Community oncologists are turning to AI, but not all of them use predictive models

Only 19% of community oncologists surveyed use AI-powered predictive models in practice, even though 83% believe artificial intelligence (AI) can make cancer care more accessible, according to the new 2026 Oncology Care Index from Johnson & Johnson (J&J).

The survey, conducted online from May 21, 2026, to July 6, 2026, by The Harris Poll for J&J, included 109 community oncologists and 100 practice administrators, the company said in a news release.

The results were not weighted and reflect only those who completed the survey.

How many community oncologists use AI in practice?

The survey found that nearly half of surveyed community oncologists (49%) said they use AI tools. J&J compared the oncologist figure with 19% of healthcare providers in all practice settings in 2025. The 2025 survey covered a broader group that included urologists and advanced practice providers, so the numbers are not directly comparable.

June Lanoue, president of hematology at J&J, told Managed Healthcare Executive (MHE) that the same trend holds in hematology, where more than half of patients receive care in community settings. Only 10 of the 109 oncologists surveyed were hematologists.

AI can help at several points in multiple myeloma care, added Prerna Mewawalla, M.D., director of the Stem Cell Transplantation and Cellular Therapy Program at Allegheny Health Network (AHN), in an interview with MHE. Mewawalla worked with J&J on the index. She said tools can summarize records a patient brings for a second opinion, draft visit notes so physicians can talk with patients instead of typing and screen electronic health records to match patients with clinical trials.

Looking ahead, Mewawalla said AI could help catch AL amyloidosis sooner by linking signs such as nerve damage and heart muscle disease, since the condition often is diagnosed late. For literature searches, she pointed to OpenEvidence, which she said summarizes recent evidence with references and is free for physicians and advanced practice providers.

What is holding back AI adoption in oncology?

The news release described the main barriers as structural. Lack of reimbursement for AI tools was cited by 44% of community oncologists. Another 43% pointed to limited formal training and 41% to insufficient funding. In the survey, 97% of respondents also said payer barriers delay or prevent patients from receiving the therapies they believe are most appropriate.

Community oncologists “are not waiting to be invited to the AI conversation,” said Stephen (Fred) Divers, M.D., chief medical officer at American Oncology Network, in the release.

Mewawalla said there is a difference between knowing about AI and having a tool that is “validated, integrated, reimbursed, and useful in everyday practice.”

She added that tools must work with the electronic health record, meet federal privacy rules and be accurate, and practices need the money and training to use them.

Where do community oncologists lack confidence in new immunotherapies?

Among community oncologists, 92% said their practice is very or somewhat comfortable managing new immunotherapies, and 94% said it is very or somewhat equipped to offer them. Fewer said they are very comfortable (43%) or very equipped (35%). The top barriers were staff training and expertise (38%), financial and reimbursement resources (38%) and infrastructure for adverse event management protocols (37%).

According to Mewawalla, clear protocols can make chimeric antigen receptor (CAR) T-cell therapy and bispecific antibody treatment workable in community practices, even if side effects such as cytokine release syndrome and neurotoxicity can be overwhelming for new teams. Safeguards include choosing patients carefully, giving them wallet cards that alert emergency and intensive care staff and setting up one number to call. Anyone answering a 2 a.m. call should have step-by-step instructions for each side effect grade, Mewawalla said.

Trial options are another gap. About 6 in 10 community oncologists (60%) said their practice offers some trial options but a limited variety. Mewawalla encouraged that trials should include more patients who have other health conditions and be built so patients can take part locally instead of traveling to an academic center. She also said telehealth visits are not allowed in trials and that this needs to change.

J&J is investing in continuing medical education and specialized nurse educators for community care teams to help close these gaps, Lanoue said. The company also issued a request for proposals for not-for-profit groups working on the barriers the index identified.

Kevin Hamill, company group chairman of J&J Innovative Medicine North America, said in the release the gap between what community oncologists believe AI can do and what they can actually do with it today “isn’t closing on its own.”

“That’s not a confidence problem: it’s a systems problem, which is why we’re investing directly in areas that can help move them from readiness to real impact,” Hamill said.

One example is AHN, which offers bispecific antibodies at its community sites, including in Erie, about two hours from Pittsburgh, Mewawalla shared. Patients can receive the initial ramp-up doses outpatient, and the network trained staff across sites to manage side effects, she said.

Travel can strain finances, caregiver availability and work schedules, and some patients miss effective treatment because they cannot make the trip, Mewawalla said, adding that she partnered with J&J on the index because “a patient’s zip code really shouldn’t determine the quality of cancer care they receive.”