It takes many years to successfully discover and develop a new drug and Artificial Intelligence (AI) and Machine Learning (ML) hold promise to help generate therapies in a shorter timeframe by allowing scientists to conduct research more efficiently leveraging analytics, data, and technology to develop new concepts. However, this will require a change in mindset and drug development operations.
 
By definition, AI and ML are a branch of computer science, statistics, and engineering that uses algorithms or models to perform tasks and exhibit behaviors such as learning, making decisions, and making predictions. ML is considered a subset of AI that allows models to be developed by training algorithms through analysis of data, without models being explicitly programmed.
 
AI/ML will likely play a critical role in drug development, and the FDA plans to develop and adopt a flexible risk-based regulatory framework. The FDA has seen a significant increase in the number of drug application submissions using AI/ML components over the past few years, with more than 100 submissions reported in 2021.¹ These submissions range from drug discovery and clinical research to postmarket safety surveillance and advanced pharmaceutical manufacturing. 
 
With the growth in AI and its promise to improve drug development, life sciences companies must ensure there’s a proper foundation in place before adopting AI/ML into their processes and clinical trials. Identifying the right data and learning data sets is extremely important when it comes to leveraging AI. Raj Indupuri, CEO & Co-founder of eClinical Solutions, explains why people, processes, technology, and data need to work hand-in-hand to increase operational efficiencies. –KB 
 
Contract Pharma: What foundations are needed to transition into Artificial Intelligence successfully?
 
Raj Indupuri: Embracing and implementing AI properly requires modern infrastructure to enable advanced technologies. It requires robust data infrastructure to handle any data, whether it is structured, semi-structured, or unstructured. Some examples include labs, imaging, medical writing, scientific literature, or protocols. The real power of AI is the ability to harness data and extract insights for complex use cases that are generally hard for humans to process and glean information. It’s important to identify the right use case to start, ensure you have the right data and talent, and take small steps before applying it at a large scale when considering AI initiatives. There are many exciting use cases for AI/ML in clinical research, from identifying new drug targets, improving trial design, predicting disease progression, clinical operations optimization, and data management optimization, including data review automation and data mapping automation. So to realize the vision of AI it’s imperative to ensure one starts with the right use case along with data and infrastructure to realize the true potential of what AI can bring.
 
Contract Pharma: What are the common pitfalls to avoid when starting an AI initiative?
 
Raj Indupuri: With the right use case, data, and infrastructure, ensuring you have the right team and talent with the necessary knowledge and skills to develop, deploy and manage ML models is pertinent. Stakeholders and employees must be trained to use AI-driven tools to augment actions and decisions, troubleshoot simple problems, and recognize when the AI algorithm is underperforming. We must not discount the fact that it requires a significant change for people to adapt the processes to utilize AI. It’s also essential for users to build trust in these models, and this can be achieved by providing explainability and additional information on the model’s inner workings. This will ensure that errors can be discovered and biases can be managed, preventing faulty operations.
 
Contract Pharma: How can drug developers recognize when an AI algorithm is underperforming, and how does governance play a role in this?
 
Raj Indupuri: Understanding and improving AI model performance go hand in hand. In many companies, teams are devoted to testing and verifying AI model performance, tasked with reviewing predictions and verifying then labeling whether those predictions were accurate. This ground truth labeling ensures the human-in-the-loop and is critical to measuring and monitoring performance to create a feedback loop mechanism for models to learn and adapt continuously.
  
Also, governance is key to success and higher-performing models. How do you manage all these datasets, and who can access what? How do you ensure and communicate interpretability and explainability to ensure that these models are providing the intended results; what is the governance around testing and verifying that the outputs of these models are high-performing? You need to answer these questions and establish processes to deliver high-performing models.

1.     https://www.fda.gov/science-research/science-and-research-special-topics/artificial-intelligence-and-machine-learning-aiml-drug-development