Artificial Intelligence and Its Role in Vaccine Efficacy and Pharmacovigilance

Welcome to Health Connect, the podcast for health professionals where we will share the latest news and information on science and technology in the medical field. In this episode, we will talk about artificial intelligence and pharmacovigilance in the field of vaccines.

Did you know that the use of artificial intelligence and machine learning can help improve pharmacovigilance as we know it now?

A pandemic is boundless and capable of causing immense morbidity and mortality. Globally, there have been several pandemic outbreaks, such as the Black Death, the Spanish flu, Cholera, influenza, and more recently, COVID-19, to name a few, and they are capable of causing social and economic disruption.1

The increasing challenge of modern medicine to constantly improve and catch up with evolving human diseases requires an urgent and thorough approach for the benefit of global health.2

Predicting an epidemic and pandemic is a challenge that requires a large amount of resources and time. However, with current advancements in medical science, studying the propagation of dangerous diseases is made possible.1

During the COVID-19 pandemic, a number of vaccines against the virus were developed and found to be effective, and they were rolled out at an unprecedented speed and scale across the world.3

Active surveillance demands huge resources, manpower, and time.1 Thus, a major component of vaccine pharmacovigilance strategies should be the use of robust surveillance systems to monitor for adverse effects following immunization. This is particularly important given the persisting levels of vaccine hesitancy, especially now that new vaccine technologies are being employed for the first time.3

Pharmacovigilance is fundamentally a data-driven field, as it requires the collection, management, and analysis of large amounts of data from a wide range of diverse sources.4 

The primary type of data used in pharmacovigilance are individual case safety reports, which are records of suspected adverse events collected via multiple channels, aggregated and organized into large databases, and constantly monitored to detect safety signals.4

Individual case safety reports come from various sources, including chatbot interactions, electronic health records, published literature, patient registries, patient support programs, or even directly from patients via social media. Reports are collected worldwide and characterized by heterogeneity in format, language, and unique characteristics of the underlying healthcare systems. Adverse events must be identified and analyzed in order to find potential emerging safety issues in medicines and vaccines.4

The central challenge of pharmacovigilance is to interpret large amounts of heterogeneous data to quickly and reliably detect safety signals that require escalation and triage.4

Given the rise of artificial intelligence and machine learning across many fields of science and medicine over the last decade, many have speculated that these same technologies could help with the aforementioned challenges of pharmacovigilance.4

Artificial intelligence can be used to achieve surveillance with the efficient utilization of resources. Machine learning and deep learning are being incorporated into various healthcare segments and are found to be more effective when compared to human resources.1

Nevertheless, caution must be used when attempting to extrapolate the success of machine learning in other areas compared with pharmacovigilance, since there are specific factors that may account for the recent success of machine learning that may or may not be present for pharmacovigilance applications.4 

Outstandingly, deep learning methods are a type of machine learning that are scalable and can train on large amounts of data through the use of graphical processing units and continue to improve. In addition to scalability, the modular nature of deep learning brings the added benefit of easily incorporating domain-specific knowledge, often called an inductive bias, to point the model in the direction of good or parsimonious solutions. Without large data and inductive biases, numerous studies with deep learning are often no better than traditional statistical models. 4

Although image recognition is not a common task in the current pharmacovigilance pipeline, deep learning models known as convolutional neural networks offer an example of how potent the combination of large data and domain knowledge can be.4

In this regard, Kim and colleagues developed a machine learning-based active surveillance system for detecting possible factors that can induce adverse events using health claims and vaccination databases.5

The researchers used two large databases, one of which contains flu vaccination records for the elderly, and the other contains the claim data of vaccinated people. They developed a machine learning model based on a case-crossover design to predict the health outcome of interest events, namely anaphylaxis and agranulocytosis, using a random forest model.5

The model was trained well, with an area under the curve of over 99% for the training data, and an accuracy for predicting the occurrence of health outcomes of interest on the test data of around 70% for both, anaphylaxis and agranulocytosis.5

The results of the study demonstrated that, by linking those two national health record databases, the model can help establish a system for conducting active surveillance on vaccination.5

It also demonstrated that the system based on machine learning for vaccine adverse event monitoring was able to detect which features affected agranulocytosis and anaphylaxis, using the elderly flu vaccine cohort data from two databases.5

CÁPSULA
Influenza has a detrimental impact on individuals and society, causing nearly 650,000 deaths and millions of hospitalizations every year around the world. Vaccines are the best available means for influenza prevention and control, reducing the frequency of morbidity, severity of infection, and mortality. However, the protection induced by vaccines depends on annual re-vaccination to confront new circulating strains. This represents an unmet need to increase the protection of influenza vaccines and help prevent future pandemics.6

