Why can't pharmaceutical companies look up to the current AI pharmaceutical?

The results are ethereal, too much touting AI, lack of pharmaceutical business logic, and now some phenomena of AI new drug research and development make pharmaceutical companies and pharmaceutical investors "unaccustomed".

”

Author | Liu Haitao

Edit | Wang Yafeng

From finding pre-clinical candidate compounds to the ensuing cooperation of pharmaceutical companies, from continuous large financing to winning tens of billions of huge orders, the siege speed of AI new drug research and development enterprises has refreshed the cognition of various pharmaceutical companies, investors and AI circles.

But at the same time of rapid growth, the doubts faced by ai new drug research and development are also multiplied. Behind the high-frequency financing, is the AI new drug research and development industry reaching maturity, or a bubble carnival boosted by capital?

Focusing on the current situation of AI new drug research and development, we recently talked with Quark Capital partner Leo Han Lianyi, Northern Lights Venture Partners Song Gaoguang, and a senior executive Zhao Xing (pseudonym) who has worked in multinational pharmaceutical companies for many years, and other pharmaceutical company leaders and investors to share their views on AI pharmaceuticals.

1

AI pharmaceuticals, heaven or earth?

According to statistics, in 2021, there were 77 financing incidents in the global AI+ pharmaceutical industry, with a cumulative financing amount of 4.564 billion US dollars, and the number of financing events and financing amounts jointly refreshed the financing record of the past years. Compared with 2020, the growth rate of financing in 2021 will reach 152%.

Is there a bubble under the prosperity? During the dialogue of "Medical and Health AI Nuggets", all three respondents expressed concerns about AI new drug research and development projects.

This concern mainly focuses on the difficulty of assessing the future value of AI new drug developers.

In the past two years, although there have been schrodinger Schrodinger, Exscientia, Relay and other AI pharmaceutical companies in foreign countries have completed listing, from the current stock price, most companies have not maintained a super high market value, and Exscientia's market value has shrunk by 50% compared with the listing.

Even with such a revolutionary breakthrough as AlphaFold2, the new drug molecules of some companies have successively announced PCC (preclinical candidate compounds), but specific to each company and each cooperation, the specific effect of AI drugs can still not be measured.

The traditional biomedical system has its own set of rules for value assessment.

Zhao Xing said: "The value of enterprises ultimately returns to products and services, which has little to do with whether ai is used. ”

Yang Kun, a veteran of pharmaceutical investment and a partner of Gaorong Capital, once told Leifeng Network: "At present, many AI research and development drugs are in the pre-clinical stage, once developed to the clinical stage, according to the valuation of new drug research and development or AI companies, there will be great differences, and even AI pharmaceutical companies may face the differentiation of heaven and landing." ”

If it is the latter, in the context of the collective decline of ai business prospects, their development will face great doubts and distrust.

In addition, as a traditional industry, biopharmaceuticals do not adapt to the concept of the Internet and the way of playing the Internet will also affect the final evaluation of the value of AI pharmaceuticals.

If the former is the former, it means that the self-research platform of AI pharmaceutical companies needs more achievements that can stand up in the pharmaceutical field, such as high-value innovative drugs and sufficiently versatile tools, and the difficulty behind this can be imagined.

2

What are the core values of AI Pharmaceuticals?

The value of the AI platform needs to rely on the cooperation of pharmaceutical companies to prove that in the dialogue process of "Medical and Health AI Nuggets", many pharmaceutical industry groups have heard dissatisfaction with AI pharmaceuticals.

Leo Han Lianyi, who has worked in the National Biotechnology Information Center for many years and is currently a partner at Quark Capital, said: "Before talking about the future of AI, pharmacists and pharmaceutical practitioners will first pay attention to what is the business logic of AI pharmaceutical companies, where the core competitiveness is, and what kind of track it ultimately affects, whether it is crystal form prediction or small molecule drug screening, how AI finds the service relationship of drug research and development is the key." ”

Taking drug research and development enterprises as an example, the business logic is to make innovative drugs, and the core competitiveness lies in the market prospects of pipelines, and the number of high-value pipelines is not enough. In the same way, CRO is able to provide professional R & D, production and other outsourcing services.

But obviously, in the eyes of many pharmaceutical companies or investors, most AI new drug research and development companies do not have similar clear logic.

Liu Wei (pseudonym) is a pharmaceutical industry analyst, she believes that in addition to publicizing the optimization ability of algorithms, AI changes the rules of the drug research and development game, but also how to change the rules of the game, which rules to change, whether it is acting in the field of large molecules or small molecules, whether it can optimize the synthesis path of compounds, or can achieve the optimization of dosage forms.

"Take AI to help pharmaceutical companies discover small molecule compounds as an example. In the view of pharmacists, those preclinical achievements are relatively weak, and their frequent work needs may not be to find new possible molecules, but to pass the molecules that can't be passed as soon as possible, the earlier they are killed, the more money they save, and finally they are dragged to the clinical stage to verify, then the loss is too much. ”

Therefore, for pharmaceutical companies, the big problem at present is that they cannot understand the business logic and core competitiveness of AI new drug research and development enterprises.

The entrepreneurs of AI Pharmaceutical also need to think, can not just follow the heat to tell stories, not to use AI to solve problems, but to solve what problems to solve, how to solve, the idea must be clear, to Party A to show out.

3

Traditional AI methods are also very good

In the traditional pharmaceutical industry, pharmaceutical companies do need a lot of suitable tools to solve problems. However, in practice, the platform provided by AI new drug research and development is very different from the needs of pharmaceutical companies.

