Many of us interact with artificial intelligence (AI) in our daily lives, from searching online and receiving personalized shopping recommendations to booking travel, making reservations and asking questions to digital assistants at home. The use of artificial intelligence has been growing significantly due to its ability to quickly perform complex tasks and solve problems without human intervention.
AI’s ability to quickly scan and synthesize large amounts of information is also being adopted by many healthcare programs, including for drug discovery and development.
The traditional process of drug discovery and developing a new drug can take more than a decade and cost billions of dollars. Most drug candidates ultimately fail to show efficacy in clinical trials, and Big Pharma often forgoes investments in early-stage development of drug candidates even before the clinical trial stage due to cost concerns. As we learn more about the complexities of human disease and human biology, it’s clear that we need more sophisticated tools.
The advantages of artificial intelligence in drug development
The use of AI in drug development has many benefits, including the ability to speed up discovery while reducing the need for extensive laboratory work and shortening the clinical trial phase.
AI is also used to search and find existing drugs that may be effective against newly discovered diseases. During the rapidly escalating and deadly Covid-19 pandemic, researchers are applying AI to evaluate existing drugs that can be used alone or in combination with other drugs to treat Covid-19 without causing serious complications.
Despite the recent interest, the use of computing and data science in basic scientific research and clinical trials has not developed as rapidly as in other fields; however, the evidence for its benefits has been growing.
Here are some potential advantages of using artificial intelligence in drug development:
- Target identification: AI can be used to sift through vast amounts of structured and unstructured data to enhance disease understanding and more accurately identify relevant drug targets.
- Lead compound generation: AI can help screen trillions of molecules to identify the most promising ones; or predict protein structures and interactions to increase the likelihood that compounds will be effective.
- Predicting efficacy and safety: AI can be used to predict absorption, distribution, metabolism, and excretion (ADME) properties in silico, or to assess whether pharmaceutical compounds that work in animal models are suitable for use in humans and predict safety issues.
- Patient selection: AI can help determine which populations are most likely to respond to treatment to develop the right inclusion and exclusion criteria and biomarkers.
- Preventing patients from dropping out of clinical trials: AI can be used to develop “patient companions”—digital personalized messages to support patients during trials to encourage them to stay engaged and seek feedback. AI can also provide tailored responses to participants’ questions.
- Find the perfect place to conduct clinical trials: Since some sites have difficulty recruiting or enrolling patients in trials, AI can be used to understand these demographic variables and see what trials are being run by competitors at those sites.
The application of artificial intelligence in clinical trials
Importantly, AI combined with computational biology and modeling can be used to form synthetic control arms of research to compare experimental drugs to standards of care, other drugs, or drug combinations. This is a relatively new and growing field known as “in silico” clinical trials, in which computer models of a specific disease form virtual cohorts to test the safety and/or efficacy of new drugs.
By using computer simulations, we can combine different parameters of disease, age and gender, for example, one at a time and in different combinations to test the potential effectiveness of a drug in thousands of virtual patients. We have the potential to see thousands of changes. We can test multiple compounds, which is not possible in a traditional clinical trial setting. Therefore, we can achieve statistically significant results with fewer unsuccessful trials.
AI could also alleviate the need for animal testing on some levels. In silico, modeling for translational medicine allows drug developers to better understand how a drug responds in humans, rather than seeing how animals react and hoping the same effects translate in humans. Almost 90 percent of drugs that show promise in animal studies are not safe or effective in humans.
The ability of artificial intelligence to reduce the cost of drug development
Finally, there is the cost factor. We know the costs involved in developing new and better treatments. By being able to design medicines faster with better efficacy and fewer side effects, drugmakers can develop and ultimately sell medicines in a more affordable way.
I believe that within 5 to 10 years of greater use of AI in drug development and testing, we will see a complete change in the economics of drug development. This is especially important for narrowly-targeted therapies and orphan disease therapies—diseases that affect a small number of patients, where investment in drug development may not be balanced against the rate of return.
The increasing use of artificial intelligence will also create opportunities for new small companies to succeed in drug development.
Further improvements and use of artificial intelligence in drug development are critical to continuing patient-centered approaches to care, improving and extending people’s lives—even those with rare but serious diseases without substantial financial incentives to invest Time and resources to develop effective new treatments in traditional ways.
Photo: metamorworks, Getty Images
