AI pharmacy, above the tuyere: bubble or future?

The "Double Ten Rule" has been circulating in the pharmaceutical industry, that is, it usually takes ten years and billions of dollars for an innovative drug to be born and put on the market for treatme.

How to reduce research and development costs, improve the success rate, and shorten the research and development cycle has become the top priority of the entire pharmaceutical indust.

The emergence and continuous maturity of AI pharmaceutical technology has provided a new solution for the "cost reduction and efficiency increase" of innovative drug research and developme.

How does AI make medicine?

Although today's AI pharmaceutical technology has shown that it can bring value to new drug research and development, it is only a tool for new drug research and development, and there is still a long way to go to automatically develop new drugs or replace manual lab.

At present, among all aspects of new drug research and development, drug discovery is the most mature and most mature aspect of AI pharmaceutical applicatio.

The discovery of traditional new drugs needs to determine the target of a disease first, and researchers design and screen the most suitable molecules according to the targ.

This process is difficult and the success rate is extremely l.

New drug development flow chart

After analyzing and reading a large amount of physical and chemical data, journal literature results, clinical data and other big data, AI can predict disease targets and drug targets with the help of machine learning, deep learning, natural language processing and other technologi.

At the same time, AI can also assist researchers to generate molecules, screen molecules, predict the performance of drug candidates, such as drug absorption, metabolism, toxicity, adverse reactions, e.

Compared with traditional drug discovery, AI technology can reduce nearly 40% of the preclinical research time in the new drug development process by virtue of big data and algorith.

Recently, Insilicon announced the discovery of new drugs twice within half a year using AI technolo.

Insilicon completes AI drug discovery in 18 months

In addition to the application of AI pharmaceutical technology in the field of preclinical drug discovery, some companies have begun to explore the application of AI in the clinical stage of higher co.

By matching information such as medical records, medical literature, and pathological content actively uploaded by patients with the information of the tested drugs, AI technology helps new drug development companies find suitable test patients and improve the efficiency of recruiting patients for clinical tria.

Wang Ning, a bioinformatician from Arcus Biosciences, once said that patient response prediction is likely to become the next direction of AI pharmaceutical development, that is, AI predicts drug efficacy through specific biomarke.

At present, AI pharmacy is only used as an auxiliary tool in some aspects of new drug research and development, but its powerful computing power can save a lot of time in drug discovery, and can also save a lot of experimental cos.

Above the tuyere, the heroes rise together

The powerful advantages of AI pharmacy have attracted various capita.

In the first quarter of 2022, there were 42 financing events in the field of AI pharmaceuticals, with a cumulative financing amount exceeding US$4 billi.

Among them, there were 13 cases in China, 21 cases in the United States, and 8 cases in other countries such as Euro.

Investment and financing activities still mainly occurred in China and the United States, accounting for 80% of total financing even.

Venture capital pouring into AI/ML drug discovery companies

In terms of R&D licensing, the market was more active, with 243 collaborations signed, with up-front payments and options exceeding $1 billion, and a total announced collaboration value of $37 billi.

Among them, in January of this year alone, the amount has reached about 10 billion US dolla.

Among them, AI pharmaceutical start-ups such as Jingtai Technology and Insilico Intelligence are favored by capit.

Jingtai Technology

Founded in 2014, it is one of the earliest AI pharmaceutical start-ups in Chi.

It has received 6 rounds of financing, with a total financing amount of US$785 milli.

Its core technology platform is the ID4 intelligent drug research and development platfo.

By combining quantum physics, artificial intelligence and cloud computing technology, it can accurately predict various important characteristics of dru.

In addition to the early crystal forms, it now also includes activity, druggability, toxici.

and other indicators, so as to comprehensively accelerate the efficiency and success rate of drug preclinical resear.

At present, Jingtai Technology has reached cooperation with a number of pharmaceutical companies, including formal paid cooperation with 7 of the world's top ten pharmaceutical compani.

It has led dozens of drug discovery projects, most of which are "first in clas.

.

Insilicon Intelligence

Founded in 2014, it has received a total of 6 rounds of financing, with a cumulative amount of over 300 million US dolla.

