When will AI technology help the pharmaceutical industry break "Inverted Moore's Law"?

In 2020, the first year of the COVID-19 outbreak, the US Food and Drug Administration approved only 53 new drugs. In the same year, the global pharmaceutical industry’s overall drug R&D investment reached nearly US$200 billion. This means that the average cost of each drug approved in 2020 is close to $3.8 billion. A study published that year gave a relatively conservative estimate of the cost of new drugs, arguing that although the cost of new drugs has increased dramatically over the past decade, the specific range is still between US$314 million and US$2.8 billion. The study also found that the median total research and development expenditure invested in bringing a new drug to the market is close to US$1 billion, while the average is estimated to be around US$1.3 billion. In addition, the average launch cycle of new drugs is 10 to 15 years, of which about half of the time and investment is spent on the clinical trial stage, and the remaining half of the cost is used to support preclinical compound discovery, testing and supervision. As for why each new drug costs so much and has a long cycle, the reasons include lack of clinical efficacy, lack of commercial benefits and improper strategic planning. In short, all these complex factors have turned the effectiveness of the pharmaceutical industry into a metaphysics. Many people have even become skeptics due to the high cost of launching new drugs, questioning why the pharmaceutical industry has made significant progress since its technical level and management capabilities have improved significantly. You will still be stuck in the current predicament and unable to extricate yourself.

These are the proponents of what’s known as “Inverted Moore’s Law,” in which the cost of developing new drugs has grown exponentially over the past few decades despite improvements in technology. Inverse Moore's Law states that the cost of developing a new drug doubles approximately every nine years, discounting the effects of inflation. This observation proposes a law similar to diminishing returns. According to the concept of economics, if a certain input in the production of a certain commodity is increased while all other inputs remain unchanged, the overall situation will eventually Reaching a critical point - if you continue to increase input, the corresponding output will begin to gradually decrease. The term "Inverse Moore's Law" was proposed by Dr. Jack Scannell and colleagues in "Nature·Reviews·Drug Discovery" in 2012.

Inverted Moore’s Law naturally points to the famous Moore’s Law. This conceptual observation from the 1960s found that the number of transistors on large-scale integrated circuits doubled every two years or so. Moore's Law is named after Gordon Moore, co-founder of Intel Corporation, and is his observation and summary of historical trends.

Dr. Scannell emphasized that there are four main reasons for the current predicament. First of all, regulatory agencies have increasingly higher standards for therapies; regulatory agencies are less and less able to bear risks, which increases the cost and difficulty of research and development; a money-throwing mentality, which relies on flooding resources to promote projects , it is easy to cause project overruns; and then there is brute force cracking of basic research, that is, overestimating the possibility of using crude trial and error to break through basic research problems.

Despite all the difficult factors we face, we will one day defeat the challenge of Moore's Law, and a powerful weapon that will determine the outcome of the battle is AI. The good news is that someone has already taken the first steps toward exploring this path.

Dr. Scannell and his co-scientists are calling on pharmaceutical companies to appoint a chief drug officer who will be responsible for summarizing the causes of failures at each stage of the development process and publishing the results in scientific journals. Currently, pharmaceutical companies rarely even publish failed experiments or clinical results, and most have not yet thought of appointing dedicated executives to handle the valuable information in failed cases. However, Dr. Scannell emphasized that in order to break the inverse Moore's Law, companies must first change their R&D processes. Collaboration and information sharing are of course a good starting point, but in the pharmaceutical industry, there is only one way to truly break Moore's Law - AI.

In the past few years, people have made many attempts to use AI to break the inverse Moore's Law. Now, many institutions such as Exscientia and Insilico Medicine are sprinting towards this end.

Headquartered in Oxford, Exscientia is a global pharmaceutical technology company that puts patients at the center and accelerates drug discovery through AI technology. Last year, the company announced that the first immuno-oncology molecule designed by AI had entered human clinical trials. In the project, Exsientia is collaborating with Evotec to develop A2a receptor antagonists for adult patients with advanced TB, using the former's Centaur Chemist drug discovery platform. This is not the first attempt made by Exscientia. In 2020, the company announced an obsessive-compulsive disorder treatment drug designed by AI-driven software and has entered Phase I clinical trials.

In addition, there is Schrodinger, who have developed the most advanced chemical simulation software in the pharmaceutical industry. Schrodinger recently received FDA approval to study its computer-designed treatment for non-Hodgkin's lymphoma in early-stage trials. The company's platform, based on machine learning technology, classified 8.2 billion potential compounds within 10 months and ultimately identified 78 compounds that could successfully pass preclinical experimental synthesis and screening. Now, the company has plans to launch a Phase I clinical study and begin recruiting patients with relapsed or refractory non-Hodgkin lymphoma.

Meanwhile, Utah-based Recursion Pharmaceuticals is also using AI technology to find new uses for existing drugs. Last year, Roche and Genetech established a project collaboration with Recursion to jointly explore new areas of cell biology and try to develop new treatments in the fields of neuroscience and oncology indications. Through the collaboration, the two companies will use Recursion's AI drug discovery platform to conduct comprehensive screening of new drug targets, thereby accelerating the development of small molecule drugs.

In Insilico, a leading anti-fibrosis drug candidate has also successfully completed Phase 0 clinical research and officially entered the Phase I clinical stage. The new target of this drug candidate was discovered by the Pharma.AI platform. The total time from target discovery to the launch of the first phase of the project was even less than 30 months, which has set a new record for the speed of new drug development in the pharmaceutical industry.

Don’t forget that AI technology will also play a role in brain-computer interface, deep learning, human-computer interface, machine learning and other intelligent simulation scenarios. These concepts have existed for decades. Early medical AI systems once relied heavily on clinical knowledge and logical rules provided by medical experts, but now trained supercomputers can complete these tasks on their own.

In order to break the inverse Moore's Law, data scientists and medical scientists must jointly plan achievable use cases, apply AI technology to various clinical trials, and combine AI technology with existing technologies that will replace/complement it. Contrast. In this way, AI is expected to smoothly enter the clinical trial ecosystem, effectively reducing R&D failure rates and costs while rapidly improving the industry's drug discovery and development processes. Today, almost all large pharmaceutical companies are using internal original research algorithms, cooperating with AI vendors, or directly acquiring AI vendors/technology to enrich their product portfolios and drug discovery pipelines. The partnership statements from Massive Financing and multiple pharmaceutical companies also tell us that the industry has high expectations for the application of AI tools in the drug research and development process. There have been many changes in this field. It is hoped that in the next few years, companies will be able to combine better investment strategies with advanced AI technology to break the "curse" of inverted Moore's Law in one fell swoop.

The above is the detailed content of When will AI technology help the pharmaceutical industry break "Inverted Moore's Law"?. For more information, please follow other related articles on the PHP Chinese website!