The drug development process is an expensive and difficult task that can take up to decades. Yet, for more than 50 years the drug development pipeline has remained virtually unchanged since its development in the 1960s. With rising R&D cost, attrition rate and strictness in regulations, the pharmaceutical industry needs a new solution. Could AI be the panacea that ails the Pharma Industry ?

Vishnu Vettrivel
Vishnu Vettrivel

Vishnu Vettrivel, the founder of Wisecube.ai, was recently interviewed by Angela Lai a researcher in biomedical entrepreneurship, on the topic of AI application in drug discovery and development.  Here are the excerpts from that interview:


Can you tell me a little bit about Wisecube.ai?

Wisecube is a hybrid cloud AI platform company focused on life science research acceleration. We build AI platforms that empower scientists to be more accurate and speed up the development process. 

What do you think is the biggest challenge in the drug development process? 

There are so many aspects of the human body to consider when it comes to drug discovery, for example, enzymes, proteins, metabolic pathways etc.  How can we find the molecules that fit all the criteria? How can we predict their effects? Can we predict them even before we try them on people?  Can we be sure about our prediction? What are the appropriate research techniques to use? There are just so many dimensions to consider and data to be processed before scientists can find the right therapies. That is why drug discovery takes such a long time.

Do you think the implementation of AI could improve such problems?

“If you want to solve the drug discovery problem, you have to solve the data problem first!” 

So there are two dimensions of data. The first dimension has only a single set of data points, for example, pictures of cats and dogs. These types of data could easily be learned just by providing the AI with examples of each. 

The drug development data, however, belongs to the second dimension of data that has multiple sets of data points, for example, metabolic pathways, genes, proteins, toxicities, etc.

The challenge to improving the drug development process lies in solving the second dimension: How do we combine data sets? How do we learn from different sets of data and make predictions?  How do we learn from all sorts of data? At Wisecube AI, we utilize deep learning to make predictions from analyzing data from multiple data sets to speed up the process of drug development.

Which stage of drug development value chain does the AI tool from your company focus on?

Our current focus is on non-clinical stages of drug development, i.e. early drug discovery, preclinical studies and repurposing of existing drugs.   

What does your AI tool do? What value does it offer? Which type of learning does it use?

What we do at Wisecube AI is that we use our AI engine to process publicly available data sets, combine data sets into knowledge graphs, and use advanced machine learning to predict possible molecule candidates or drug applications so that the time scientists require to search and screen is greatly reduced. 

Take drug repurposing as an example.  Let’s say we have an enzyme of interest, our AI will screen all publicly available literature that talks about the enzyme and through processing information relevant to the enzyme, the AI predicts using link prediction, directly or indirectly, the possible uses of this enzyme and makes suggestions. It’s like Facebook’s friend suggestion function. 

How do you collaborate with pharmaceutical companies? 

We use a licensing model. Pharmaceutical companies approach us with a scope, for example, repurposing of a certain kinase. Our AI engine will then search through all available information, build knowledge graphs and make predictions of other possible applications. With different scopes the solution will be different but the fundamental methodology is the same.

Has your AI successfully been implemented ?

Yes,we have been able to demonstrate to pharma companies that the result works. This is proven science, proven work. We can tell the scientists why AI makes certain suggestions? Why does AI recommend this molecule? There are reasons for every action and those reasons are accessible to our clients.  It is not a Black Box Model! 

Do you think in the next 2-3 years there will be drugs made from a process that utilizes AI? Do you see AI becoming the norm in drug development in the future? 

Yes, it will definitely happen, it is just a matter of time. However, AI is just an accelerator, a tool that helps narrow down the pool of targets and speed up the process,  it will not replace humans. People watch too many sci-fi movies and form unrealistic views of AI.

What advantage do you think you have over the traditional drug company? Do you think it will give you more control of the pharmaceutical industry? How will it change your control of the market? 

Human brain is great at learning and analyzing. However, our great ability becomes limited when we are faced with large quantities of data. The advantage of AI is that it can process and learn from massive amounts of data then narrow down the pool of possible results for the human. 

AI will be inevitably used in the drug development process, now, almost all big pharmas are starting to implement AI in their pipeline.  AI companies will certainly play a more and more important role and have more control in the drug development process. 

In the future, individual scientists or virtually anyone can be drug developers. You do not need to work for big pharmas to develop drugs. By using the AI platform, anyone can find drugs using their computer.  “It’s like drug mining”.

Do you think AI will radically change the drug development process? Do  you think it is a disruptive or incremental innovation? 

Yes, the human brain can not take massive amounts of data and learn from it. For example, in 2020 there were 100, 000 papers published about Covid-19, even a world leading researcher can’t possibly read all of it. AI can help narrow down and extract the important information so it becomes much easier for humans to consume the information. AI is nothing but automation. 

I believe AI will radically change the drug development process. Let us look at Cloud Service as an example. 20 years ago there were no such services, companies or entrepreneurs needed to rent a database with private service providers to store their data. But look at the cloud service now! 

Drug development is the same. Pharma companies are starting to outsource research efforts to external labs. The next question becomes how can individuals contribute? 

I believe in the future, AI will lead to ‘”Democratized” drug discovery where anyone can be a drug hunter.   By using the online platform, individuals identify drugs, protect them with IP and sell or license them to big pharmas.  This is how I think drug development would be in the future.

Wisecube AI along with other third party developers will provide the tools to empower the scientist, but we will always still need the human in the loop.