Six jobs for AI in myeloma care, from Prerna Mewawalla, M.D.

Nearly half of community oncologists (49%) now use AI tools, but only 19% use AI-powered predictive models, even though 83% believe AI can make cancer care more accessible, according to Johnson & Johnson’s 2026 Oncology Care Index. The Harris Poll conducted the survey for J&J among 109 community oncologists and 100 practice administrators.

Prerna Mewawalla, M.D., director of the Stem Cell Transplantation and Cellular Therapy Program at Allegheny Health Network (AHN) in Pittsburgh, worked with J&J on the index. In this interview with Managed Healthcare Executive, the hematologist-oncologist, who specializes in multiple myeloma and AL amyloidosis, walks through six places AI can help across a myeloma patient’s care, from the first visit onward.

Cutting the paperwork

The first is data review and synthesis. Patients who come for a second opinion often arrive with stacks of lab results, pathology reports and scans, and AI can review them and summarize what the physician needs to know. The second is documentation and workflow. Mewawalla uses ambient documentation for note-taking, which she said lets her talk with patients rather than type during the visit, though she still reviews each note for accuracy. Next, she expects AI to place the orders she mentions during a visit, such as a PET-CT or lab work, and leave them unsigned for her review.

Keeping care on schedule

Third, AI can support guideline-based care. After autologous stem cell transplant, myeloma patients need specific follow-up at one, three, six and 12 months, and AI can flag a missed milestone. It can also surface NCCN-recommended options for a newly diagnosed patient. Fourth is clinical trial matching. A community oncologist may see 20 or more patients a day, Mewawalla noted, and can’t track every open trial for every cancer. AI integrated with the EHR could screen patients and alert physicians to those who may be eligible.

Fifth is literature review. Tools such as OpenEvidence return referenced answers to clinical questions, so physicians can check the underlying evidence.

Spotting what’s easy to miss

The sixth, and the one Mewawalla sees as the future, is pattern recognition. Amyloidosis can mimic many other conditions, and diagnosis is often delayed in ways that hurt prognosis. A tool that connects neuropathy, abnormal lab results and cardiomyopathy could prompt a physician to consider amyloidosis sooner.

“I think AI can be a great tool and a great adjunct to clinical practice,” she said.