CCAB Blog – Artificial intelligence and drug development

By CCAB - July 15, 2019 2:19 pm (leave your thoughts)

Adding artificial intelligence to drug development

by Karen Ramlall
Communications Manager

It may be a difficult concept to put into words, however, artificial intelligence is all around us and is very much a part of our daily lives. Think Siri, music streaming services, self-driving cars – artificial intelligence makes these and many other services possible.

So, what exactly is artificial intelligence (AI)? According to, it is “…an area of computer science that emphasizes the creation of intelligent machines that work and react like humans. (1)

AI is a broad term which incorporates machine learning and deep learning. The human-like intelligence and decision-making ability of the machines is based on the massive amounts of data that are fed into computer systems.

To demonstrate, let’s use Siri as an example. Siri uses AI in a couple of ways to allow it to respond to our queries. AI is used to develop Siri’s speech recognition system, which is based on models that are trained using a wide range of voice samples.

Once Siri recognizes what you’ve said, AI algorithms are used to recognize the intent of what you’ve said and, subsequently, Siri is able to respond. What’s really amazing is that Siri and similar voice-controlled assistants continuously learn about their users and, eventually, are able to anticipate their needs.

As impressive as it is to have Siri respond to our everyday requests by speaking just a few words into our phone, AI can, of course, do so much more. It is currently being put to work to address pressing problems in health care.

Improving health care with AI

AI is already having a tremendous impact in health care in a number of ways.

For example (here’s where the robots come in), AI is being used to improve the precision and effectiveness of surgery using robotic instruments. One study showed that an AI analysis of pre-operative data can make surgery more precise and can lead to a 21% reduction in a patient’s hospital stay. (2)

Deep learning is also being used to assist doctors by suggesting the best course of treatment and flagging potential dangers. One company has developed software which uses AI for a tool used in nursing homes. The tool can analyze vast amounts of information, far exceeding the amount of information an individual doctor can read, and makes patient-specific care recommendations which improve outcomes. (3)

AI in drug development

Some especially exciting innovations using AI are taking place in the development of new drugs. There are efforts underway to accelerate and improve what is happening along every segment of the development pipeline – from discovery to clinical development.

Lab of the future: At Phenomic AI, the company has developed AI-based techniques to interpret large amounts of microscopy data to identify differences between cells and sort the data into categories. These types of differences are difficult to interpret by eye or even with existing computational techniques. The software aims to understand how thousands of genetic mutations and drugs affect cell health.

The Phenomic AI technique is faster and more efficient than traditional techniques because it has used deep learning to automate the processing of the large datasets. Once an experiment is completed, the results are uploaded to the cloud, automatically analyzed and, the next day, can be explored and interpreted by scientists. The founders of Phenomic AI say they are looking to build the “lab of the future” by reducing the time between data gathering and data interpretation.

Panorama of drug interactions: The lock and key analogy is a well-worn one when it comes to describing how a drug is developed and works against a disease. Very generally, it describes the formula of one protein, one disease, one drug. However, a drug does more than one thing when it is put into a complex biological system – like a cell or the human body. In fact, it can interact with upwards of 300 off-target proteins. And determining these off-site effects is the challenge that Cyclica has accepted.

The company is using a powerful combination of AI approaches which allows them to take a panoramic view of how drugs will interact with all the pieces of biology in disease development. To do this, Cyclica is using machine learning and deep learning to augment computational biophysics (applying computers to model and predict how drugs will interact with the core components of biology). By allowing scientists to get a more comprehensive view of the landscape of drug interactions, it helps them take more precise development steps and decrease drug discovery time.

Cyclica has supercharged the lock (the biological component for which the drug is being designed) and key (the drug) approach to drug development. The knowledge-based machine and deep learning approaches use a wealth of data about both the lock and key to form predictions about whether the key will work. Combined with computational biophysics, scientists can gain insight into how the key is going to fit and whether or not it will open the lock.

More than 9 million scientific papers analyzed: Antibodies are the workhorse of experiments in biological sciences but ongoing problems with the quality of antibodies can result in a range of complications, chief among them is the impact on reproducibility of experiments. There are estimates that 50% of commercial antibodies could not be used in the anticipated application, which has a significant impact on research productivity. (4)

BenchSci, a Toronto biomedical start-up, is offering a solution. The company has developed an AI platform, which uses machine learning combined with other sophisticated approaches. The platform helps researchers find reliable antibodies based on data analyzed from scientific papers.

Using biologist-trained machine learning models, the platform reviews biomedical papers to understand which antibodies have been successfully used in which experiments, allowing scientists to select the most suitable antibodies for their planned experiments. The platform has analyzed close to 10 million papers and more than 6 million antibody products.


Speed. Efficiency. Accuracy. Just a few ways AI can change and potentially transform drug development. By taking full advantage of a wealth of high-quality data, scientists may be able to develop more precise drug therapies in a shorter amount of time, which is not just transformative for drug development but for the patients who can benefit from the new therapies.


  1. Artificial intelligence. Retrieved from
  2. The 3 most valuable applications of AI in health care. (2018, April 22). Retrieved from
  3. The 5 most amazing AI advances in health care (2018, May 14). Retrieved from
  4. Michael G. Weller. Quality issues of research antibodies. Analytical Chemistry Insights.2016; 11: 21-27


Leave a Reply