How the pandemic has impacted the use of AI in R&D
Do you remember the first time writing an academic paper or thesis? We’ve all been there (or most of us). Struggling to find good papers, feeling like we’re digging through the whole internet without satisfying results, spending too much time in front of the screen…sounds familiar?
Technology is changing at an unprecedented rate and many fields have been impacted by it – one of them being R&D. More data was generated between 2014 and 2015 than in the entire previous history of the human race, and that amount of data is projected to double every two years. And our ability to analyse the amount of collected data is very limited.
The COVID-19 crisis has been a change accelerator in the past year and a half. The pandemic is creating risks, but it’s also creating opportunities to reinvent how researchers do their work. Due to the pandemic biopharma companies quickly turned to AI to do research for vaccines and drugs, and conduct virtual trials. Projects that may have been planned for a few years out were suddenly being implemented immediately to address the pandemic. This movement put AI into the spotlight.
How the pandemic has impacted the use of AI in R&D
Finding relevant articles faster
One of the most crucial tasks in the R&D process is finding relevant data for later use. And with millions of papers available online it’s difficult to quickly dig through it all. Artificial intelligence enables researchers to discover relevant literature faster and more accurately. Instead of using only key words, researchers can formulate their own problem statement or use an existing text from the specific domain. For example, you can upload a clinical paper about a concrete disease to the Iris.ai tool, which will find the most important terms in that paper, identify synonyms and hypernyms, and match that to scientific documents in the databases. This is often called a content-based recommendation engine. The engine scores the literature found from 0 to 100% (100% being a duplicate) based on how similar it is to the initial paper uploaded.
AI and vaccine development
It can take years to develop a vaccine, but during the pandemic we were able to do vaccine tests already after three months from the first reported cases. This was done at a record speed thanks to AI tools that helped researchers analyse data faster and more accurately. For example, a ML model developed by Janssen and MIT data scientists played a key role in the clinical trial process for the Johnson & Johnson Covid-19 vaccine. Developers of the vaccine, leveraged real-world data and applied artificial intelligence and machine learning to help guide the company’s research efforts into a potential vaccine.
Viruses have hundreds of protein components and ML models are able to go through this data fast and point out the most immunogenic subcomponents that the vaccine should be focused on. The way of how AI was used in the past year will revolutionize how vaccines are made in the future.
Clinical trials
Pharma companies usually choose to do Randomised Clinical Trials to assess potential new medicines. However, it has become more expensive and complex over time due to a large number of participants needed to ensure sufficient statistical power and the duration of follow‑up which could be even 5 years.
Advances in AI can help us to rethink how we do clinical trials – it can reduce clinical trial cycle times while improving the costs of productivity and finding new ways to discover and develop potential new medicines. For example Novartis used AI and combined clinical trial data from a variety of internal sources to predict and monitor trial cost, enrolment and quality. In consequence, the company reported a 10-15 % reduction in patient enrolment times in pilot trials.
The Covid-19 also accelerated the adoption of telemedicine and remote monitoring solutions, which has driven interest in virtual clinical trials. The potential benefits of virtual trials include reduced costs, a wider network of eligible patients, and better patient retention rates. For example the company “Science 37” offers end-to-end clinical trial services using its virtual Metasite model. It leverages a network of investigators, mobile nurses, and study coordinators with the aim of making studies more accessible to patients.
Post-market surveillance
Managing and reporting adverse events (AEs) is getting more and more challenging for pharmaceutical companies. As the number of AEs is expected to increase dramatically more products are introduced to the market and pharmacovigilance teams need new technological solutions to help them keep up with the pace. AI tools can help with the data collection of case and clinical trial reports, scientific literature and even social media reports, the review of the collected data, the extraction of all relevant data points, such as adverse events and lastly, signal detection. At Iris.ai we have developed a set of modules for automation of post market surveillance – systematic literature reviews, including continued surveillance, advanced categorization and classification of all papers as well as extraction of the key data. You can read more here.
Key takeaways
The COVID-19 pandemic had a big impact on the use of technologies in R&D, especially in drug development. Artificial Intelligence was put into the spotlight and changed the classical way of how researchers work. AI-enhanced technologies have unparalleled potential to collect, organise and analyse the increasing body of data. Artificial intelligence can be used in many ways – literature review, vaccine development, clinical trials and post-market surveillance.
