The Role of AI with Tumor Boards

Multidisciplinary tumor boards rely on evidence to help make the best recommendations possible for patients. The 3 million peer-reviewed publications related to cancer and more than 250 U.S. Food and Drug Administration-approved oncology drugs make keeping up with the latest developments a virtually impossible task. Doctors would have to spend approximately 21 hours each day to keep up with new professional insights, according to one estimate.

This deluge of scientific evidence holds great potential in that it can help clinicians personalize care plans for patients. One of the ways clinicians collaborate to personalize care for patients is through tumor boards, which are groups of physicians and other healthcare professionals with different specialties that meet regularly to discuss cancer cases. But vast amounts of new information present a problem for tumor boards: How can they efficiently incorporate the most relevant evidence into their decision-making?

Artificial intelligence (AI) is emerging as a decision support tool for tumor boards. It combs through evidence and data to find the most relevant information for that specific patient case. Recent studies highlight how tumor boards are using AI to:

• Inform care decisions: AI can review relevant data and evidence, incorporate standard clinical guidelines and present this information in an actionable way. This information may directly affect care decisions. For example, in a retrospective review of 1,000 tumor board cases, clinicians changed their care decisions in 13.6% of the cases after reviewing information from an AI solution. They changed care decisions because AI identified newer treatment options (55% of cases), more personalized options (30% of cases) or new insights from genotypic and phenotypic data (15% of cases).
• Match patients to clinical trials: One study showed that tumor boards help increase physician awareness, clarity and enthusiasm about clinical trials. However, manual screening of eligibility requirements for clinical trials is time-consuming and typically results in a low percentage of eligible cancer patients being enrolled. In a study of an AI-powered clinical trial matching system in an ambulatory oncology practice, oncologists used the system to help increase enrollment in clinical cancer trials from an average of 3.5 to 6.4 patients per month, about an 80% increase.
• Consider genomic evidence: As tumor boards strive towards more personalized treatment options for their patients, they seek to understand the genomic profile of each patient’s tumor. At the University of North Carolina, researchers found that AI identified genomic events of potential significance — ones that had been missed by molecular tumor boards — in 32% of patients. Additionally, AI genomic variant interpretation tools have been shown to complete analysis and compile a pathology report of potentially clinical actionable insights in as few as 10 minutes.

These are just some examples of the role AI can play with tumor boards. While it’s just the beginning of the journey, and there are still challenges ahead, this early evidence of progress is encouraging. When tumor boards harness the capabilities of AI, the scientific evidence shows they can help improve clinician workflow and personalized care for patients.

About the author: Jeffrey T Lenert, M.D., MBA, FACS, is associate chief medical officer of Oncology & Genomics at IBM Watson Health. Prior to joining IBM Watson Health, Lenert was medical director of the Breast Care Center and attending surgeon at Walter Reed National Military Medical Center in Bethesda, Maryland.

Get the best insights in digital health directly to your inbox.

Related

AI Performs Similarly to Humans in Identifying Cancerous Lesions

Digging Deeper: Ethical Use of AI in Radiology

AI in Tandem with Radiologists More Accurately Identified Breast Cancer