- AI and deep learning will change the world through drug discovery and the eradication of disease.
- Insilico Medicine is a Baltimore-based company focusing on next-generation AI and blockchain technologies for drug discovery, biomarker development and aging research
Hollywood has been speculating on the subject via the big screen for decades. Indeed one of its most famous seemingly prophetic films opened some 34 years ago as a near-indestructible humanoid cyborg returned from 2029 to assassinate a waitress, whose unborn son would lead humanity in a war against the machines.
“Listen, and understand. That terminator is out there. It can’t be bargained with. It can’t be reasoned with. It doesn’t feel pity, or remorse, or fear. And it absolutely will not stop, ever, until you are dead.”~Kyle Reese, “The Terminator”
And though the ramifications of the unknown to our pre-computerized world scared us a bit at the time, we’ve found that The Information Age and shift from the Industrial Revolution has been far more beneficial than destructive.
One company born of the digital or computer age is using artificial intelligence and deep learning for drug discovery, biomarker development and aging research, and its founder couldn’t be more hopeful.
Want to publish your own articles on DistilINFO Publications?
Send us an email, we will get in touch with you.
Insilico Medicine is a Baltimore-based company focusing on next-generation AI and blockchain technologies for drug discovery, biomarker development and aging research. Through bioinformatics, research and development offices in six countries around the globe, 49 employees and more than $12 million in venture funding, Insilico and Founder and CEO Dr. Alex Zhavoronkov aspire to extend healthy longevity through innovative AI solutions for drug discovery, aging research and preventing and/or curing disease.
Zhavoronkov said the company’s value is mainly in its intellectual property or the molecules Insilico is creating that could eventually be sold for billions of dollars. “We are a sizable player in AI one of the top 100 AI companies in the world,” he said.
But as with most ideas that change the world think Steve Wozniak, Steve Jobs and Apple Computers—profits are rarely the focus. Zhavoronkov said his founding Insilico was a “calculated decision” based on “basic mathematics.”
Consider the quality-adjusted life year. The QALY, as it’s referred to by scientists, is a generic measure of disease burden, including both the quality and the quantity of life lived. The QALY is used in economic evaluation to determine the value for money of medical interventions, thus one QALY equates to one year in perfect health.
The US National Library of Medicine and the National Institutes of Health says the QALY calculation is simple: “the change in utility value induced by the treatment is multiplied by the duration of the treatment effect to provide the number of QALYs gained. QALYs can then be incorporated with medical costs to arrive at a final common denominator of cost/QALY. This parameter can be used to compare the cost-effectiveness of any treatment.”
“You have to look at the number of quality years of life you’re adding. If you include every individual on the planet, and you are adding one year of quality life to each of them, you’re generating 7.5 billion QALYs,” Zhavoronkov said. “That is the most altruistic decision to make. So if you go after aging, you are affecting the quality of life for everyone on the planet. The benefit of aging research generates the most benefit to society as a whole. If I can personally give one year of quality life to everyone, my life was very well spent.”
Zhavoronkov acknowledges that the battle against aging is not an easy one and will likely require the most refined tools modern science has to offer. This is why, says the biogerontology researcher, that AI may be among the most promising mechanisms scientists can apply to the cause.
Working mainly in the fields of biogerontology and regenerative medicine, Zhavoronkov co-founded Insilico Medicine in 2014. He holds two bachelor degrees from Queen’s University, a master’s degree in biotechnology from Johns Hopkins University, and a PhD in physics and mathematics from Moscow State University. He is the director of IARP—the International Aging Research Portfolio—and of the UK-based charity Biogerontology Research Foundation. He has been head of the Regenerative Medicine Laboratory at the Centre for Pediatric Hematology, Oncology and Immunology and adjunct professor at the Moscow Institute of Physics and Technology.
At Insilico, Zhavoronkov pioneered the applications of generative adversarial networks and reinforcement learning techniques for generating new molecular structures with properties to combat disease. He was the first to develop the deep-learned multi-modal predictors of age for drug and biomarker development. Zhavoronkov set up the research and development centers in the United Kingdom, Korea, Russia, Hong Kong and Taiwan and launched multiple biomarker initiatives including Young.AI. Since 2012 he has published over 90 peer-reviewed research papers and books including The Ageless Generation: How Biomedical Advances Will Transform the Global Economy.
Zhavoronkov said Insilico focuses on three areas: aging research, drug discovery and the study of disease and identification of their molecular targets.
Research about what makes us age and how to fight it as long as we can is high on Zhavoronkov’s list of priorities. “Aging research is my personal favorite, and I spend most of my time and effort to try to identify targets for aging and age-reated diseases. Currently aging is not considered to be a disease in the health care system, and so it gets less attention. I want to develop interventions that keep people in the most youthful states and maybe reverse some of the damage of aging.”
Insilico is using the latest techniques in AI for drug discovery, including generative adversarial networks or GANs.
