Could artificial intelligence help treat rare diseases?
The healthcare industry has struggled to find therapies for rare diseases. Now, artificial intelligence and machine learning offer potential breakthroughs in treatments for patients.
To date, the healthcare industry has had limited success in developing treatments for rare diseases. Reassessing their approach to drug discovery and development has opened the door to artificial intelligence (AI) and machine learning (ML). AI and ML offer augmented approaches and healthcare is already using such technology to support diagnosis and optimising patient treatment in a number of settings. “If you think of the digital data that we have access to now, and if AI and ML can be used together with human expertise to recognise patterns between disease pathophysiology, symptoms, targets, and drugs, the possibilities are enormous,” says Krishnan Nandabalan CEO of InveniAi, who seek to transform access to innovation for healthcare and other industries.
The difference between AI and ML
AI and ML are often spoken about in the same breath, but there are nuanced differences between the two. AI focuses on using computers to do jobs that are typically done by humans and accelerating the rate and accuracy with which they can be performed. ML, on the other hand, is really about developing algorithms and models to help computers identify patterns to predict outcomes.
Fewer patients with rare diseases means they are hard to research
Nandabalan believes the applications of both AI and ML can go further than diagnostics, and that they hold particular promise for the development of treatments for rare diseases. “Only about 10% of drugs that make it to clinical trials are successful. For rare diseases it’s half or even a third of that,” he says. “The reason we have less success is the fact that there are fewer patients, so it’s much harder to gain the insights you need to develop treatments.”
“Only about 10% of drugs that make it to clinical trials are successful. For rare diseases it’s half or even a third of that.”
Without these key insights, issues of risk and safety become stumbling blocks. This is where AI and ML can help to bridge the gap. ML offers the capability to take data from a whole range of sources and uncover common threads, hidden connections and other patterns that give fresh insight into possible treatments and the impact they could have on patients.
Uncovering hidden connections and insights by using AI and ML
While no drugs have come to market as a result of using AI and ML as yet, Nandabalan believes it’s only a matter of time. “As more and more data is being accumulated, AI and ML can uncover new insights,” he says.
Speaking of his company’s own research he continues: “We have already used AI and ML to uncover hidden connections between basic biological processes and pathophysiology that allowed us to identify existing drugs that can be used to treat common and rare diseases. For example, our technology identified BXCL701, a DPP8/9 and FAP inhibitor, as an immuno-oncology candidate that is expected to enter Phase Ib/II trials for treatment-emergent neuroendocrine prostate cancer (tNEPC), a rare hormone-refractory segment of prostate cancer.”
Treatment concept to approval can take 10 years
At present it takes around 10 years to take a new treatment from concept through to approval. Nandabalan believes this time could be halved with the application of AI and ML, especially when repositioning drugs. With a reduction in timescales costs would also fall, which is so often a limiting factor in the development process and certainly a major setback for rare diseases where there are already limited therapeutic options.
While it can be easy to overhype technology, many believe AI and ML are the next step in the evolution of the drug discovery and development process. Big pharmaceutical companies are investing in the technology and, with gathering momentum, all the signs suggest that big breakthroughs are just around the corner.