AI & VACCINE DISCOVERY
AI & VACCINE DISCOVERY
Scope of Use (only research)
- Machine learning approaches have been used (link) for identifying antigens from protein sequences which assist B-cell and T-cell antibodies in their ability to bind and attack invaders.
- Besides, a drawback of AI models in the context of Covid-19 is that their real utility remains largely untested. In Covid-19 research, AI-based models are theoretical (link).
- One of the most direct applications of ML and other AI-based strategies in vaccine discovery and development is to identify the presence of antigenic peptides presented by MHC-II (molecules which induce antigen-specific responses, which are central to vaccine-induced immunity).
- In this context ML has been reserached to develop programs such as MARIA (link) that predicts antigen presentation.
- AI tools that have been researched in the context of Covid-19 vaccine discovery include MARIA, NetMHCPan4, Long Short-Term Memory network, deep-learning Recurrent Neural Networks and MoDec (link).
- In general, various AI tools have been researched (link) to analyze SARS-CoV-2 viral peptide presentation on MHC molecules from patients to understand natural immunity with the aim to discover COVID-19 specific immune response and assist in designing an effective vaccine.
- An example of AI-related tools in that context is the Vaxign-ML-based reverse vaccinology tools to predict targets that could be used to develop a safe and effective COVID-19 vaccine.
Technological robustness and efficacy (no evidence)
- In recent years the successful application (link) of ML has revolutionalised many fields of science including vaccine discovery.
- In relation to the AI-based tools used in the context of Covid-19 vaccine discovery and development there are not adequate evidence regarding their robustness and efficacy.
Impact on citizens and society (blue=pos, red=neg)
- We consider this type of AI use to have a predominantly positive impact on citizens and society.
- Validation, generalization, explainability, interpretability, risk mitigation, fairness, and inclusiveness are some of the key challenges (link) in making AI-based decisions in medical and public health settings such as the vaccine discovery.
Governance and Accountability (importance)
- In clinical and healthcare settings, transparency, privacy, fairness, safety and liability are major challenges (link) in terms of ethical and regulatory aspects of AI.
- The issues concerning bias and lack of transparency should be dealt by engaging different stakeholders in the Covid-19 vaccine discovery process.
- Explainability and interpretability are two important factors that need governance to monitor and enhance AI algorithmic fairness, transparency and accountability.
- “Ethical auditing” could be deployed to examine the inputs and outputs of AI algorithms and models for bias and potential risks.
 Goodman, K.; Zandi, D.; Reis, A.; Vayena, E. Balancing risks and benefits of artificial intelligence in the health sector. Bull. World Health Organ. 2020, 98, 230–230A.
 Cath, C. Governing artificial intelligence: Ethical, legal and technical opportunities and challenges. Philos. Transact. A Math. Phys. Eng. Sci. 2018, 376.