Zhao Xing said: "For pharmaceutical companies, our goal is to tend to solve problems, that is, engineering capabilities, and we will not pay much attention to tools from current artificial intelligence companies, but also traditional companies." ”

At present, there are many tools that work well in Pfizer's drug development, such as tools that integrate existing scientific research results and reports:

When pharmacists discover a new compound, they can quickly find all the relevant reference articles and reports, provide references, and many people will be willing to use such tools.

In Zhao Xing's view, many AI new drug research and development platforms have not reached such ease of use, and there is a general phenomenon of insufficient engineering capabilities.

Taking the application of AI to target recognition and molecular generation as an example, in addition to the faster discovery of targets and compounds, pharmaceutical companies will be more concerned about whether the data behind the platform is accurate, whether the relationship between the platform and known targets can be accurately analyzed, and whether these results can be presented easily.

Behind the scenes, pharmacists are more concerned about the data quality of the AI new drug research and development platform: Behind the current AI screening results, is there a large amount of high-quality data to support, and is the coverage data credible?

Zhang Lei (pseudonym), the head of domestic R&D at a multinational pharmaceutical company, also said that AI seems to assist R&D personnel in making more comprehensive decisions. However, it is difficult for users to verify the optimality and rationality of the results given by different AI models, and the unreliability of any one model will amplify the risk of drug development failure in the later stages.

Further downstream, using AI to predict drug generation and toxicology, and USing AI to predict clinical trial results, from the current results, are not very reliable.

In addition, in the eyes of pharmacists, the new technology of AI new drugs is not so pure.

Zhao Xing said: "Everyone may see such a huge new achievement as AlphaFold2, but the new trend of AI pharmaceuticals is not one-size-fits-all, nor is it a revolutionary technology that has suddenly emerged in recent years." Perhaps in the process of drug research and development, which are AI and which are not AI, even entrepreneurs can't say clearly. ”

For example, the definition of artificial intelligence naturally includes machine learning and deep learning.

Statistics, linear regression, random forests, and neural networks in machine learning have been used for many years in the fields of modification and modification of lead compounds, peptide synthesis, and toxicity prediction in drug research and development.

Around 2002, Han lianyi worked at the National Biotechnology Information Center in the United States.

According to his recollection, many databases and drug research and development work have used support vector machines and neural networks to solve pharmaceutical problems. However, even so, these technologies only solve part of the problem, and the breakthrough is not large.

For the current pharmaceutical companies, or experts from medicine, they feel that the emergence of the current AI DD or AlphaFold is just a new tool in the original AI application, and even in many scenarios, the past thread returns, the random forest is also used, and even the effect will be better.

Zhao Xing said, "Suddenly raise a sign and talk to investors or pharmaceutical companies about which type of AI pharmaceutical companies are new and which companies are not, this logic is very reasonable." Even some AI pharmaceutical achievements in biological computing or genomic analysis, how much pure deep learning can account for and how meaningful it is, remains to be investigated. ”

4

When Internet VCs face pharmaceutical VCs

In the AI pharmaceutical outlet, there are also conflicts in the views of different investors.

Although in 2021, the number of financing events and financing amounts in the AI+ pharmaceutical industry have set a new funding record; the types of VCs behind these financings are also very distinctive.

According to statistics, there are 8 institutions participating in more than 5 transactions, namely Sequoia Capital, Wuyuan Capital, Fengrui Capital, Gaorong Capital, Source Code Capital, Baidu, Zhen Fund, and BAI Capital.

Therefore, many investors in the pharmaceutical industry believe that there is a direct relationship between the rising heat of the field and the blessing of Internet giants and the influx of Internet capital.

Song Gaoguang, partner of Northern Light Venture Capital, said: "At present, Internet capital is the largest in terms of AI pharmaceuticals, both in quantity and amount, and it is more grouped and more optimistic. But the investors who do medicine and the investors with internet or TMT backgrounds have a very different understanding of this track. ”

For investors who focus on medicine, the commercialization of AI in many scenarios is far from reaching the high expectations after that; in pharmaceutical investment, they are more willing to look at the underlying logic, rather than dancing guns and sticks above, making a lot of fancy and impractical things, and the advancement of technology should ultimately be empowered in products.

Song Gaoguang said: "In the view of VCs and investment managers who focus on drug injection, it is difficult to reflect the value of AI new drug research and development enterprises without candidate drugs or pipelines in hand. ”

For this consideration, pharmaceutical VCs are almost all waiting and trying, and drug delivery is still their main theme, not AI.

They are also not optimistic about the development trend of AI new drug research and development.

An investor from a pharmaceutical venture fund said: "The capital behind it will decide that the development goals and measurement indicators of AI new drug research and development enterprises will be closer to Internet thinking: data theory, scale, growth, and rapid pursuit of high valuation, whether such a trend will lead to a retracement of the old road of other medical AI tracks, it is difficult to say." ”

5

epilogue

Any kind of auxiliary or enabling tool can gain recognition in the industry, which is a matter of life and death.

According to the report of The Journal of Dug Discovery Today and eKay Capital, after the accumulation of AI new drug research and development in previous years, especially in 2022, the overall financing visible to the naked eye is shifting to the middle and late stages.

Even if there are still new entrepreneurs in the end, more funds and VCs have said that the center of funding in the coming year is looking at the middle and late stages of AI new drug research and development.

Although capital investment is the future, is the imagination space, contrary to the traditional story and the valuation supported by the story; but from the perspective of business reality, the results of the current AI new drug research and development are obviously not satisfied by pharmaceutical companies and the pharmaceutical community.

Perhaps, starting from this year, from the perspective of financing and industry development, AI new drug research and development enterprises and entrepreneurs have been needed to come up with not only slogans and academic breakthroughs, but also change.

For more articles in the field of AI and digitalization of "hospitals + pharmaceutical companies + medical insurance", please pay attention to Leifeng's channel "Medical and Health AI Nuggets".

END