Skip to Main Content

The days of the blockbuster drug might not be over, but they are dwindling. As markets become more crowded, pharmaceutical companies are focusing their drug development strategies on specialized populations, such as individuals with rare diseases or those with genetic subsets of chronic conditions like cancer and diabetes and cardiovascular diseases.

Identifying a specific subgroup of patients from a niche population poses challenges for many companies.


The integration of artificial intelligence (AI) and machine learning (ML) in a drug development and launch playbook offers new possibilities to find the right subpopulation(s) for product targeting as a way to ensure a successful launch.

With the growth of big data and exponential advancements in technology, we are seeing more opportunities to apply artificial intelligence and machine learning in decision-making for drug development. Indeed, we believe that AI and ML represent powerful, disruptive tools that will transform the future of drug development, empowering manufacturers with faster and better insights that will improve drug launch success.

At the same time, we recognize that the pharmaceutical industry often wrestles with understanding how to apply these technologies to support their business objectives and how best to deploy them within an already established drug development.


So, what is the transformational power of AI and ML?

Over the past decade, artificial intelligence and machine learning have become woven into our daily lives. In our homes, AI and ML power voice-recognition devices that control lights and thermostats, tell us the weather, schedule appointments, and answer questions. On-demand content providers use AI and ML to suggest new programming based on individuals’ viewing habits. Artificial intelligence and machine learning help Internet search engines target ads based on our browsing history, while they help e-commerce companies provide us with more personalized shopping experiences. Clearly, these technologies have evolved from buzzwords and hype to fundamentally transforming the way we live and work.

That same power is beginning to be used in the pharmaceutical industry to arm companies with the kinds of insights needed to improve the success of a drug launch. And companies are gaining access to these insights in less time than with traditional models and methods.

From the preclinical phase all the way through clinical development and launch, artificial intelligence and machine learning can help companies navigate the development pathway with greater guidance and understanding. They can provide insights on the right patient sub-population for a specific compound, the prescribers who are most likely to adopt that treatment once approved, and the drug’s potential performance once in the market. This kind of information is especially critical when exploring a therapeutic area in which there are limited or no treatment options available and thus no market data to use as a guide.

How does this work?

In the prelaunch phase, investigators can use artificial intelligence and machine learning to identify which patients are more likely to progress to severe disease with greater predictive accuracy than other models. This can help define patient subpopulations that might achieve greater benefits versus risks with an investigational product, which can increase the odds of success downstream.

Effective and fast market access is an essential component for a successful launch. Real-world anonymized patient-level data and machine learning models that predict a product’s potential impact on patient outcomes can be incorporated into documentation for a health technology assessment, providing additional insights that can support negotiating optimal pricing and reimbursement rates with payers prior to market launch.

Machine learning models can predict physicians’ prescribing potential by examining current prescribing behaviors across all diseases and patient mixes. With superior capabilities to handle the complexity of a vast volume of data, proprietary models can predict if a physician will prescribe a new product over a six- to 12-month time period with 80% accuracy. This capability can help drug makers optimize their promotional efforts and better target doctors during the launch period.

These aren’t just theoretical possibilities. We and our colleagues at IQVIA have developed a platform that incorporates anonymized patient longitudinal data, payer and health plan data, and data about health behaviors and characteristics of health care providers across the U.S.

Working with one pharmaceutical company, we delivered a 25% lift for the launch of a second-line indication in the cancer treatment market. Once a patient starts the treatment, keeping them on the therapy according to the regimen improves the patients’ outcome. Our algorithm identifies candidate patient profiles to improve patient persistence and compliance which are sent to nurse educators or other patient engagement programs.

Artificial intelligence and machine learning can be powerful game changers for the industry. These capabilities can power better decision-making and improve efficiencies within the entire clinical development ecosystem.

Yilian Yuan is senior vice president for data science and advanced analytics at IQVIA. José Luis Fernández is the company’s senior vice president for consulting services.