We've updated our Privacy Policy to make it clearer how we use your personal data. We use cookies to provide you with a better experience. You can read our Cookie Policy here.

Advertisement

How Machine Learning Can Power Drug Discovery

Listen with
Speechify
0:00
Register for free to listen to this article
Thank you. Listen to this article using the player above.

Want to listen to this article for FREE?

Complete the form below to unlock access to ALL audio articles.

Read time: 3 minutes

A recent assessment of the success of the drug development process put the percentage of phase I programs that make it to approval at just 13.8%. That’s not great, and while that figure is not as bad as some earlier estimates, nonetheless points to a crisis of attrition within the industry. Netramark, a Toronto-based start-up, claim they have a solution to that crisis. Leveraging various machine-learning based methods, Netramark aim to give new life to failing and failed drugs by revealing specific sub-populations that treatments may prove successful in, whilst reducing clinical trial size to enable savings for industry. It’s an exciting idea, and we recently spoke to Netramark founder Dr. Joseph Geraci to discover more about the company's technology, and how machine learning and quantum computing may make drug development easier for clinicians, scientists, and patients.

Laura Mason (LM): What is NetraMark's proprietary technology and how can it be used to identify potential drug targets?

Joseph Geraci (JG): NetraMark is an AI company with dedicated solutions for pharmaceutical companies with a unique machine learning platform for: 

Drug development: stratification, personalization and placebo response prediction

Drug resurrection and repurposing

Drug discovery: novel targets and molecules via classical and quantum computation. 


Our ability to understand a disease through various genetic signatures of its constituent patient population puts us in the unique position of being able to target specific sub populations. By identifying which specific proteins we should target to produce a therapeutic effect, we are helping to usher in personalized medicine. Once this is done we can discover molecules that could be used to heal or treat very serious complex diseases such as certain cancers and Alzheimer's disease. 

LM:  What sets you aside from other companies using machine learning?

JG:
We use many methods from machine learning. We created a new paradigm of machine learning based on topology, dynamical systems, random matrices, and self-organization that is capable of "seeing" into complex patient populations. Our methods can also learn from small data sets - which are typical of a clinical trial process. In fact, with our methods we can decrease the size of a clinical trial, instead of making it bigger as deep learning-based methods demand! This translates into deep savings. Further, we provide a way for humans and machines to work together to discover what is driving disease through an augmented intelligence driven process. 

LM: What are NetraMaps and how do these benefit clinicians, scientists, and patients?

JG:
NetraMaps are the vehicle through which human experts like doctors and scientists can see patient populations in a powerful new way. They are able to identify certain patients and the way they relate to each other according to different factors like personality, disease progress, genetic makeup and various other ways. This allows these experts to design treatments that are tailored and thus much more likely to be safe and effective for patients. 

LM: What advantages does augmented intelligence provide?

JG:
Machines haven’t yet progressed to the point of being able to unilaterally make decisions and perform tasks that only human intelligence is capable of, especially in areas where life changing decisions are made. Augmented intelligence is the solution that’s available to us now, and a modest and authentic way forward is to build systems that allow collaboration between humans and machines. Human creativity combined with the insights that massively directed computation provide can have a significant impact on patient care.  

Ruairi Mackenzie (RM): Can you outline how quantum computing comes into play for NetraMark?

JG:
Quantum computing is still in its infancy, but progress has been made by using near term devices to aid in machine learning. The way this works is by treating current machine learning solutions from Rigetti, D-Wave, and IBM as oracles that our classical computers can call on for help. There are certain types of machine learning that can be exposed to these quantum oracles to take advantage of a certain type of heavy computational lifting that the quantum-based computers can begin to provide help with. These will be used in two ways for NetraMark: 

1)    We are going to use these to understand what variables are driving patient populations better – this can lead to superior treatment options in the future

2)    Drug design – molecules are inherently quantum mechanical entities and thus the idea that quantum computers can help in their discovery and design has been with us for over a decade.  

RM: On that note, can you tell us anything about your collaboration with D-Wave?

JG: 
D-Wave and NetaMark are working together on the above two points. We wrote a paper that used the D-Wave machine to learn how to classify lung cancer patients. This paper will be available over the next two weeks. 

RM: Your presentation at CNS touched on the mathematics powering Netramark’s technology, which may be somewhat alien to biomedical scientists. Is there a risk of a knowledge gap being created between the tools processing data and the scientists analyzing that data?

JG:
We’ve found that the scientists who are interested in the medical consequences of using machine learning aren’t overly concerned with how the machine is working but they indeed want to feel comfortable with how decisions are being arrived at. However, we work collaboratively with researchers to be that bridge, so to speak. We translate the process into a comprehensible format so they understand how we’re going from A to B.  

Joseph Geraci was speaking to Laura Mason and Ruairi J Mackenzie, Science Writers for Technology Networks