Beyond R&D: AI’s potential to transform the vaccine ecosystem
By: Amandeep Singh, Sanket Shelar, Amudha Sharma, Sara Ackermann, Regan O’Marra, Keely Anderson and Adi Natu (ZS Vaccines Center of Excellence)
Artificial Intelligence empowers machines to perform a myriad of tasks traditionally associated with human capabilities. In the realm of healthcare, AI and machine learning (ML) have proven to be transformative, accelerating research and development (R&D) and commercialization of drugs and vaccines, improving accuracy in medical diagnosis and allowing personalization of treatments for rare diseases. The integration of AI extends across the entire healthcare value chain, contributing to unprecedented advancements.
The impact of AI in healthcare is profound, with estimations suggesting potential annual savings of $150 billion in United States healthcare costs by 2026. While AI’s presence in various industries predates recent years, its significance has skyrocketed, garnering widespread attention, with the advent of groundbreaking platforms like ChatGPT. Unlike conventional AI models, ChatGPT, a generative AI platform (gen AI), distinguishes itself by not merely “knowing” information but by possessing the unique capability to “create” information. This innovation has spurred increased interest in exploring AI’s potential, leading experts to contemplate its applications in the vaccine ecosystem.
AI in vaccines
One of the earliest applications of AI use for vaccine development occurred in 2018, when a team of researchers from Australia’s Flinders University developed a new flu vaccine using a ML algorithm called Search Algorithm for Ligands (SAM) to analyze compounds of interest from a vast library. The vaccine, named FLU-v, showed promise reducing flu symptoms and enhancing immunity in a phase II clinical trial.
Since then, the application of AI in vaccines has gained momentum, and now extends across more than 10 modalities including research, clinical development, epidemiology, distribution and marketing. Table 1 highlights some applications of AI in vaccines.
Table 1:
In addition to its applications in the aforementioned areas, we anticipate a pivotal role, in particular for gen AI, to address critical areas of concern around vaccine misinformation, promoting equity and mitigating vaccine hesitancy.
Gen AI can help identify and tackle vaccine misinformation at the source
The World Health Organization (WHO) recognizes misinformation as an urgent threat to public health with potential lethal consequences.
The dissemination of online misinformation has had a detrimental effect on vaccine uptake. Many studies have reported vaccine refusal linked to misinformation and conspiracy theories. For instance, a 2022 study by Pierri et. al., in the U.S. predicted an approximately 20% decrease in vaccine uptake across states, and an approximately 67% increase in hesitancy rates across democratic counties attributed to misinformation.
The Times of India reported that approximately 91% of COVID-related fake news was attributed to internet-based sources, with 85% of misinformation generated through social media. Another study conducted by the Royal Society for Public Health in the UK found that two-fifths of 2000 participants encountered negative messages about vaccination on social media platforms.
Given this scale, it is crucial to curb the spread of misleading, false, and exaggerated information at its source to build trust in vaccines and improve vaccination rates. A study by Brewer et al. found that correcting misinformation about vaccine safety increased vaccine acceptance by 10.2% among vaccine-hesitant adults in the U.S.
Gen AI can be leveraged to identify, track, address, and prevent the spread of misinformation.
Machine learning algorithms and complementary automated text reading and classification programs can accurately distinguish real or true information from misinformation. AI can be used for data and content analysis, fact checking, sentiment analysis, monitoring, alerts, warning systems and more. Furthermore, the multi-lingual capability of AI can be leveraged to create chatbots or information systems to examine region specific sources of information.
Various AI tools are already in use to identify and curb the spread of false data. For instance, “The Dark Crawler (TDC)” has been customized to acquire datasets that are representative of COVID-19 misinformation from social media and online news sources. TextBlob is another AI application that was used for analyzing tweets and building stats around public perception of COVID-19 vaccination. Researchers have also employed the latent Dirichlet allocation model for topic modeling of X — previously known as Twitter — data. Other examples of such tools include WHO’s Misinfo Watch, Vigilance.ai, Meedan’s Check Global and Jigsaw’s Perspective AP.
There is however a need for more sophisticated AI tools.
While several databases help identify and tackle sources of misinformation, many AI tools face limitations accessing encrypted, secured, or hidden data on the dark web. Moreover, ethical, and legal concerns regarding personal data impose constraints on their potential.