The latest round of financing occurred in June 2021, with an amount of up to 255 million US dolla.

It has three core platforms involving the whole process of drug development, namely PandaOmics for target discovery, Chemistry42 for molecule generation, and InClinico for clinical trial design and predicti.

In 2021, Insilicon announced the discovery of new drugs twice within half a year, and took the lead in entering a new stage of clinical tria.

Just on April 14 this year, Insilico once again announced that the company nominated a preclinical candidate compound designed by Chemistry42 and targeting synthetic lethal target deubiquitinase (USP

In addition to AI pharmaceutical start-ups, traditional big pharmaceutical companies and Internet giants have joined the AI ​​pharmaceutical layo.

Traditional pharmaceutical companies usually invest in cooperati.

For example, Fosun Pharma and Insilicon have reached a cooperation to jointly promote the research and development of AI drugs for multiple targe.

Insilico received an initial payment and milestone payments of US$13 milli.

This is the largest advance payment received by an AI pharmaceutical company in China so f.

The cross-border entry of Internet giants has made AI pharmaceuticals even more popul.

In foreign countries, Google has already entered the market, and domestic Alibaba, Tencent, and Baidu will accelerate the deployment of AI pharmaceuticals after 2020, and Huawei has also established a medical intelligent body "EIHealt.

Not only that, ByteDance, another upstart in the Internet industry, also began to deploy AI pharmaceuticals last ye.

HUAWEI CLOUD EIHealth at the 2020 World Artificial Intelligence Health Cloud Summit

The entry of Internet companies representing the top AI level into AI pharmaceuticals means that it may bring revolutionary and subversive changes to the indust.

How to break the data dilemma

Unlike traditional pharmaceutical technologies, AI pharmaceuticals rely heavily on big data - all AI technologies need to be trained and learned based on massive amounts of da.

But the biggest difficulty and challenge at present is the lack of high-quality da.

Different from the application of AI in the field of image recognition, the amount of image data is large and the acquisition is relatively simple, but the total amount of data required by AI pharmaceuticals is small, and most pharmaceutical companies are reluctant to share data for the purpose of confidentiali.

At present, most of the data sources of AI pharmaceutical companies come from public information, such as published medical literature, public target libraries, public data from pharmaceutical companies, scientific research institutions or colleges, e.

, but the most important high-quality data comes from pharmaceutica.

Enterprises, this part of the data is not easy to obta.

At the same time, these data also need to be organized into a format readable by the AI ​​platform, which takes a lot of time and manpow.

In addition to data, algorithms are another k.

Now there are many kinds of algorithms in AI companies, which one is really effective? This also requires a lot of experiments to veri.

The more streamlined the algorithm, the more precise the compound produced, and the higher the success ra.

The current situation is that traditional pharmaceutical companies have a lot of data, but lack technology in AI algorithms; AI pharmaceutical start-ups and Internet giants have strong AI algorithm and AI development experience, but lack data support and no pharmaceutical experien.

Therefore, cooperation between pharmaceutical companies and AI pharmaceutical startups and Internet giants may become an option in the futu.

In January last year, British AI pharmaceutical startup Exscientia announced that its AI-designed immuno-oncology molecule EXS21546 entered human clinical tria.

At present, Exscientia has reached cooperation with well-known pharmaceutical companies such as Bayer, Sanofi, and GlaxoSmithKli.

Insilicon has also reached cooperation with many first-class biopharmaceutical companies such as Pfizer, Astellas, Johnson & Johnson's Janssen Pharmaceuticals, and Taisho Pharmaceutica.

In this model, AI pharmaceutical companies are more of a CRO-like ro.

However, in the context of the sharp decline in the profits of innovative drugs, how much are pharmaceutical companies willing to pay for AI pharmaceuticals?

References

Prospectus of Rico B.

Source: official website

"What should the biopharmaceutical industry expect when AI pharmaceuticals enter the clinic? ", WuXi AppTec, 2022-02-23

"The total amount is 26 billion yuan! 2021 Global AI Pharma Financing Inventory", Zhiyao Bureau, 2022-01-27

"Duan Hongliang: Current Situation, Technology and Challenges of AI Pharmaceuticals", Zhiyaobang, 2021-02-08