Compared with traditional methods, artificial intelligence (AI) is advantageous because it can help discover new molecular compounds or drug targets faster, thus speeding up the drug development process. At the same time, it can more accurately predict the subsequent experimental results of new drugs, thereby increasing the success rate of each stage in the drug development process as much as possible. From a corporate perspective, the use of AI technology can enable large pharmaceutical companies and biotechnology companies to simplify drug development endeavors, and ultimately significantly reduce drug costs and development time.

In the past few years, large pharmaceutical companies are collaborating with AI startups to develop new drugs and treatments, with a belief that AI is not only a tool for compound discovery, but also a more versatile tool for promoting biological research, discovering new biological targets, and developing new disease models.

To sum up, AI has been applied to almost all processes and stages of drug development, mainly in the following aspects:

Target identification
Target identification is a key step in drug development and one of the most complicated steps. At present, most of the known drug targets are proteins. Using machine learning methods, extracting features from the original protein information, and constructing accurate and stable models for function inference, prediction and classification have become important methods for target research. AI can help to extract genomics, proteomics, metabolomics and other omics data from patient samples and massive biomedical data, and use deep learning to analyze the difference between non-disease and disease states, and it can also be used to discover proteins which may cause diseases.

Phenotype-based drug discovery
In the past thirty years, target-based drug discovery has been the main method of drug discovery. In recent years, phenotypic-based drug discovery (direct use of biological systems for new drug screening) has attracted attention. Machine learning can associate cell phenotypes with the mode of action of compounds in phenotypic screening, and obtain clusters of targets, signaling pathways, or genetic disease associations. The powerful image processing capabilities of AI can integrate all the morphological features of biological systems, systematically study the potential modes of action and signal pathways of drugs, and expand the biological understanding of diseases.

Molecular generation
Machine learning methods can generate new small molecules. AI can learn the laws of compound molecular structure and durability by studying a large number of compounds or drug molecules, and then generate many compounds that have never existed in nature as candidate drug molecules according to these laws, and effectively construct a certain scale and high quality. Molecular library.

Chemical reaction design
One of the areas where AI is currently making progress is the modeling and prediction of chemical reactions and synthetic routes. AI can map molecular structure into a form that can be processed by machine learning algorithms, form multiple synthetic routes based on the structure of known compounds, and recommend the best synthetic route. Conversely, in the case of a given reactant, deep learning and transfer learning can predict the result of a chemical reaction. AI can also be used to explore new chemical reactions.

Compound screening
AI can model the relationship between the chemical structure and biological activity of a compound, and predict the mechanism of action of the compound. A typical example is that MIT researchers discovered new antibiotics based on deep learning. The researchers trained a deep neural network that can predict molecules with antibacterial activity, screened more than 100 million compounds within a few days, ranked the compounds based on the model’s prediction scores, and finally determined the structure of 8 antibiotics that were hugely different from the known antibiotics.

ADMET properties prediction
Unsatisfactory pharmacokinetic properties are one of the main reasons for the failure of drug development in the clinical research phase. Deep learning can automatically identify the relevant characteristics of the compound, evaluate the hidden relationships and trends between multiple ADMET parameters in the data set, and predict the cell permeability and solubility of the compound.

Clinical trials
The stage where the most funds are invested in the development of new drugs is the clinical trial stage. AI has potential for application in clinical trial design, management, and patient recruitment. Natural language processing technology can extract information from various structured and unstructured data types to find subjects who meet the criteria for entry into clinical trials; it can also be used to correlate various large data sets to find potential relationships between variables, and meanwhile, improving the matching of patients and trials.

Pharmacovigilance
AI will have an impact on traditional pharmacovigilance. With stricter regulatory requirements and increased patient safety awareness, the workload and cost of pharmacovigilance have greatly increased. AI can automate the entire process of adverse drug reactions from receipt to report, optimize pharmacovigilance work and reduce costs. Based on the AI system, it is also possible to carry out drug risk assessment through predictive capabilities.

In addition, the application of AI in drug development also includes the prediction of physical and chemical properties, drug redirection, and application in formulation development.

Author's Bio: 

Protheragen MedAI is missioned to solve difficulties and challenges faced by pharmaceutical industry and accelerate drug discovery process by applying AI into processes and attempts like personalized healthcare and various other medical applications. To this end, a few AI-powered drug discovery platforms have been developed, including CADD Platform, AIDD Platform, and Experimental Validation Platform. All these platforms are conducive to the drug R&D projects of pharmaceutical companies. The good news is that Protheragen MedAI has successfully accomplished a few projects in the past few years, which helped pharmaceutical companies simplify R&D steps and at the same time helped streamline R&D costs and other input.