Technology

How Artificial Intelligence is Improving Cancer Research

biotech
Credit: Photo by National Cancer Institute on Unsplash

Cancer is one of the most complex diseases that affects humankind, making it an incredibly difficult condition to study and even more difficult to treat. In fact, cancer is so complicated that the world’s foremost experts of the early 20th century knowingly simplified cancer research -- it was the only way they could develop any usable body of scientific knowledge, full well knowing they were unable to possibly understand every single factor that influenced cancer. 

Today, though, humanity has a powerful new ally in the fight to eliminate cancer: artificial intelligence. AI and machine learning algorithms gather the data needed to understand cancer holistically and support cancer researchers by providing them with a deeper level of insight than ever available before.

Why AI is useful in cancer research and treatment

AI is helpful in nearly all facets of cancer research and treatment, contextualizing the vast troves of data required to understand cancer and providing insights to human researchers and healthcare providers that are easy to understand. 

To understand cancer (and, in turn, more effectively treat it) there are many factors to consider. These include:

  • Tumor mapping: Every tumor is unique and contains a heterogeneous cluster of cells. To destroy a tumor, you first must know what types of cells make up that tumor and in what proportions. To prevent the cancer returning after initial treatment, all these cells must be eliminated.
  • Patient genetics: Patient genetics plays a key role in cancer and cancer treatment. Depending on a patient’s hereditary background, their cancer may develop and they may respond to treatments in a particular way. Understanding patient genetics is critical to choosing the optimal treatment plan.
  • Patient medical history: A patient’s medical history, the entire profile or phenotyping of the patient from age, sex, previous diseases and treatments and responses, as well as the laboratory tests run and results recorded are all added to the overall picture of the cancer and how to treat the cancer.
  • Patient lifestyle habits: Whether a patient smokes, drinks, exercises, or eats well all factors into how cancer develops and how a patient may respond to treatment. There are dozens of environmental and lifestyle factors that can play a role.

For humans, all the factors in each of these areas combine to an order of magnitude that would take a lifetime to understand. But AI can make these calculations much more quickly while working around the clock. What once took years of research by expert teams to ferret out can now be determined in a matter of hours by machine learning algorithms that improve as they work. 

Rather than replacing human cancer researchers and healthcare professionals, AI is instead a critical support tool that frees these experts up to move their work forward more quickly. For example, by analyzing a detailed tumor map against a patient’s individual profile, then examining historical data about similar patients’ responses to certain drugs, machine learning algorithms can quickly identify the optimal available treatment plans and drug formulations for an individual patient -- this could greatly improve the survivability and quality of life of many cancer patients.

How biopharmaceutical companies can benefit from AI

Developing drugs intended for cancer treatment is a timely and costly process. First, companies must discover drugs that have strong promise as a cancer treatment, which can take significant research and development in the lab. Then, after demonstrating a baseline safety and efficacy internally, companies must begin the uncertain process of attaining U.S. Food and Drug Administration (FDA) approval. Artificial intelligence can help in both these arenas.

Drug discovery

The first challenge in getting a cancer drug to market is to identify one with actual potential in the first place. The drug discovery process is a research intensive phase that can result in a lot of trial and error. In some cases, this can amount to a lot of time and money wasted.

Machine learning algorithms can be trained to review vast databases of information and help improve the discovery process. For example, Predictive Oncology’s machine learning tools known as PeDAL and CoRE are able to analyze a wide range of information, including patient genetics, behavioral habits, environmental factors, cancer type, tumor details, and more. 

By running countless scenarios against a database of 150,000 deidentified patient drug responses maintained by Helomics, a POAI subsidiary, these algorithms are able to identify optimal drug formulations for specific types of patients with specific types of cancer. That saves researchers a lot of time in the lab and allows them to pinpoint their best candidates for development and approval sooner. 

This technology brings the patient into early drug discovery research, which is an important difference than conventional methods, such as relying on laboratory in vitro assays and animal models that don’t represent the patient. In vitro assays and animal models can be used for safety testing and some efficacy, but relying solely on screening and animal models is one of the primary reasons why drug approval, even after compound selection for cancer clinical trials is an abysmal ~7.3% from 1994 to 2020, according to the Pharmaceutical Manufacturers Research Association.

Drug development

Once a company has developed a cancer drug, they need to test it and gain approval before they can sell it on the market. The average cost of the FDA approval process for a single cancer drug is about $650 million, and there is no guarantee that investment will lead to approval. For the cancer drugs that do make it to market though, the return is massive: the average cancer drug drives $1.658 billion in revenue. 

For pharmaceutical companies, AI means a higher likelihood of attaining FDA approval and a faster time to market. When the U.S. market for cancer drugs is worth nearly $130 billion annually and only 80 cancer drugs were approved by the FDA from 2000 to 2017, the opportunity for AI to remake cancer drug discovery and development is immense.

Artificial intelligence can support drug development in a way that reduces overall costs and improves the odds a company attains approval. In many cases, AI can streamline the entire process too, saving time in addition to reducing overhead. AI can predict how cancer cells become resistant to treatment and offers recommendations for adjusting formulations accordingly. It can also be used to manage chemotherapy and support healthcare providers in developing optimal treatment plans.

When it comes to the fight to eliminate cancer, AI isn’t just another tool in the cancer researcher’s toolbox. It is an indispensable ally that has the opportunity to completely change how we treat cancer, a game changer in humanity’s fight against one of the most insidious and confounding diseases facing our species. With AI on our side, we’re now poised to win that fight.

The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.

Mel Engle

Mel Engle is CEO & Chairman of the Board at Predictive Oncology (POAI), a knowledge-driven company focused on applying artificial intelligence to personalized medicine and drug discovery. Their mission: to eliminate cancer.

Read Mel's Bio