Vaccine Mediator

AI-powered online mediation platform to provide accurate vaccine information as well as facilitate the resolution of vaccine injury disputes.

Samuel Dahan

Seventy to ninety percent of a population must be vaccinated or have antibodies resulting from infection to arrive at the safe harbor known as herd immunity, wherein the whole community is protected. Public acceptance of vaccines during vaccination campaigns is therefore critically important in controlling pandemics. However, there are concerns that herd immunity may not be achieved, as data on vaccine attitudes indicate that there is a worrying increase in people who are reluctant or even outright refuse, to accept recommended vaccinations. Even before the COVID outbreak, the World Health Organisation declared vaccine hesitancy a major global health threat. The current pandemic has only exacerbated mistrust in science and public institutions, with 38.7 million people following anti-vaccine groups on Facebook and Instagram, and only 50% of Americans committed to receiving a vaccine -- with the communities most at risk from the virus being the most hesitant. It is also clear that anti-vaccine activists are disseminating misinformation about vaccines in order to foment fear and exacerbate political divisions. Whilst the outright rejection of vaccination, though on the rise, remains relatively low, the broader phenomenon of vaccine hesitancy and distrust in science, wherein people delay or selectively choose vaccinations, is increasing rapidly.

We hope this system will benefit governments, the research community, members of the public, and potential victims in several ways. First, messaging about our research will be immediately integrated within social media platforms to promote vaccine acceptance. Contacts with policymakers will ensure that our research is incorporated into their internal discussions. Public engagement will also be advanced through focus groups, as well as an intensive digital marketing campaign designed to enhance the accessibility of the solution.

Secondly, the system would improve access to justice by helping alleged victims to access accurate, free, anonymised independent information and dispute resolution processes drawn from reliable medical sources. It would assist regulators in fulfilling their ethical obligation to protect those who have assumed the risk of taking vaccines for the benefit of others. In addition, considering the distrust many public institutions are currently experiencing, the positive impact of proactively establishing such a system will include improving the success of the immunisation programme while reducing the overall cost to society.

Our solution will contribute to the public good in two ways: it provides a free-to-use-platform for access to reliable health information, and it offers legal guidance which is otherwise unaffordable for many people. First, the Vaccine Mediator platform is designed to increase vaccine uptake by promoting public acceptance of vaccines, thanks to an open-access search engine for vaccine misinformation. The tool assesses whether a user’s claim is drawn from false information, a root cause of public mistrust and vaccine hesitancy. Second, the platform helps to promote access to justice by facilitating access to vaccine injury remedies at no cost to injured victims. It is worth noting that these two services would contribute significantly to the public good at any time, but they will have an outsized impact in the urgent context of the current pandemic.

Our pilot solution is expected to have a significant impact on the general population in that herd immunity will only be achieved when at least 70-80% of the population is immunized. Recent research suggests that some sociodemographic groups are disproportionately affected by exposure to misinformation. In addition, scientific-sounding misinformation is more strongly associated with declines in vaccination intent. 

We argue that misinformation is only the trigger. The most common reason for vaccine hesitancy amongst minority groups is concern about potential adverse effects. Studies indicate that this mistrust is deep-seated, even among minority healthcare workers who also show lower vaccine uptake in the UK and the USA.  

 Our outreach campaign will therefore focus on specific groups that are targeted by misinformation campaigns. A rapid and transparent reporting system for adverse vaccine effects, coupled with a dispute resolution system, would improve communication and potentially reduce vaccine hesitancy among minority populations. For instance, systems such as the Yellow Card Scheme and V-safe have already empowered individuals to provide real-time information on side effects, and have helped to establish a trusted, public-led vaccine-safety monitoring programme.  Our system will build on those programs, bolstering vaccine acceptance and uptake in our target populations.

