The use of artificial intelligence technology in clinical decision making is still in an early phase. But recent studies indicate that AI has the potential to help improve the way clinicians treat cancer, according to IBM Watson Health.
One study—which focused on Manipal Hospitals in India and was presented over the weekend at the American Society of Clinical Oncologists (ASCO) annual meeting in Chicago—showed that physicians on a multidisciplinary tumor board changed their treatment decisions in 13.6% of cases based on information provided by Watson for Oncology, the tech giant's AI platform for cancer care.
Of those cases where the tumor board changed their treatment decisions and made a different recommendation, 55% of the time it was because Watson for Oncology provided more up-to-date, evidence-based information on newer treatments than what the physicians knew on their own, Nathan Levitan, M.D., chief medical officer for oncology and genomics at IBM Watson Health, told FierceHealthcare.
"We all know physicians face a nearly impossible task keeping up with all the emerging literature and these physicians found the information provided by Watson for Oncology so compelling they actually changed their treatment decisions," he said.
"The real power of leveraging AI technology is to manage a volume of information that the human brain can’t encompass at one time. Technology doesn’t tell a doctor what to do, it doesn’t make a diagnosis, it presents evidence-based treatment options to enable the doctor to be his or her best self in caring for that patient," Levitan said.
That study was based on a multidisciplinary tumor board 's blinded evaluation of 1,000 breast, lung and colorectal cancer patients. The study found that decisions on cancer treatment also changed due to the AI platform providing more personalized alternatives (30%) or new insights from genotypic and phenotypic data and evolving clinical experiences (15%).
That study was among 22 scientific studies IBM Watson Health presented at ASCO this year demonstrating progress in using AI to provide clinical decision support for cancer care.
The AI platform was trained by specialists at Memorial Sloan Kettering, according to IBM Watson.
The company has faced scrutiny as it has encountered challenges in its work to bring AI to oncology. A Stat report last July, based on internal IBM documents, indicated the Watson supercomputer often produced erroneous cancer treatment advice and that company medical specialists and customers identified “multiple examples of unsafe and incorrect treatment recommendations” as IBM was promoting the product to hospitals and physicians around the world.
With this latest study presented at the ASCO annual meeting, lead investigator SP Somashekhar, chairman of surgical oncology at Manipal Hospitals, said it builds on previous studies and suggests that AI decision support holds substantial promise to reduce the cognitive burden on oncologists, which is a significant problem impacting physician burnout today.
"We consider Watson for Oncology to be an important tool to support decision making, and this study suggests that AI could help reduce variability of care," Somashekhar said in a statement.
Ongoing progress in leveraging AI in clinical care
The studies Watson Health presented at ASCO this weekend demonstrate that the AI platform provides value in improving patient confidence in treatment plans and annotating genomic variants and identifying clinical interventions, the company said.
In another study showcased at the conference, Watson for Genomics was found to identify clinically actionable genomic variants that had not been identified in manual interpretation in a third of patients at a hospital in South Korea. "This helps suggest that the labor-intensive manual curation of such results could be augmented with tools like Watson for Genomics," the company said.
In a third study, physicians from Beijing Chaoyang Integrative Medicine Emergency Medical Center’s oncology department reported patients had a better understanding of their disease and treatment options after Watson for Oncology was incorporated into the consultation process. This led to improved levels of patient engagement and stronger patient confidence in their care plans because patients had a comprehensive view of treatment options, Levitan said.
IBM studies also suggest that machine learning can be used to automatically identify relevant clinical publications and may reduce the time clinicians spend finding pertinent evidence for their patients.
"As an oncologist, my email inbox is full of medical information and it is difficult to assimilate all of this at the point of care. Our data shows that this can be a very powerful tool to curate the literature and bring to it the physician the evidence that is most relevant to the decision at hand," Levitan said.
Medical researchers at IBM Watson also see the potential for the company's AI technology to help reduce variability in cancer care and improve cancer outcomes globally.
"With 18 million diagnoses globally each year, cancer is a devastating disease that has a heavy human toll, as well as a high health system cost," Levitan said. "Patients often face grueling and confusing treatment regimens, while oncologists sift through reams of medical literature and genomic data to identify the best care plan for each individual patient. All the while, researchers are hamstrung by trials that too often fail due to low patient recruitment."
Data, analytics and AI can be used to address these pressing health challenges, Levitan said.