The hope of The Human Genome Project was that it would herald a new age of precision medicine. However, the challenge turned out to be more complex and nuanced than had been imagined. Of nearly 25,000 human genes, only 2,418 have been associated with specific diseases, explaining only a small fraction of all human pathologies. In 2020, we will begin to harness the power of artificial intelligence (AI) to create new, life-saving medicine.
In the past decade, we have learned a great deal about the complexity of diseases. We now understand that biology arises not only from DNA, but also from proteins, metabolites, different cell types and the multiple interactions between all these elements. Disease happens when something goes wrong in this complex biological network of reactions. If we are to understand them, we need to look at this system as a whole and for this we will need AI.
As Hal Barron, chief scientific officer at GSK warned in 2018, the volume of data resulting from current gene-mapping experiments is becoming intractable. “It’s definitely overwhelming for any human to think about how to deal with this,” he said. Machine learning soon emerged as the only way to find patterns in the data.
GSK has now partnered with Dundee-based Exscientia, which has developed AI algorithms that can design new molecules based on pharmacological data. In April this year, it announced the discovery of a new molecule that can be used to treat chronic obstructive pulmonary disease.
Other pharma giants are also seeking to reinvent drug discovery by collaborating with AI-biotech startups. Merck, for instance, has partnered with Atomwise, a startup that applies convolutional neural networks – the same technology used in image and speech recognition – to model chemistry interactions. Its software can simulate and analyse molecular reactions and predict how they might act in the human body. Atomwise now simulates more than ten million compounds each day.
OccamzRazor, the startup that I founded in 2016, is using natural language processing to tackle Parkinson’s disease. Every day, our algorithm, built in partnership with the Stanford’s AI research lab, reads scientific reports and data, extracts scientific information and continuously maps the biological network of Parkinson’s. We recently became the first entity to map all the information that science knows about this disease. Now, we’re starting to pinpoint the cellular dysfunctions that lead to it. Soon, we will be partnering with pharma companies to develop treatments for Parkinson’s.
In 2020, we will see the first clinical evidence showing the efficacy of drugs developed with AI. These could include drugs such as BPM31510, discovered by Boston-based startup Berg for the treatment of pancreatic tumours and carcinomas, and BenevolentAI’s ongoing trial for a drug to treat sleepiness in Parkinson’s patients. These drugs will be the first of many designed and discovered by machine intelligence.
There’s plenty of hype around AI and, just as happened with The Human Genome Project, we will see some disappointments. But in the case of drug discovery, this technology has already had a significant effect. Next year, artificial intelligence will be increasingly crucial for medical breakthroughs.
Katharina Sophia Volz is the founder of OccamzRazor
More great stories from WIRED
🚙 SUVs are worse for the planet than anyone realised
⏲️ Science says we should work shorter hours in winter
🐘 The illegal trade of Siberian mammoth tusks revealed
🙈 I ditched Google for DuckDuckGo. Here’s why you should too
📧 How to use psychology to get people to answer your emails