Once the Canadian pilot program proves successful and we have amassed sufficient reliable data on side effects, we will begin engaging with American and European users (especially in France and the UK). However, we can only deploy a full AI solution once we are able to access regulatory data on causation and injury compensation, which is only possible through partnerships with the Public Health Agency of Canada, the CDC, and the MRHA in the UK. Eventually, our solution could be deployed in developing countries through the GAVI-COVAX AMC scheme, for which 92 low- and middle-income economies are currently eligible.  In particular, we hope to focus on countries that are most affected by COVID (India, Pakistan, Nigeria and Cameroon).

The solution should also be scaled for other vaccines. Vaccine hesitancy and resistance are on the rise in general, leading, for instance, to a 300% increase in the incidence of measles in the first three months of 2019 compared to the same period in 2018, according to the WHO. While it will require additional research, we believe we can adapt our work on the COVID vaccine to other vaccines as well.

We have chosen two metrics to assess the impact of the proposed solution and pilot project: the number of users that have consulted and completed the tools (i.e. completed the form in its entirety), and how many disputes were resolved. We have been collecting that information for our pilot tool – MyOpenCourt – via Google Analytics.

Please note that in the case of the Vaccine Mediator, we will take into consideration two other variables, namely whether users have decided to follow the recommendations of the mediator, and any discrepancy between our prediction and decisions of the Canadian Vaccine Injury Compensation Scheme (currently being established). 

We need to overcome three challenges in order to deploy a successful solution.

First, developing legal AI applications is both costly and technically challenging. Hiring a cohort of qualified lawyers and medical scientists to process the data represents a significant cost. In addition, algorithms for misinformation rectification are still somewhat limited, so further work is necessary to optimize their performance.

Second, we only have access to survey and self-reported data, which is always biased as it may be skewed by external forces such as press reports and thus produce misleading spikes in reported side effects. We believe that mining datasets such as those curated by the Clinical Practice Research Datalink (with over ten million primary care records) would generate more meaningful data on correlation and causality. Our success will depend on convincing regulators to focus on primary care and regulatory data. 

Finally, considering current levels of mistrust in science and expertise, we anticipate user reluctance. As a result, we will devote significant time and resources to user engagement strategies in order to convince people to trust this system as a neutral self-help solution.

3 different centres will be working on this project divided thematically between:

ODR / Data analytics: Conflict Analytics Lab at Queen’s and McGill Universities

Social sciences/history of medicine: University of Oxford / Dublin

Law & ethics: British Institute for Comparative and International Law (BIICL), and the University of Manchester

Besides offering financial support, The Trinity Challenge can help us engage with relevant stakeholders, including data holders such as regulators, social media companies and medical practitioners. This would help to guarantee the success of our solution insofar as it is highly dependent on the quality of the data. 

Crucially, The Trinity Challenge would also assist the Vaccine Mediator solution with outreach and user engagement, both in terms of the visibility afforded by the association with The Trinity Challenge and in terms of the potential partnership opportunities facilitated by the Challenge, as described below in 31.

We would like to partner with national vaccine injury compensation schemes in the UK, France, and Canada, as well as international-level programs such as the COVAX No-fault Compensation Programme. We are currently in discussions with several such organisations, including the Public Health Agency of Canada, in order to optimise the impact of our project. 

We would also like to collaborate with The Trinity Challenge Member organisations. The Bill & Melinda Gates Foundation would be an excellent partner. Their Vaccine Development and Surveillance project, with its focus on vaccine education and the development of innovative technical solutions in this area -- including data-driven modelling and forecasting capabilities -- relates closely to our work. Similarly, Doctor Evidence would be a valuable partner given their data synthesis expertise in the healthcare setting. We would also be interested in exploring the addition of a behavioural science component to our project through a partnership with The Behavioural Insights Team (bi.team). Partnerships with these organisations would enhance the development of our predictive system, thus assisting in outreach and the dissemination of our solution. We would also like to engage with Facebook in order to leverage our own tools for combatting vaccine misinformation.