Drug research and development has always been a high investment, high-risk, and long-term field. A new drug takes an average of 10-15 years from research and development to market, costing over 2.6 billion US dollars, and the failure rate during clinical trials is as high as 90%. Artificial intelligence is fundamentally changing this landscape, bringing unprecedented efficiency improvements to the pharmaceutical industry.
At the forefront of drug discovery, AI molecular screening technology has demonstrated astonishing potential. Traditional methods require screening millions of compounds one by one, taking months or even years. Virtual screening systems based on deep learning, such as DeepMind's AlphaFold series and Insilico Medicine's PandaOmics platform, can complete computational screening of billions of compounds within a few days, reducing the number of candidate molecules from millions to dozens. In 2024, the anti fibrotic drug discovered by Yingsi Intelligence based on AI platform has entered phase II clinical trials, becoming the world's first drug to enter the clinical stage completely discovered and designed by AI.
Generative AI has opened a Xintiandi in the field of molecular design. Similar to ChatGPT generating text, AI molecular generation models can "design" entirely new candidate molecular structures. These models learn the structure and activity data of millions of known molecules, and are able to generate novel molecular skeletons with specific pharmacological properties and low toxicity. The deep learning model developed by MIT's research team can design candidate molecules targeting specific targets within hours, while traditional methods require several months.
AI also plays an important role in optimizing clinical trials. Patient recruitment is one of the most time-consuming processes in clinical trials - about 80% of clinical trials are delayed due to insufficient recruitment. AI algorithms can analyze electronic health records (EHR) and genomic data, automatically identify patients who meet the experimental conditions, and improve recruitment efficiency by 3-5 times. In addition, AI can predict which patients are more likely to respond to specific treatments, thereby optimizing trial design and reducing the required sample size.
China's AI pharmaceutical industry is also rapidly catching up. By the end of 2025, there are over 80 AI pharmaceutical startups in China, and several leading pharmaceutical companies such as Hengrui Pharmaceutical and Shiyao Group have established internal AI drug research and development platforms. The National Medical Products Administration has also released the "Guiding Principles for Artificial Intelligence Assisted Drug Development Technology", providing a policy framework for the standardized development of AI pharmaceuticals.
In the future, with the integration of multimodal AI and synthetic biology, we are expected to see "digital drugs" designed entirely by AI - from target discovery to molecular design, from toxicology prediction to clinical trial protocols, the entire process is driven by AI. This is not only a technological advancement, but also a revolutionary leap in the cause of human health.
[Reference source] This article comprehensively summarizes the research and development progress and pharmaceutical industry public information released by companies such as British Silicon Intelligence and DeepMind. <｜end▁of▁thinking｜>