New approaches and technological developments in systems immunology, that include a combination of multi-omics technologies and novel computational methods for integrative analysis, have been showing potential to help discover new depths of influenza immunity by examining roles and interrelationships of clinical, biological, and genetic factors in the control of influenza infections. These advances could lead to a deeper understanding of key factors mediating protective immunity of influenza, and could help develop vaccines that are able to induce long-term immunity against divergent viral strains and prevent future pandemics.6
FIN DE LA CÁPSULA

In this data-driven world, innovative measures to analyze and recommend effective and efficient systems, approaches, techniques, and methods to combat viral diseases more efficiently and cost-effectively becomes highly important for an improved health system; especially for developing countries with limited access to adequate resources.2

Such measures must therefore be applied in vaccination production and distribution, since vaccines have been identified as the most potent method to curb the spread of viral diseases such as the coronavirus. With the ongoing distribution of vaccines and with its accompanying challenges; it is thus also fitting to highlight the need for further optimization of vaccine distribution, especially to poorer regions of the world and priority populations.2 

The development and continuous distribution of the COVID-19 vaccines offers great hope going forward toward the eradication of the virus. The rate of vaccinations is crucial in this effort; therefore, it is pertinent to analyze and forecast how quickly vaccines can be distributed across the world and as needed to ensure that the virus is eliminated effectively and efficiently when dealing with a global pandemic.2

National, regional, and global strategies influenced by economic, geographic, and climactic conditions are required to effectviely fight a pandemic at the global level and to reach herd immunity. Effective and equitable distribution of vaccines must be carried out uniquely taking into account the situations and circumstances within each country.2 

Efforts in the battle against COVID-19 and other viruses should consider multiple factors to create a multidimensional predictive testing and vaccination approach, in both rural and urban communities, and make use of remote diagnostics and treatment techniques using data driven technologies, to ensure that testing, diagnosis, and treatments or vaccinations will be carried out efficiently and effectively at minimal cost.2

Modern technologies, such as artificial intelligence can be used to interpret huge amounts of data for analysis, aiming for an even more collaborative, productive, and proactive healthcare industry. Various algorithms can be used to analyze data from many sources, including information on the spread of communicable diseases like COVID-19, the distribution of vaccines, genomic data, health records, preventive measures, and unique geographical and demographic situations.2 

The recent developments in artificial intelligence are fascinating, especially its application in healthcare, pharmaceutical research and drug and vaccine development. It is therefore highly likely that AI-enabled technologies will continue to evolve and offer many opportunities in various sectors of healthcare and pharmaceutical research, which could be game-changers in futuristic research.1

Thanks for joining us on this episode of Health Connect. Don’t miss out on our next episode. 
Discover more medical news on Viatris Connect.

 

References:

  1. 1. Bhattamisra SK, Banerjee P, Gupta P, Mayuren J, Patra S, Candasamy M. Artificial Intelligence in Pharmaceutical and Healthcare Research. Big Data Cogn. Comput. 2023;7,10. Available at: https://www.mdpi.com/2504-2289/7/1/10
  2. 2. Kalu CK. Analysis of COVID-19 Vaccinations and Symptom Mapping Diagnostic Technique for Viral Diseases: Using Data Analytics, Machine Learning, and Artificial Intelligence. Inquiry. 2023; 60:469580231164480. Available at: https://journals.sagepub.com/doi/10.1177/00469580231164480
  3. 3. Hussain Z, Sheikh Z, Tahir A, Dashtipour K, Gogate M, Sheikh A, et al. Artificial Intelligence-Enabled Social Media Analysis for Pharmacovigilance of COVID-19 Vaccinations in the United Kingdom: Observational Study. JMIR Public Health Surveill. 2022;8(5):e32543. Available at: https://publichealth.jmir.org/2022/5/e32543
  4. 4. Kompa B, Hakim JB, Palepu A, Kompa KG, Smith M, Bain PA, et al. Artificial Intelligence Based on Machine Learning in Pharmacovigilance: A Scoping Review. Drug Saf. 2022;45(5):477-491. Available at: Artificial Intelligence Based on Machine Learning in Pharmacovigilance: A Scoping Review
  5. 5. Kim Y, Jang JH, Park N, Jeong NY, Lim E, Kim S, et al. Machine Learning Approach for Active Vaccine Safety Monitoring. J Korean Med Sci. 2021;36(31):e198. Available at: https://jkms.org/DOIx.php?id=10.3346/jkms.2021.36.e198
  6. 6. Tomic A, Pollard AJ, Davis, MM. Systems Immunology: Revealing Influenza Immunological Imprint. Viruses. 2021;13,948. Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8160800/

Os links para todos os sites de terceiros são oferecidos como um serviço aos nossos visitantes e não implicam endosso, indicação ou recomendação do Health Connect. Os artigos vinculados são fornecidos apenas para fins informativos e não visam implicar uma atribuição pelo autor e/ou editor. O Health Connect se isenta de qualquer responsabilidade pelo conteúdo ou pelos serviços de outros sites. Recomendamos que você analise as políticas e condições de todos os sites que escolher acessar.

NON-2023-10041
NON-2023-2512