Your Source for Data Science

How AI is transforming drug discovery

Pharmaceutical companies need innovation as new drug approval takes a decade and time in the market is shortening. Artificial Intelligence is transforming drug discovery and development.



At the advent of Fourth Industrial revolution, we have created a very powerful brain in machines. Every Pharmaceutical corporation is looking to implement Artificial Intelligence which learns from our mistake and identifies patterns from enormous data, in a way that’s beyond human capability. 

According to Cisco, the healthcare industry is expected to be the fastest growing industry in data generation. With connected healthcare and applications such as health monitors, medicine dispensers, first-responder connectivity, and telemedicine, it will be the fastest-growing industry segment, at 30% CAGR. It also gives us a clue that organizations that possess maximum data might have an upper hand in reaping the fruits of AI.

The primary application of AI in Pharmaceutical is to address the most leading cause of global deaths – chronic diseases.

50 % of Americans are estimated to be suffering from at least one chronic condition. Cancer is the second leading cause of death in the world and responsible for 8.8 million deaths in 2015. Globally, nearly 1 in 6 deaths is due to cancer. In 2017, there will be an estimated 1,688,780 new cancer cases diagnosed and 600,920 cancer deaths in the US alone.

Drug discovery is an obvious choice for AI, as quality and timely medication can lower the number of causalities from chronic diseases. Drug discovery is an area where AI can play a major role; it can predict how molecules will behave and how likely they make a useful drug by using modern supercomputers and machine learning systems.

AI systems claim to deliver drug candidates in roughly one-quarter of the time and at one-quarter of the cost of traditional approaches.

It will not only save money on superfluous tests but also shorten the time to drug discovery. The average time for a new drug approval in the United States is 12 years and cost more than $1 billion.

On the contrary, the average window of time in which a drug remains on the market before competing products arrive has fallen to just 4 years. This gives no choice for pharmaceutical companies to try innovative ideas to fill the gap.

In my previous article, I have shared a video on AlphaGO software that beat the world’s top human player at the ancient strategy game GO. However, the new AlphaGo Zero that trained itself entirely on reinforcement learning (which is to play against itself) and thrashed the older AlphaGo by 100 games to zero.

AlphaGo Zero learned completely from scratch, with no knowledge of how humans play the game” DeepMind chief executive Demis Hassabis said at a press conference held by the scientific journal Nature in October 2017.

He also said that the company is now planning to apply an algorithm based on AlphaGo Zero to other domains with real-world applications, starting with protein folding. To build drugs against various viruses, researchers need to know how proteins fold.

DeepMind is a London-based artificial intelligence company owned by Alphabet Inc (parent company of Google). Alphabet has set its eyes on the healthcare sector. Its life-sciences arm Verily has partnered with Novartis, Sanofi and GlaxoSmithKline Plc in improving various areas of Pharmaceutical and Healthcare.

IBM is also working on the concept of game-winning strategies; their program is based on ancient Japanese board game, Shogi.

When chemists design a new drug, they not only need to design a target molecule or compound, but they also need to look at the reaction pathways to synthesize that target molecule or compound.

Providing an automated solution to reaction pathway discovery is not a trivial task, and has yet remained unsolved. So, instead has been done manually. It is a time-consuming, repetitive task that can result in sub-optimal solutions, or even failure in finding reaction pathways due to human error. This is why I am looking for a way to use artificial intelligence to help automate this process.”

  • Akihiro Kishimoto, research member at IBM Research – Ireland

You must have understood by now, why technology companies are opting for game-winning strategies of AI in drug discovery. Still, let me break it down for you.

The task of finding reaction pathways in chemistry is similar to game-winning strategies. The algorithm focuses on the best combination of moves leading to a solution and finds the most promising strategy or pathway – instead of examining all combinations, which is manually done today.

Though, the reason almost every Pharmaceutical company is exploring ways to use Artificial intelligence is beyond drug discovering.

Another application of AI can be seen in investigating biological systems to find out how a drug might affect a patient’s cells or tissues. It’s helping Drug makers to better identify patients for clinical trials and therapies that are most likely to work for them, this also improves the chances of medication to get approved by regulatory agencies such as the FDA (Food and Drug Administration).

