Drug Discovery – a Real Deal

Since long time ago, medicines have been  high priced, money, research and resources go into producing  them. Studies show that after a drug is approved there is an estimated added cost of $2.9 billion dollars and more than two years is spent in drug development. The pharmaceutical industry is actively looking into AI solutions which can tackle these issues.

What Can AI Accomplish?

May be it’s too early, to get an FDA approved drug, discovered solely using AI. Drug companies are focusing into the future – pushing the envelope by looking at AI to help them find and develop new medications that are faster and more cost effective. The idea is that AI is more adept to pattern recognition.  It can sift through data and vast amounts of new and existing genetics, it can help unravel complex biological networks.  With cognitive computational hardware and deep learning systems in place, the goal is that the time and cost that goes into developing a drug will significantly decrease.  

AI in pharma is enabled by a huge medical data pool. Pharma and medical researchers have data on different drug experimentations, research papers, extensive library of synthetic and theoretical drug compounds and drug testing results, coupled with EHR, genomic profiles, medical imaging and clinical trial details.

Using in-memory cognitive computations, drug discovery and new usages of existing drugs can be accelerated. Deep learning algorithms can assimilate and correlate data; map relationships between theories and experimentations; discover patterns and compile number of hypotheses to conclude. In addition, AI can help to validate and eliminate a number of variables using criteria set to rule out evidently failing hypotheses, and generate the ones with high percentage of success rate.

AI At Work

Insilico Medicine is currently using a deep learning technique called Generative Adversarial Network (GAN) to create prototype images of cancer-like molecules, that can accelerate cancer-fighting drug discovery. GAN uses two competitive neural network that compares real data and generates new data, which is homogeneous to the real one. Following this, a discriminative model then brings the output by comparing both the newly generated data and the real data. The company has claimed to feed various forms of medical data to its network and imagined around 69 different molecules with cancer-fighting potential already. Moreover, they are using the platform to test drugs on gene expressions of different types of human cell, and classify them categorically for therapeutic usages.

Exscientia, who focuses on drug design, uses vast resources of chemical structures in pharmacology and bioassays data, patient information and research literature for their AI platform. The tools after filtering all the data (considering various medical parameters like pharmacokinetic and toxicity variables) generates a new molecule design. The new molecule design is tested further in depth, to refine it into a new optimized molecule ready for clinical trial.

The latest trend in big pharmaceutical companies like Astrazeneca, GlaxoSmithKline, Sanofi and Merck is to collaborate with ‘Drug Discovery Expert AI’ startups like Benevolence, Berg, twoXAR, Atomwise, Numerate, Exscientia and InSilico Medicine.

Some well known established AI companies like Deep Mind will also start experimenting in protein folding using algorithms based on AlphaGo Zero, as protein folding happens to be the fundamental principle in drug discovery against various viruses at large.

AI is Here Only to Enable

AI proves to be an important tool that can be used at every stage of drug discovery and development process, its data simulation and analytical powers, help to come up with revolutionary drug ideas. It will definitely boost R&D efficiency with optimized efforts, time and cost, as well as anticipate side effects before human trials, resulting into higher success rate. It is not intended  AI to replace human involvement in chemical synthesis, laboratory works, trials, regulatory approvals or actual production scene. AI will be an enabler, and not a replacement, in near future.

Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>