How is AI Transforming the Drug Discovery Lifecycle?
Drug discovery has been a crucial element in combating diseases over time. However, did you know most drugs cannot pass the drug test? Because most drugs lower the secretion of Cytochrome P450 (CYP450). This set of enzymes are produced in the liver. The drug can also be toxic for humans if the secretion of these enzymes is obstructed. Hence, pharma companies rely on conventional tools to predict if a drug sample will inhibit CYP450 in patients.
However, the process is tedious. An approved drug takes year-long procedures, starting with conducting chemical analysis on animal trials and gradually moving to humans. Moreover, the accuracy of predictions is wrong almost one-third of the time. In human trials, we observe an increase in side effects. Thus, the drug companies result in loss of money, man effort, and time.
For context, the 1 trillion global pharmaceutical industry has been investing in the productivity of drug development for over two decades. The investment is growing with time but the return on that investment is decreasing. Thus, to deal with the inefficient system drug companies are adopting new technology to overcome this shortcoming.
You may be interested in Is AI-based Automation an Alternative to Primary Medical Care?
AI is at the forefront of new technologies. Though, the real question is how can an algorithm help in drug discovery? To answer this we first need to understand the techniques of AI. The AI system is not pre-programmed with an analytical route; rather the system operates on a technique called Machine Learning (ML). In this technique, we feed organized and labeled sample problems and solutions to the system. Then, the algorithm develops computational approaches for producing the same solutions. Deep Learning is a technique that works with raw, unstructured data but requires large volumes of data. To answer the question, the use case below gives us a glimpse into the capabilities of AI in drug discovery.
Bristol-Myers Squibb’s AI model successfully accomplished the business challenge. It eliminates potentially toxic drug samples. Also, it focuses on samples that have a stronger shot at making it through multiple human trials to the U.S. Food and Drug Administration approval (FDA). So, the system eliminates the incompatible information and finds patterns in data that correlate with CYP450 inhibition.
“ The program boosted the accuracy of its CYP450 predictions to 95 percent — a sixfold reduction in the failure rate compared with conventional methods.”
- Saurabh Saha, Senior Vice President of R&D and Global Head of Translational Medicine for all disease areas at Bristol Myers Squibb
Many scientists in the field think that AI in Drug Discovery is starting to live up to the hype and eventually, it will improve drug development in several ways:
By Raising the “Hit Rate”
The AI technology collects and analyzes a lot of data needed for clinical trials. Thus shortening the drug development process. Also, it raises the hit rate, which means the percentage of candidates that pass the clinical trial stage and gain regulatory approval.
By Identifying more Promising Drug Samples
Bristol-Myers Squibb’s AI system is a great example of how an AI system can quickly screen out a toxic drug sample. Subsequently, multiple human trials and drug regulatory approvals sanctioned the results of the AI system as safe drug samples.
By Speeding up the Overall Process
In the pandemic era, we are well aware of how a deadly virus can disrupt our lives. Therefore, the process of drug discovery must be rapid and effective. AI sorts and cross-references the data more quickly. Conventionally, it takes hours or even days to plan a lengthy chemical synthesis. Today, AI tools can shorten the long process to just a few minutes.
Cutting-edge AI innovations are creating opportunities to study the complexity of diseases. By generating data, models, or novel drug sample data, design or redesign drugs, running preclinical experiments, establishing biomarkers, designing and running clinical trials, and even analyzing the real-world experience. Nonetheless, transforming the drug discovery lifecycle is just the tip of the iceberg.