Disrupting drug delivery with AI
Drug discovery is complicated, costly, and time consuming. Finding potential new treatments is traditionally based on trial and error, which draws out the process. However, armed with artificial intelligence, medical researchers could transform drug discovery. Once fed with relevant data, algorithms can combine information from multiple sources to make predictions about the effectiveness of a newly discovered drug (otherwise known as a candidate). Taking out the guesswork means lower resource expenditure, and less reliance on lengthy clinical trials.
While AI shows huge promise in the discovery of new drugs, it seems that many scientists are still unaware of the technology’s potential. A BenchSci survey of 330 drug discovery researchers found that 41 per cent of respondents were unfamiliar with the uses of AI. That said, various HealthTech companies have made notable progress in the space and are now working with large pharma firms to disrupt drug discovery.
1) Deep Genomics
Founded in 2015, Canadian startup Deep Genomics believes that biology is too complex for humans to understand. However, by using a vast artificially intelligent library of over 1000 molecular compounds, the company wants to unlock ‘a new universe of genetic medicines’. The library, named Project Saturn, functions as a toolkit for Deep Genomics’ geneticists, biologists, chemists, clinicians and drug developers. It is able to find new drug candidates and offer predictions on how they could alter genetic diseases. Project Saturn also matches participants with relevant clinical trials.
Insilico aims to find optimal drug candidates before clinical trials so that the time and effort needed during the process is reduced. The company uses generative adversarial networks in which two artificially intelligent models compete to evaluate potential drugs and their effectiveness. Insilico has partnered with WuXi AppTec, a Chinese pharmaceutical giant with an existing laboratory infrastructure and extensive network. This means that Insilico can focus on discovery without shelling out for expensive labs.
Exscientia is a UK startup founded in Oxford in 2012. It claims to be the first company to automate drug design with an artificially intelligent platform augmented with knowledge from human ‘drug hunters’. In 2017, Exscientia announced a strategic research collaboration with Sanofi to discover and develop therapies for metabolic diseases like diabetes. Metabolic diseases are difficult to find treatments for as they often lack single targets that respond well to drugs. Exscientia’s AI platform addresses this issue by identifying and validating combinations of drug targets that could work together.
US biotech company Berg has developed an artificially intelligent platform that can identify previously unknown cancer mechanisms by modelling diseased human cells. The platform, known as Interrogative Biology, is able to identify naturally occurring molecules in cancer metabolism and, as a result, suggest drugs that could target those molecules more effectively. Instead of working on prior hypotheses, Interrogative Biology takes its cues almost entirely from data to profile an entire disease in the context of biological and patient metrics. In 2018, Berg began a phase II clinical trial for people with advanced pancreatic cancer.
BenevolentAI is an international biotech firm using an AI platform fed by research papers, clinical trial data, patents, and patient records. It functions as a search engine, combining and comparing the different relationships between data sources and producing knowledge graphs of a medical condition and the genes associated with it. By putting data into context, BenevolentAI gives scientists the best possible starting point when exploring drug candidates. In late 2017, the company released groundbreaking research into treatments for amyotrophic lateral sclerosis (ALS). Out of five potential treatments, four were shown to be promising, with one proving to reduce neurological symptoms in mice.
Nuritas is taking an especially novel approach to drug discovery. The Irish startup, founded in 2014, uses deep learning to predict whether bioactive peptides can help to alleviate human health conditions. Bioactive peptides are organic substances made up of amino acids, and can be found in plant and food sources. Instead of being classed as a pharmaceutical drug, they fall under the category of ‘nutraceuticals’. Unlike pharmaceuticals, nutraceuticals are less heavily tested and regulated. It goes without saying that the effectiveness of nutraceuticals is still in question, but Nuritas believes that hundreds of millions of years of evolution have made bioactive peptides the most potent healers.
Toronto based Cyclica aims to empower scientists with cloud based and artificially intelligent tech. While traditional methods have led to significant drug discoveries, treatments interact with multiple proteins once inside the human body which can affect the safety and efficiency of the drug. As such, Cyclica’s models are ‘drug-centric’ – they consider the proteins that the drug may interact with as well as the target itself. The startup is currently working in partnership with multinational pharma company Bayer to build advanced drug discovery programs for small molecules.
Effective drug discovery has long been held back by the costs and considerations that come with the creation of life changing treatments. The startups listed here show that there is scope for AI to disrupt the traditional process and replace guess work with data driven precision.
It’s worth noting that in most cases these young companies aren’t working alone… Many have partnered with larger businesses to tap into their resources. In addition to this collaborative approach, the combination of AI with developing areas like genomics and nutraceuticals may also play a key role in breaking down the barriers to more effective treatments.
Interested in innovation? Then please take our survey – a collaboration between D/SRUPTION and Digital Catapult on the most successful and challenging methods of innovating.