Many believe that AI has the potential to complement the scientific community in major breakthroughs, especially in developing vaccines for COVID-19. Deepak Jha, Deputy General Manager- AIPF (Artificial Intelligence Platform), NEC Technologies India elaborates on the role AI can play in drug and vaccine development and associated research, in conversation with Raelene Kambli
Why is AI so important in the COVID-19 era? And how important it can be for drug discoveries in these times?
Traditionally, the drug discovery process is very long (typically it takes 8-10 years from research, development to market), complicated and expensive. AI could play an important role by expediting the overall test process from years to months by evaluating different scenarios across parameters at once. AI-based drug screening techniques can potentially help to understand essential part of the virus (like protein structure) to determine how the virus functions will play an integral role in designing drugs and accelerate the developments of vaccines to combat against COVID-19.
So how does an AI system really function in drug discovery?
AI helps in building a self-learning platform which is nurtured on the historical clinical and pharma data and real-time information. Drug design initiatives are majorly dependent on the molecular structure of the drug. Insights are drawn from the information engines to create novel drug candidates or repurpose the already available ones. AI and Machine Learning techniques can be used to predict potential targets and find the most potent among them based on different parameters like opportunity, expected response reliability and safety.
This information can be then fed into the drug design system which improves the compounds with the needed properties beforehand so that they are fit for production. Experimental data from selected compounds are fed back into the model from time-to-time to enhance the system.
British start-up and leading pharma tech company, Exscientia, is one of the first to automate drug discovery through AI. They have developed an OCD drug DSP-1181 using AI with a joint venture with Sumitomo Dainippon Pharma and completed the exploratory research phase in under 12 months which otherwise takes five years when done manually.
Though we have not witnessed a vaccine or drug yet which can cure the COVID-19 patient, many research organisations are expecting some positive outcome soon with their trust in AI for drug discovery.
Is there any option available to find entirely new molecules or repurpose existing drugs through AI? How can AI trawl through literature, study the DNA of the virus and consider the suitability of various drugs?
Yes, most of the researches are focused for the same. Google’s Deepmind utilised AI technology to predict the structure of the six proteins associated with COVID-19 which can be the baseline for further research and help in developing vaccines and also upscaling treatment. Similarly, initiatives are under progress to find new drug candidates by using the drug discovery information engine to create novel compounds that have high infinity power and could bind with the proteins associated with Corona and inhibit its property to replicate.
For example, researchers have found many similarities between COVID-19 virus and the 2003 SARS Virus and based on the existing data that caused SARS, AI learning models can be created to predict drug structures that could potentially treat COVID -19. It may potentially take lesser time for rigour testing and can help in repurpose existing drug.
Are there any surprising finds that AI has enabled researchers to discover in this pandemic?
Research labs across the globe are coming up with many suggestions on treatment and appropriate drugs/vaccines for Corona which needs to go through rigorous testing to determine feasibility, safety and reach. Coming to the question, yes, there have been some surprise findings like this virus invades the brain tissues. Maybe this the reason why many patients express the loss of their sense to taste and smell. Another interesting prediction that was made by collating the patient data through AI was that COVID-19 holds the potential to damage the reproductive system of both men and women.
We all know that presently AI cannot work alone so what are the approaches or strategies that researchers need to take if they opt for AI solutions for drug discoveries?
AI works on clean and enriched datasets as one builds and execute models on transformed data. A good amount of data, released by various health agencies and organisations is available on open platforms. However, most of the data is still lying in siloes with individual organisations like pharmaceutical agencies and their likes or are lost in the intellectual property or old information systems in various research labs. The foremost and the much-needed strategy in these times is to collate this disparate drug data so that AI researchers can apply their algorithms to derive actionable insights. For this, the world agencies and policymakers need to step up to push big pharma companies and research labs to join forces with smaller research organisations and pool data sources.
AI still has a big role to play in the fight against COVID-19, especially from a diagnostic and pharmaceutical point of view. To predict more accurate results, researchers need huge sets of good data which is currently not available and are therefore constrained by lack of data or too much noisy and outlier data. Organisations across the globe are taking necessary initiatives to use AI for testing which is encouraging, however, there is a need for more diagnostic testing.
AI can help a great deal but it ultimately depends on the clinical trials to validate the effectiveness of a drug against the pandemic.