How AI is Revolutionizing Drug Discovery and Development

AI can considerably transform the drug development process.

Drug discovery and development have always been challenging processes because they require large budgets and extensive timelines. Consequently, the growth of pharmaceutical companies is directly dependent on their R&D budgets. However, considering the recent boom in artificial intelligence (AI), one would not be wrong to assume that AI can play a significant role in revolutionizing the drug discovery and development process.

AI offers the advantages of accelerated drug discovery timelines, improved clinical trial design, quality assurance, drug repurposing, and drug combination analysis. Therefore, AI can considerably transform the drug development process.

How is AI transforming the R&D process?

Let’s first understand how a typical pharma R&D process works: First, researchers screen patients or organisms to identify target molecules, for example, a protein that is associated with a specific disease. Then, researchers thoroughly screen molecular libraries to identify molecules that may bind to these target molecules. Next, these identified molecules are thoroughly tested to transform them into potential candidates. Finally, these candidates undergo laboratory tests, animal tests, and clinical trials before being sold.

While this has been simplified, the pharma R&D process is quite challenging and drug discovery breakthroughs are rare. In general, the pharma R&D process may take anywhere between a few weeks to decades. AI can play a significant role in changing these timelines:

Source

Identification of targets

AI can be used to identify target molecules and binding molecules. AI may be much faster at identifying these molecules than an average researcher. Machine learning algorithms can be used to predict proteins, genes, and mechanisms involved in specific diseases.

Lead optimization

AI can be used to identify potential molecules that may directly affect disease-causing organisms (e.g., bacteria, fungi, micro-organisms, etc.), proteins, genes, and mutations. AI models can also be used to understand or predict the efficacy of these molecules before actual laboratory testing.

Preclinical failure detection

AI algorithms can be used to predict the potential toxicity of formulations and molecules early in the development stage, which can help avoid late-stage failures. AI algorithms can analyze molecule structures and mechanisms to identify toxicity.

Expedited clinical study

AI models can be used to streamline the patient recruitment process. Patient data can be input into these models, which can then identify appropriate candidates for the clinical study. This can streamline the patient recruitment and selection process, ensuring the entire process happens without bias.

These are only some of the roles that AI can play in drug discovery and development. Ultimately, the role of AI in drug discovery depends on the technology chosen. Currently, there are many AI-powered technologies, but here are the most popular ones.

AI Technologies in drug discovery: Key AI-powered technologies used in drug discovery

Machine and deep learning

ML algorithms can be used to analyze datasets to predict the biological activity of compounds. DL models like convolutional neural networks (CNNs) and recurrent neural networks (RNNs) can be used to identify trends and relations between datasets.

QSAR models

Quantitative structure-activity relationship (QSAR) models can be used to predict the activity of compounds based on their structures. ML-enhanced QSAR models can be used for larger datasets with complex structures.

NLP models

Source

Natural language processing (NLP) models can be used to conduct literature surveys. These models can analyze patents, clinical trial reports and published studies to find relevant information. This allows vast the literature review process to be squeezed into a sort period, comprehensively integrating data from multiple sources.

Computational chemistry

Currently, computational chemistry is being widely used to predict structures because it saves time and allows accurate structure prediction. These algorithms allow researchers to explore molecular interactions and bond properties at the atomic level.

MD simulations

Molecular dynamics (MD) simulations are another AI-powered technology that is being increasingly used in the pharma R&D process. MD simulations allow researchers to predict the change in the behavior of molecules with time. For example, MD simulations can be used to predict the change in molecular behavior when the molecule is exposed to specific stimuli.

Here are six companies that are currently using AI-powered technologies for drug discovery breakthroughs:

Antidote provides an AI-powered platform where patients and medical researchers are matched.
Atomwise reduces the cost of drug development using their supercomputers, which predict which medicines might work using their database.
Turbine.AI identifies the best drug for a target tumor and identifies biomarkers of the tumor. It also identifies combination therapies for the tumor.
Row Analytics provides a platform that uses AI and data analytics to consider multiple genetic variants for various diseases.
Deep Genomics is using DL to decode the meaning of genomes.
Insilico Medicine aims to integrate AI into the entire drug discovery and clinical trial process.

However, despite these advances, AI is still in a nascent phase, and it still faces challenges.

Challenges facing AI in drug discovery

Data privacy and regulatory requirements

AI-powered technologies are heavily dependent on data. However, global data laws are becoming increasingly stringent, and pharmaceutical companies have to strictly adhere to these laws, which can make it challenging to rely on AI.

Poor data quality

Source

AI can only provide solutions when it has a large and high-quality dataset available. However, reliable data in drug discovery is scarce, and the available data is of poor quality. To fuel AI in drug discovery, the quality and quantity of available data need to be improved.

High costs

While AI can help save time and decrease costs during drug development, the implementation of AI technologies is not cheap. Companies need to hire specialized staff with technical expertise to train models. Often, this may be just as expensive.

No standardization

While most pharmaceutical companies follow protocols and have established data collection methods, these are not standard across different companies, countries, and regions. For example, a common survey question is “How much does X hurt?” The patients have to choose between a scale of 1 to 10. However, there are no standardized definitions of these numbers, which means that even if data is collected from all companies, it cannot be compared.

Conclusion

AI can change the drug development landscape for the better. As highlighted in the article, AI-powered technologies can expedite the expensive and time-consuming drug development process that most R&D teams have to go through. This will considerably decrease the development timeframe and costs for pharmaceutical companies.

Various AI-powered technologies can be integrated to improve drug discovery; however, this integration may create challenges because most of these technologies are still at a nascent stage and require further development. Still, as data shows, AI has a huge potential in drug development and discovery.

FAQs

Can AI be used post-drug discovery?

Yes, AI can also be integrated into monitoring systems such as smartwatches to swiftly identify potential safety concerns. Consequently, medical interventions can be made at the right time.

Can AI decrease the cost of the pharma R&D process?

Yes! AI can be used to expedite some steps in the pharma R&D process. In turn, this will decrease the manpower dedicated to completing these steps, which decreases the overall cost of the process.

Can AI be used for drug-drug interactions?

AI can be used to predict drug-drug interactions. This will allow pharmaceutical companies to include appropriate disclaimers with their products. Furthermore, HCPs can also be warned not to prescribe these products together.