Statistics show that less than 12 percent of all drugs entering clinical trials end up on the market, and the average cost of developing a new drug is $2.6 billion.
Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, its research and development products include information robot, editing robot, writing robot and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.
Developing new drugs is a long and inefficient task. Statistics show that less than 12 percent of all drugs entering clinical trials end up on the market, and the average cost of developing a new drug is $2.6 billion.
Drug developers have to test a variety of different compounds and chemicals, a process that costs too much time and money by making mistakes. With so many molecules to test, researchers have had to use pipetting robots to test thousands of variants at a time, then select the most effective variants for animal models or cell cultures, with the hope that some will eventually make it to human clinical trials.
With the cost of trial and error so high, drug developers are increasingly turning to computers and artificial intelligence to narrow down the range of potential drug molecules and save time and money on subsequent testing. To identify genes that encode proteins that have great potential as drug targets, they are pinning their hopes on algorithms. New algorithmic models, including one published recently in Science Translational Medicine, add a new level of complexity to narrow down the range of relevant protein, drug, and clinical data in order to better predict which genes are most likely to bind proteins and drugs.
“Drug development can fail for many reasons.” “However, one of the main causes is the failure to select the right target for the disease.” A drug may show initial promise in early studies of cells, tissues, and animal models, but these early studies are often too simple to be controlled by randomized blind trials. Scientists use these results to predict which proteins could be drug targets, but because these studies tend to be small and short, there are many factors that can cause misjudgments.
Instead of relying on these limited tests, however, Hororani’s team built a predictive model that combines genetic information, protein data structure and known drug action processes. In the end, they came up with nearly 4,500 potential drug targets, double the number of human genomes that had previously been predicted to be ready. Two clinicians then combed out 144 drugs with the right shape and chemistry that could bind to other proteins in addition to those already identified as targets. Because the drugs have already passed safety tests, that means they could be used quickly to treat other diseases. Time is money for drug developers.
Researchers estimate that about 15 to 20 percent of the cost of a new drug is spent in the discovery phase. Typically, that means hundreds of millions of dollars and three to six years of work. Now, there are hopes that AI can cut that process to a few months and drastically reduce the cost of development. There are no drugs on the market that AI systems picked out in the first place, but they’re on track.
One of Hingorani’s collaborators is Benevolent’s vice President of Biomedical Informatics. Benevolent, a UK AI company, has just signed an agreement with Janssen (a subsidiary of Johnson & Johnson) to acquire and develop drug candidates for clinical trials. They plan to begin phase IIb trials later this year. (In phase IIa, a small number of subjects are enrolled to determine the appropriate dose; IIb expands the sample size on the basis of effective group A, and clarifies the effectiveness and safety of dosage.)
Other pharmaceutical companies are also following suit. Santen signed an agreement with Palo Alto-based twoXAR last month to use twoXAR’s AI technology to identify drug candidates for glaucoma, according to Lei. This comes just weeks after Pharnext and Galapagos, two European companies, announced a collaboration to develop AI models for finding new treatments for neurodegenerative diseases.
But Derek Loewe, a longtime drug development researcher, wrote on the personal blog of Science that he was skeptical of such pure calculations. “I don’t think it’s impossible in the long run.” “But if someone told me they could predict the activity of all these compounds, THEN I would probably think that’s nonsense. I want to see more evidence before I believe.”
Companies like twoXAR are trying to build that evidence. Last fall, they teamed up with Stanford’s Asian Liver Center to screen 25,000 drug candidates for adult Liver cancer patients. They used computer software they developed to combine genetic, proteomic, pharmaceutical and clinical data to screen for 10 potential drugs.
Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.
Samuel So, director of the Asian Liver Center, was surprised by the results because several of the drugs screened by computer software matched the predictions of laboratory researchers, So he decided to test all 10 candidates. One of the most promising drugs, which kills five different types of liver cancer cells without harming healthy cells, is now being prepared for human trials. Currently, the only drug for the same cancer has taken five years to get FDA approval, and twoXAR and Stanford have been using it for only four months so far.
What’s exciting is that even small advances in an industry with such a high failure rate can move billions of dollars, not to mention the lives that could be saved. But until drugs discovered through AI systems are actually sold, the industry’s research and development model will not fundamentally change.