Read the article: The application of artificial intelligence in drug research and development

Artificial intelligence has flourished over the past decade, driven by revolutionary advances in computing technology, which have revolutionized the ability to collect and process vast amounts of data
.
At the same time, the cost of new drug development and the price of new drugs are also prohibitive for R&D companies and patients
.
The research and development of a new drug is a very expensive and long process, and the success rate is still very low.
According to the survey, the average research and development investment of each drug is 1.
3 billion US dollars, and the average research and development time of each non-tumor drug is 5.
9-7.
2 The research and development time of oncology drugs is as high as 13.
1 years, and only 13.
8% of the drugs that can be successfully approved are finally approved
.
Because of its predictive ability, artificial intelligence can effectively improve the success rate of drug development, which is absolutely attractive to new drug development companies
.
Artificial intelligence can increase the likelihood of successful drug development in many ways, including new target identification, drug candidate selection, prediction of chemical and physical properties of compounds, and prediction of protein structures
.
1 Identification of drug targets Targets are the basis of new drug development.
At present, we have discovered many drug targets, but compared with undiscovered drug targets, it may be only a drop in the bucket
.
The process of drug target discovery is generally time-consuming and labor-intensive, so if we can predict the target in advance through the computer, it is crucial to shorten the target discovery time
.
Kumari et al.
improved the random forest algorithm by combining bootstrap sampling and successfully differentiated drug targets from non-drug targets
.
2 Screening of active compounds In the human body, drugs can act on multiple targets at the same time, but when they act on non-targeted receptors, certain side effects may occur
.
Therefore, we need to screen compounds to screen out compounds with high biological activity on specific targets
.
Artificial intelligence can speed up our screening, which in turn can speed up the development of drugs and make products available to patients faster.

.
3 Prediction of compound properties An important factor affecting the success or failure of drug development is the selection of compounds with excellent properties, especially related properties such as bioavailability, biological activity, and toxicity
.
The clinical failure of many drugs is due to the poor physical and chemical properties of the drug, so the properties of the drug itself are crucial to whether the drug can successfully pass the clinical market
.
Therefore, in the early stage of drug development, it is necessary to conduct detailed physical and chemical properties research, and we can use artificial intelligence technology to predict drug absorption, adverse reactions, toxicity and other properties
.
For example: Newby et al.
constructed a decision tree model to predict the role of compound permeability and solubility in the oral absorption of drugs
.
4 Prediction of protein structure The biological mechanism of protein is determined by its encoded one-dimensional amino acid sequence and three-dimensional structure
.
Protein misfolding is known to be common in many diseases, including neurodegenerative diseases such as type 2 diabetes, Alzheimer's, Parkinson's, Huntington's and amyotrophic lateral sclerosis
.
Therefore, it is of great value to develop methods that can accurately predict the three-dimensional protein structure to help new drug discovery and understand protein folding diseases
.
AlphaFold, developed by DeepMind, is an artificial intelligence network that can be used to determine the 3D structure of proteins based on their amino acid sequences
.
Beck et al.
developed a deep learning-based drug-target interaction prediction model, termed molecular translator-drug target interaction (MT-DTI), for predicting binding affinity based on the chemical and amino acid sequences of target proteins, without its structural information
.
5 Precision Medicine According to statistics, 57% of failed clinical phase 3 trials were due to insufficient efficacy, and the main factors were failure to use the correct dose and to identify an appropriate target patient population
.
Therefore, precision medicine has become the focus of drug development in the pharmaceutical industry
.
And we can use artificial intelligence tools to predict doses and identify groups of patients who will benefit most from treatment
.
Atomwise, a Silicon Valley company in the United States, used IBM supercomputers to screen treatment methods in a molecular structure database and evaluated 8.
2 million candidate compounds for drug development
.
In 2015, Atomwise successfully identified two drug candidates that could control the Ebola virus in less than a day by applying artificial intelligence algorithms based on existing drug candidates
.
At present, many pharmaceutical giants have also deepened their cooperation with some Al companies, which all means the importance of artificial intelligence for pharmaceutical research and development.
The following table shows the cooperation between some pharmaceutical giants and Al companies
.
Compared with traditional drug development techniques, screening drugs through artificial intelligence methods is more efficient
.
The routine screening process may take several months, plus the cost of hundreds or hundreds of yuan per compound
.
With the help of artificial intelligence, virtual compounds can screen libraries of billions of molecules in a matter of days
.
And artificial intelligence tools can predict the physical and chemical properties of drugs in just a few days
.
However, at the current stage, the data available for artificial intelligence mining is still relatively small, and it is necessary to generate enough massive data to make better use of this technology
.
It is believed that in the near future, the market scale of drugs developed by artificial intelligence will become larger and larger, and artificial intelligence will make great efforts in the field of medical research and development! References: [1]Machine Learning and Artifcial Intelligence in Pharmaceutical Research and Development: a Review [2]Artificial intelligence to deep learning: machine intelligence approach for drug discovery[J].
Molecular Diversity, 2021:1-46.
[3] Liang Li, Deng Chenglong, Zhang Yanmin, et al.
Application and challenges of artificial intelligence in drug discovery [J].
Advances in Pharmacy, 2020, 44(1):10.
[4] Ding Boxiang, Hu Jian, Wang Jifang.
Artificial intelligence in drugs Application progress in research and development [J].
Shandong Chemical Industry, 2019, 48(22):4.
[5] Huang Fang, Yang Hongfei, Zhu Xun.
Application progress of artificial intelligence in new drug discovery [J].
Advances in Pharmacy, 2021 45 Volume 7, Pages 502-511, CA, 2021.