AI Pharma, a gimmick or the future?

Gimmick, or the future? This is a question that everyone often asks in the face of AI pharmaceuticals, but it is not a problem faced by AI pharmaceutica.

The DeepMind team hit the industry with an article published by the AlphaFold 2 algorithm in the journal Nature, which reported 350,000 predicted macromolecular protein structures, theoretically containing 95% of the protein population of huma.

But how far is the future? The answer given by the scientists is that they are down-to-earth and strive for the day and night, and achievements cannot cover up proble.

On May 19th, the live broadcast room of [Friends of Rubik's Cube] specially invited .

Feng Jianxin: As a practitioner with an IT background, looking at the results of AlphaFold 2 from an IT perspective, a protein is the executor of all functions in an organism, and its structure determines its function, breaking the steps that can only be found through experimen.

But whether these protein structures are accurate, it still needs to be verified by experimental means, and besides humans, there are a large number of other macromolecules that also need to be analyz.

Chang Shan: AlphaFold 2 has achieved such a breakthrough in structure prediction, which is beneficial to our drug screening, because the structure of the receptor target is very important in drug screening based on receptor targe.

The most important thing is the performance of proteins in different environmen.

Ren Zhipan: The article published in Nature by DeepMind's AlphaFold 2 is a very important breakthrough that can bring practical value in the preclinical sta.

For pharmaceutical companies, it is more important to determine the function of more proteins than to determine more protein structur.

Break through the data bottleneck and jointly build a research platform

Chang Shan: There are about hundreds of millions of chemical molecules in the international mainstream structure libra.

Ren Zhipan: For medical AI, which is expected to reverse Moore's law (Eroom's law), which is the declining clinical success rate of new drugs, data is cruci.

We only use public life science data to establish accurate aging models and disease models, and then use the models to directly predict the results of the world's first large-scale, public, and prospective clinical tria.

Feng Jianxin: Through the communication with some industry exper.

As AI technology plays an increasingly important role in the industry, the bottleneck of big data is reveal.

Today we can see that there are indeed some government-level efforts to promote the construction of data platforms, such as the medical big data platform built by Fujian Medical University in Fuji.

Combined with the data of the hospital, I started to try to do something, but a commercial platform at the commercial lev.

Who will lead and promote this matter, at least we have not yet seen a relatively mature exploration plan, and we look forward to making breakthroughs in this area as soon as possib.

Ecologicalization of production, education and research to create opportunities for overtaking

Ren Zhipan: Innovative drugs and the Internet are highly similar in nature (emphasis on innovation, emphasis on user-end data, and emphasis on product iteratio.

From therapeutic new drugs for a single aging-related disease to preventive new drugs for all aging-related diseases, it is the industry trend of global innovative drugs from a single vertical field to an aggregated platfo.

AI, like gene therapy, nucleic acid drugs, and gene editing, is one of many new technologi.

Pharmaceutical companies will naturally follow a mature set of criteria for evaluating other new technologies, which is clinical validati.

From the perspective of clinical experiments, if new targets are discovered by means of AI, new drugs are systematically developed in batches, and the success rate of unmet clinical needs is improved, then a complete clinical verification has been complet.

Using the key clinical trials of more than 300 innovative drugs around the world every year, not only can it evaluate the AlphaFold 2 like the CASP challenge (Critical Assessment of Techniques for Protein Structure Prediction), but also accurately evaluate the clinical efficacy and safety of new drugs predicted by medical AI in advan.

It can reduce the cost of clinical validation from billions of dollars to millions of dollars, and the time of clinical validation from 10 years to 1 ye.

In this way, AI will usher in a real blowo.

Therefore, we need to use the advantages of our real Internet to speed up the iteration speed, reduce the iteration cost, and obtain the speed and efficiency of the Internet, which will truly solve the biggest bottleneck for traditional pharmaceutical companies to embrace .

Chang Shan: Most AI companies are still losing mon.

Traditional pharmaceutical companies are basically in a wait-and-see sta.

Because Internet companies have some technical reserves and sufficient funds, they are more enthusiastic and more willing to make some investment in this ar.

However, to drive the leapfrog development of China's pharmaceutical industry, traditional pharmaceutical companies still need to participa.

In addition to the conservative attitude, because AI is currently mainly used in the early stages of drug discovery, but this part is the smallest cost for pharmaceutical companies, the development of the industry still has a long way to .

How does AI technology participate in drugs? Accelerating the whole process and igniting the enthusiasm of traditional pharmaceutical companies is one of the main issues we are facing at this sta.

Feng Jianxin: AI Pharma has been a very hot track in recent years, which shows that everyone is very optimistic about this matter and agrees with this directi.

Therefore, in terms of the general direction, everyone has no doubts and is clea.

But this also gives people some unrealistic expectatio.

Today's AI pharmacy is still in the early stage of developmen.

It has not fully entered the key clinical stage; another big problem i.

Was this process discovered through some serendipitous discovery, or was it a reproducible process? No verification toda.

However, from the perspective of the overall environment, China is currently very favorable to the development of AI pharmaceutica.

On the one hand, a large amount of private capital has poured in, so this industry is not short of mon.

In addition, from the policy level, the state is now encouraging and guiding innovative drugs, and the drug discovery process of innovative drugs is actually heavily dependent on AI technolo.

We have every opportunity to achieve this leap-forward development in the field of AIDD and catch up in the near futu.

Even beyond, but the time period we need to have the correct psychological expectatio.