According to Eclipse Deeplearning4j—an open-source, deep-learning project spearheaded by Skymind, GANs are deep neural net architectures comprised of two nets, pitting one against the other. GANs were introduced in a paper by Ian Goodfellow and other researchers in 2014.
According to Eclipse Deeplearning4j’s team of data scientists and deep-learning specialists, the potential of GANs “is huge, because they can learn to mimic any distribution of data. That is, GANs can be taught to create worlds eerily similar to our own in any domain: images, music, speech and prose. They are robot artists in a sense, and their output is impressive – poignant even,”
That’s exactly what Zhavoronkov is counting on. He said Insilico is able to use the technology of GANs to essentially develop two deep neural networks (DNNs) that compete with each other to learn about medicine and disease. A DNN is an artificial neural network (ANN)—inspired by our own biological neural networks or brains. Such computing systems learn and progressively improve performance on tasks by considering examples fed to them by researchers. For example, in image recognition, the computer might learn to identify images that contain cats by analyzing example images that have been manually labeled as “cat” or “no cat” and using the results to identify cats in other images. And they do this without any prior knowledge about cats. They in effect evolve their own set of relevant characteristics from the learning material that is fed to them and that they process.
But Insilico is taking deep learning a step further to combat disease. The company uses the information it gleans from these DNNs to create molecules that may not yet exist. “We train the DNN on millions of molecular structures, and then we ask those DNNs to produce new molecules with certain characteristics, such as having a certain solubility or interaction with other molecules or bio viability,” Zhavoronkov said. “We can perhaps minimize the side effects of a drug or penetrate the blood brain barrier. It’s a concept in computer science where we utilize DNNs that pretty much work like a brain, taking real data from the real world. Then this data is being passed into other layers or cortexes, and transformations occur. Just like in the brain, neurons learn certain patterns. We run data through many layers of neurons and the system learns. So you give parameters to the DNN, and it generates those molecules which will eventually comprise the drug.”
While being somewhat prophetic in its film creations of AI, Hollywood has very different goals than scientists when it comes to goodwill, Zhavoronkov said. “Many people are afraid once they see something so incredible. They think it cannot be. But we have to remember it’s still a machine that is performing a certain task. It does not have consciousness. It’s all mathematics. It can’t make conscious decisions. It acts as a classifier or a predictor. Hollywood plays on emotions. There is great potential to do good with AI. And I don’t think we will ever create a soul.”
Zhavaronkov said he believes that AI and deep learning will change the world through drug discovery and the eradication of disease. “We are trying to optimize drug discovery. Pharma is the most inefficient industry in the world. Over 90 percent of human trials fail. We want to ensure that we can either optimize Big Pharma or replace them with faster and smarter drug discovery. Pharma is relying on human intelligence. We are using systems that are outperforming humans at every level.”
As examples, Zhavaronkov said DNNs outperformed humans in image recognition in 2015 and dermatologists in classification of melanoma in 2016. He goes on to site Tesla’s autonomous car that premiered in 2016. “In such a car, we need machines to recognize another car, the road, etc., and we need them to do it very quickly. He also praised the success of the American artificial intelligence computing company NVIDIA, whose stocks jumped by some 23 percent after the company made a host of announcements at its GPU Technology Conference last year. “They set the stage for deep learning,” he said.
In studying various diseases, Insilico hopes to identify specific molecules or biomarkers that cause disease and find new drugs to stop them as well as personalized solutions for cancer and age-related diseases. “For example, we could identify specific molecules that drive cancer and then try to figure out how to change or stop them using deep learning,” Zhavaronkov said. He said Insilico is committed to transforming the pharmaceutical industry with next-generation artificial intelligence. “We are developing new tools for drug discovery and repurposing, biomarker development and pursuing novel strategies for rapid validation. Our projects combine advances in genomics, big-data analysis, deep learning and reinforcement learning.”
Insilico has so far developed a comprehensive drug discovery engine, which utilizes millions of samples and multiple data types to discover signatures of disease and identify the most promising targets for billions of molecules that already exist or can be generated
anew with the desired set of parameters.
And though Zhavaronkov admits advancements using AI won’t be as fast in health care, as scientists need to perform lengthy discovery and testing, he says the strides it has already made in other areas validates its use and provides positive expectation to applying AI and deep learning in every respect in identifying and curing disease.
Zhavaronkov said in three years he hopes to have the first molecule discovered using AI into human patients, specifically targeting rare disease. He said he is currently focusing on musculoskeletal disorders such as ALS and accelerated aging in kids through Progeria. Going forward, he hopes to take on diabetes and Parkinson’s and Alzheimer’s diseases.
He said President Donald Trump’s willingness to “slash the restraints” put on drug development by the FDA and across the government will take years off of the drug approval process for potentially life-saving treatments. “His policies are very good for patients. I think Trump really advanced the processes with the new FDA commissioner. He is really pushing the envelope in trying to deregulate the approval process and significantly de-bureaucratize the field for patient benefit.”