliament. 2020. Available from: https://committees.parliament.uk/call-for-evidence/292/living-online-the-longterm-impact-on-wellbeing/
2. Pérez-Escoda A, Jiménez-Narros C, Perlado-Lamo-de-espinosa M, Pedrero-Esteban LM. Social networks’ engagement during the COVID-19 pandemic in Spain: Health media vs. healthcare professionals. Int J Environ Res Public Health. 2020 Jul;17(14):1–17. 
3. Khan S. Coronavirus has revealed the power of social networks in a crisis. The Conversation. 2020. 
4. Saud M, Mashud M, Ida R. Usage of social media during the pandemic: Seeking support and awareness about COVID-19 through social media platforms. J Public Aff. 2020 Sep;20(4):e02417. 
5. Suler J. The online disinhibition effect. Cyberpsychology Behav [Internet]. 2004 Jun [cited 2020 Dec 6];7(3):321–6. Available from: https://pubmed.ncbi.nlm.nih.gov/15257832/
6. Straton N, Hansen K, Rao Mukkamala R, Hussain A, Grønli T-M, Langberg H, et al. Big Social Data Analytics for Public Health: Facebook Engagement and Performance. In: IEEE International Conference on Big Data. 2017. 
7. Valencia E. Web Scraping and NLP in Healthcare. Konplik. 2020. 
8. Ufert F. AI Regulation Through the Lens of Fundamental Rights: How Well Does the GDPR Address the Challenges Posed by AI? . European Papers. 2020. 
9. Graham ID, McCutcheon C, Kothari A. Exploring the frontiers of research co-production: The Integrated Knowledge Translation Research Network concept papers. Heal Res Policy Syst. 2019 Nov;17(1):88. 
10. Paul C, Brookes B. The rationalization of unethical research: Revisionist accounts of the Tuskegee syphilis study and the New Zealand “unfortunate experiment.” Am J Public Health [Internet]. 2015 Oct 1 [cited 2021 Jan 13];105(10):e12–9. Available from: /pmc/articles/PMC4568718/?report=abstract
11. UKRI. Is my study research? Medical Research Council. [Internet] Available from: http://www.hra-decisiontools.org.uk/research/ 
12. Vincent J. It’s easier than you think to craft AI tools without typing a line of code. The Verge. 2018.
13. Patel v. Facebook, Inc., No. 18-15982 (9th Cir. 2019).
14. BSA. Statement of Ethical Practice Annexe Ethics Guidelines and Collated Resources for Digital Research BSA Statement of Ethical Practice Annexe 2 Ethics Guidelines and Collated Resources for Digital Research. 2016. 
15. Sugiura L. Researching Online Forums Online Forums – Ethics Case Study 1. British Sociological Association 2016. 
16. Daws R. Google returns to using human YouTube moderators after AI errors. AI News. 2020.