Sponsored Insight
When ChatGPT launched less than two years ago, AI came into popular consciousness in a way that set the world abuzz. But as is often the case with advances in biotechnology, health care has been using artificial intelligence for years to detect disease earlier and interpret diagnostics.
As AI in health care continues to evolve and mature, recent advancements are transforming oncology and the way researchers design and develop new therapeutics, run clinical trials, bring new medicines to market, and identify and treat cancer.
At the 2024 STAT Breakthrough Summit East, Mohit Manrao, Senior Vice President, US Oncology, AstraZeneca, talked about how AI is speeding up R&D and clinical trials, identifying new, life-saving cancer treatments, and detecting cancers that may have been missed by more routine methods.
Accelerating R&D: “From 3 months to 3 days”
Through its cancer trials, AstraZeneca gathers consented patient data — including clinical data, imaging data, and multi-omic data — from more than 100,000 patients. Using AI and machine learning algorithms on the data set in addition to real-world data, they’re able to understand and create novel hypotheses of drug discovery and more accurately identify targets.
By leveraging knowledge graphs, scientists harness these vast networks of data to identify and narrow down the large number of drug chemical compounds and molecules that could potentially be used as treatments. They then select the right drugs that could work on the targets, as well as those that will be less toxic and more effective — at twice the speed of traditional methods. “With the amount of data and the machine learning algorithms available, AI not only unlocks doors to doing much more, but much faster,” said Manrao.
AstraZeneca is also leveraging AI to accelerate R&D for antibody-drug conjugates (ADCs). The company currently has one ADC in the market, six in the pipeline, and has 70% of its small molecule pipelines that go through the same narrowing-down process. “We are able to understand the antibodies that could be the targets and what could be done in three months is getting done in a matter of three weeks or three days,” said Manrao.
Improving clinical trial recruitment
Clinical trials for cancer drugs generally take 30-40% longer than for other drug trials, but AI is poised to make the process more efficient. AstraZeneca employs AI to craft diverse trial designs, leveraging real-world evidence to inform decisions regarding treatment protocols, patient selection criteria, and outcome measures.
Research teams can very quickly understand the potential cost of the trial, the number of sites needed, and the feasibility, beyond what’s possible with traditional assessment methods. “This can be accelerated for a good, or better, design of a trial and a higher likelihood of success, but also at a faster pace,” said Manrao.
AstraZeneca is also using AI to improve clinical trial recruitment by identifying diverse patient populations, and leveraging AI with electronic medical records (EMRs) to identify patients who could be eligible for a trial and should be recruited. “A wealth of data comes from this: analyzing and making predictions faster, helping to understand when the studies can read out, and where the gaps are so we can change them,” said Manrao.
Advancing health equity
Since AI is only as good as the data that feeds it, ensuring equity and representative populations in clinical trials is critical. AstraZeneca focuses its efforts on designing patient-centric clinical trials by understanding unmet needs, the issues that matter most to patients, and allowing them to provide input about clinical trial design.
“It’s absolutely critical that equity is not an afterthought. We need to ensure that we understand the determinants of health, whether they be genetic, social, environmental, or behavioral — and address the barriers that they all bring across the patient pathway from early screening to detection, diagnosis, treatment, survivorship, and care,” said Manrao.
By identifying target geographic areas or populations with higher incidences of certain diseases through AI, AstraZeneca can structure more effective clinical trials.
AI can also help reduce disparities among underserved populations by identifying patients who are undiagnosed — or who might be diagnosed later down the line and need follow-up care.
For example, AI can be leveraged to gather data on pollution incidents, behavioral elements, tobacco use, and mortality to identify high rates of lung cancer within specific counties. Using x-ray images that are already in the system, AI can identify patients who need to return for a scan because the technology suggests there is something that was missed by the naked eye that could be cancer. On this front, AstraZeneca is leveraging partnerships with Tempus, to diagnose patients with lung cancer earlier, and Qure.ai, to enhance early-stage lung cancer risk identification.
AI is also addressing health care disparities by providing quality care for individuals who face care gaps. By using AI and EMRs, patients who need a specific biomarker test can be identified and prompted by a physician about their diagnostic options, for example.
AI can also educate physicians about biomarker test and guideline treatment options they should consider, or show a digital twin of the patient and what the outcomes can be. “This will transform how we can tackle disparities in care that exist today — and with the science moving at such a fast pace, it’s only going to get broader and wider,” Manrao said.
“We want to ultimately eliminate cancer as a cause of death and this is what this transformative era of science, technology, and data converging together is unlocking for us… we as an industry have to work towards that,” Manrao said.