Berg, a US-based Biotech firm is running a 7 year study known as Project survival, where the smart machines will scan the samples and genes of hundreds of patients, for molecular fingerprints, or biomarkers, that will help to measure a drug’s impact and identify patients in which such a drug is most suitable.

The company is also helping England’s national genomics project to mine DNA and health data from thousands of British citizens for potential drug targets. The project focuses on patients with rare diseases and six common cancers. Industry partners include drugmakers such as Roche, Biogen, AstraZeneca, and GlaxoSmithKline.

Let’s look at the recent initiatives of 10 Pharmaceutical giants in Artificial Intelligence:


  • Recently closed a deal with Exscientia, Scotland-based Company, which will help GSK to search for drug candidates for up to 10 disease-related targets. GSK will provide research funding and make payments of £33 million ($42.7 million) if milestones are met.
  • Partnered with Insilico Medicine to improve the drug discovery process, thought both the companies are not making any official announcement.
  • GSK has also joined hands with Verily to explore the use of electrical signals to treat diseases.
  • Partnered with Lawrence Livermore National Laboratory to leverage AI for pharmaceutical R&D.


  • Collaboration with IBM Watson Health for the purpose of improving the treatment of breast cancer patients on June 2017.
  • Partnered with Verily, invested in a $300 million fund started by venture capital firm Medicxi in a hunt for promising opportunities in the drugs industry.

Johnson & Johnson:

  • Already working with IBM Watson’s AI technology for processing high volumes of data and offers evidence-based answers to questions posed in natural language
  • Its subsidiary, Janssen Pharmaceutical has licensed a range of drug candidates for novel Alzheimer’s medicines to BenevolentAI, a UK-based artificial intelligence company. BenevolentAI will have the sole right to develop, manufacture and commercialize these novel drug candidates in all indications and in all territories. The company intended to begin late stage Phase IIb clinical trials in mid 2017.


  • Closed a deal with Exscientia for drug discovery in May 2017.
  • Sanofi started a venture with Verily last year to tackle diabetes by combining devices and medicine

Roche: Genentech, a biotech arm of Roche Group announced collaboration with GNS Healthcare, a precision medicine company in June 2017. GNS Healthcare will use machine learning and simulation AI platform to power the development of novel cancer therapies.

Pfizer: In December 2016, Pfizer and Watson health announced a collaboration where IBM Watson will help Pfizer in Drug discovery for immuno-oncology.

AstraZeneca has research collaboration with Boston-based Berg, a specialist in artificial intelligence for drug hunting. The tie-up will focus on finding and evaluating ways of treating Parkinson’s disease and other neurological disorders.

Merck & Co. teamed up with Atomwise, the company’s deep learning technology can identify compounds that will develop medicines for neurological conditions.

Takeda Pharmaceutical signed a deal with Numerate Inc, a computational drug design company which is using its AI platform to identify and deliver multiple clinical candidates for therapeutic areas of oncology, gastroenterology, and central nervous system disorders.

Merck KGaA, The German pharmaceutical company has not only developed two drugs using computer-vision software that analyze images of cells and tissues and other AI for drawing insights from public databases of genetic and chemical information. Not only that, the Merck Company also use AI to improve demand forecasts and supply chain for pharmaceutical and healthcare products. Merck KGaA is not associated with Merck & Co. and operates in the U.S. as EMD Serono Inc.

We discussed 10 Drug Makers who already started with multiple initiatives in AI to attain better profitability and productivity from Pharmaceutical R&D. There are many who would join the race soon.

However there are few things to consider here, drug discovery is a long term investment as it takes a decade to get a new drug approval.

Adoption and innovation are slow in healthcare and there are many laws, compliances and regulatory issues which need to be considered. We still do not know how much return Pharma companies will get out of their investment as it’s still too early to get the results.

A paper submitted at Cornell University argues

Generating molecules with desired chemical properties is important for drug discovery. The use of generative neural networks is promising for this task. However, from visual inspection, it often appears that generated samples lack diversity. In this paper, we quantify this internal chemical diversity, and we raise the following challenge: can a nontrivial AI model reproduce natural chemical diversity for desired molecules? To illustrate this question, we consider two generative models: a Reinforcement Learning model and the recently introduced ORGAN. Both fail at this challenge

One thing is certain that AI has a bright future in every sector and industry but a long road ahead of it.