Experts suggest that in the short term, AI applications could focus on generating informative summaries and develop methodologies to assess data on the internet. The long-term goal would be the development of an automated ML tool to assist social media platforms, online service providers and government agencies in directly identifying, checking, assessing, and responding to sources of misinformation about vaccines on social media.
Gen AI has the potential to drive vaccine acceptance through impactful communication with stakeholders
The decisions consumers make regarding vaccinations are complex, multi-factorial and often contextual. Through a global research study with consumers, ZS experts have identified and mapped the various mental operations and cognitive factors (heuristics, mental models, etc.) contributing to vaccine hesitancy. Ultimately, incomplete, ineffective or inefficient communication among vaccine stakeholders (e.g., health care professionals, consumers etc.) can result in increase in information gaps and fuel misinformation, which impedes rational decision-making around vaccination. Thus, interventions to combat vaccine hesitancy must be tailored to the decision-drivers of specific populations to drive further vaccine uptake.
Gen AI can improve vaccine knowledge dissemination and enable positive decision making for specific sub-populations.
AI tools and datasets can be worthwhile for catering to the needs of both healthcare practitioners and patients. They can be utilized for identifying, gathering, analyzing, explaining, and communicating the most relevant and accurate information on vaccine safety and effectiveness to consumers which helps improve their decision making. Explaining complex medical information in easy-to-understand ways is crucial to build trust in vaccines. AI tools such as ChatGPT and Bing can also help to identify and communicate the factors that positively influence people’s attitudes and behaviors towards vaccination.
AI tools also play a key role in implementing targeted interventions within specific demographic segments to promote positive information about the importance of vaccination in hesitant populations through messaging, scientific engagements, and the promotion of vaccination benefits by local community leaders. Additionally, various AI-assisted initiatives like videos, consumer stories and posters promoting vaccination can effectively address and target cognitive biases and improve vaccine uptake.
Experts foresee an evolved gen AI tool to help tailor personalized HCP and consumer conversations.
Gen AI may support HCPs in having more effective, meaningful conversations to address vaccine hesitancy across their patients. Translation and transcription of scientific findings from disparate sources into a comprehensive story could lead to enhanced awareness across physicians who in turn can disseminate benefits as well as general information of vaccines across consumers. The ultimate objective for future AI tools is to guide HCP decision-making by providing personalized insights and enhancing patient interactions. Sophisticated chatbots can deliver relevant, accurate and up-to-date insights that have the potential to improve education and awareness.
Furthermore, gen AI tools may be able to analyze unstructured data from multiple sources like doctor transcripts, calls, information leaflets and PDFs to yield sanitized and valuable insights for developing marketing strategies and vaccination campaigns.
AI is here to assist human work, not replace it
The tremendous contribution of AI during the COVID pandemic is commendable and persistent efforts are in progress by vaccine stakeholders to reap more benefits out of this. More and more vaccine manufacturers are investing in using digital technology for multiple processes.
However, amidst a rapidly transforming vaccine industry landscape with amalgamation of AI into processes, it is crucial to be mindful of multiple insecurities and risks linked to AI. This is critical since AI could be manipulated to spread false information rather than to tackle it, and to target and plant misconceptions, rather than addressing hesitancy. Additionally, we need to be cautious about multiple threats like deepfakes, copyright issues and other malicious uses of gen AI technology.
The benefits of gen AI “when used in the right ways” are for all to see. Therefore, it is critical that decision-makers across the ecosystem begin to explore applications in concert with a “human-centered” approach. There is an urgent need to identify threats to vaccinations and build effective strategies to tackle them in each consumer touchpoint. Targeted, practical applications of AI in support of HCPs through pilots may be a logical first step. For instance, HCPs — in collaboration with public health officials — may be able to customize their communication strategies according to different populations to combat misinformation and more effectively motivate them to receive vaccinations. All in all, it would be exciting to witness more innovation brought by AI, ML, and other digital technologies in the coming years to increase vaccination uptake globally.
References:
2. MDPI
3. Nature
6. Nature
7. PLOS One
8. Scrip
9. ZS.com
10. Upgrad
11. Google Cloud
12. CEPI
13. GIE Journal
