Accelerating the Drug Discovery Process Using Artificial Intell…

There has been a great shift within the pharmaceutical industry to get medicines to patients faster. Focusing efforts to where they are needed the most, research and personalized healthcare for our patients. The COVID-19 pandemic has seen an urgent need to process vast amounts of data across various organisations fast. Although there are many companies using Artificial Intelligence (AI) experimentally, the pandemic has brought them to the forefront of the fight against the pandemic.

Interestingly, a company called BlueDot warned its clients about Wuhan becoming a danger zone when the virus first emerged. It used an algorithm driven by AI to look across different news reports, plant and disease networks, and official reports to detect the early warning signs. It was essentially an AI epidemiologist which could become a valuable resource to prevent pandemics in the future.

It is all about mass amounts of data. The COVID-19 pandemic has seen us share data on a global scale like never before, in an urgent scramble to search for possible treatments and vaccines. This is where AI can help in several ways:

• Repurposing existing drugs.

• Finding the best combinations of drugs to use.

• Modelling for viral genome sequences – looking at infection rates, if they can cause disease and searching for effective vaccines.

• Identifying new molecules.

It takes many years to develop a new drug, from 3-5 years to discover new molecules to 7 years in human trials which is too long for pandemics. Thus, the most common strategy in a crisis such as this one is to, ‘repurpose’ drugs. To find out if drugs already on the market for other diseases, can be effectively used to treat emerging diseases such as COVID-19, but this can also be a slow process. It requires knowledge about the disease, which may be difficult for novel viruses such as this one, and still involves multiple clinical trials to establish safety profiles. Remdesivir is an example of a repurposed drug for COVID-19 as it was originally approved for Ebola. Hydroxychloroquine was another which could have been repurposed for COVID-19, currently approved to treat Malaria, but it was not approved by the FDA. The next step is to look for drug combinations which work well together, for this, trials are currently ongoing.

Many global organisations have come together to openly share their compound libraries and data with each other in the face of the COVID-19 crisis. The Institute of Cancer Research (ICR) which has designed ‘Coronavirus canSAR’ that pulls in data from various sources such as the international Protein Databank. The use of AI algorithms makes it possible to analyse vast volumes of data and find meaningful patterns which cannot be done by humans alone. By doing so this cuts down the time and billions of pounds lost in failing discoveries. Algorithms also allow us to look at how virus proteins interact with human proteins by creating complex computerized maps. These can help with vaccine development where more accurate predictions can be made.

A blockchain can allow competitive companies to share sensitive data on a permanent ledger without revealing it all. Blockchain can be used to monitor pandemic donations, relief and material distribution, and other responses in a fast and transparent way without giving away the user’s data. The unforeseen pandemic has put mass strain on supply chains, but blockchain technology may provide the answer. Several pharmaceutical companies use the MediLedger Network which is a product verification system that helps to verify if returned drugs are authentic.

Machine learning has also been used on anonymized patient data to help identify patients at risk of developing more severe disease, and therefore predict the resourcing which would be required such as ventilators and staff. Machine learning also helps to identify personalized healthcare, moving away from the ‘one glove fits all’ approach when it comes to treatment as every patient is different and may respond to the same drug differently, or in some cases not at all. More information and the ability to share more information means physicians can make better informed decisions on patient care, ultimately saving more lives.

For those who cannot get tested, an AI tool has been designed by a company called ZOE and Kings College London and Massachusetts General Hospital to predict if they have COVID-19 based on their symptoms. Their work has been published in Nature Medicine. Their app has been downloaded by 3.3 million people and it works by using an algorithm to identify known symptoms of COVID-19 in users, against data such as their age and sex; it is said to be 80% accurate. Combining the use of this tool with other modelling, data could help to identify early signs of the disease and thus control the spread of COVID-19.

The use of AI in diagnostic and drug development is still a relatively new technology, but is rapidly changing the way we work. It must be used with human input to provide intelligent solutions to complex problems. We still have much to learn in how to utilize this tool effectively and safely, but the possibilities are endless, the future looks promising.