2026 ASH®
Annual Meeting & Exposition December 12-15, 2026 ▪ New Orleans, Louisiana

Paper Details

Abstracts, authors, affiliations, and disclosures.

Saturday, December 12, 2026
05:30 p.m. - 07:30 p.m.
MCCNO - Hall F
Access, Representation, Population Health, Technology and Procedures, Assays, Molecular Testing, Pathology, Bioinformatics, Profiling, Treatment Considerations, Supportive Care, Biological Therapies, Transplantation, Artificial Intelligence (AI), Clinical Practice (Health Services and Quality), Clinical Guidelines, Health Economics, Patient Safety, Quality Improvement, Quality of Care, Value, Diseases, Lymphoid Malignancies, Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Research, AML, APL, MDS, MPN, Clinical Research, Health Outcomes Research, Real-World Evidence, Study Population, Human, Adult
PAPER

2655 Patient perspectives about artificial intelligence in hematology and oncology: An interim analysis

Eric C. Blackstone, PhD, HEC-C, Brian Russell, MD2, Angel M. Cronin, M.S.3, David N. Sontag, JD, MBE, HEC-C4, Andrew Hantel, MD, MPH5, Nicholas Groblewski, BS6, Eliezer Van Allen, MD7, Deepa Rangachari, MD8, Gregory A. Abel, MD, MPH9
1 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
2 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
3 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
4 Beth Israel Lahey Health, Boston, Massachusetts, United States
5 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
6 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
7 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
8 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
9 Dana-Farber Cancer Institute, Boston, Massachusetts, United States

Introduction: Despite increasing support for artificial intelligence (AI) in healthcare (Angus 2025, JAMA), ethical issues surrounding its use in hematology and oncology remain. Optimally integrating AI into these subspecialties requires that both patients and clinicians trust in and feel confident with this technology, which in turn demands that patients’ perspectives about AI be understood and addressed. Patient perceptions of clinical AI use may also differ for those with blood cancers versus solid tumors.

Methods: An ongoing cross-sectional survey of patients at Dana-Farber Cancer Institute (DFCI) and Beth Israel Deaconess Medical Center (BIDMC) is assessing patient attitudes towards use of AI in hematology and oncology care and research. The survey was developed based on a previous survey of clinicians (Hantel 2024, JAMA Network Open) and cognitively tested with 12 DFCI patients prior to implementation. Adults seen at DFCI or BIDMC for either hematologic or solid tumor malignancies within the past year are eligible. Patients are identified using institutional research databases and sent surveys via FedEx with $20 incentives and opt out instructions. For this analysis, demographics and prevalence of patient viewpoints were reported using descriptive statistics. Pairwise McNemar tests with Bonferroni correction assessed differences in trust between applications of AI in cancer care, and chi-square testing assessed if these opinions differed based on having a hematologic versus solid tumor malignancy.

Results: To date, the survey has been completed by 109 of 170 eligible patients (64.1% response rate). Most patients were between 60-80 years old (67.9%), 50.0% were female, 95.4% were White, and 30.3% had blood cancers, 67.9% had solid tumors, and 1.8% were unsure. Approximately one third of patients (32.1%) reported intentionally using AI once per week or more. About half expected that AI would make cancer care less personal (54.7%). Overall, patients reported trusting AI to help doctors with recommending screening (85.0%), predicting cancer risk (78.7%), finding a clinical trial (77.1%), detecting whether cancer is getting worse (75.7%), or choosing treatment options (65.7%). As compared to those with solid tumors, patients with blood cancers were less trusting of AI for treatment decisions (51.5% vs 72.2%, p=0.04) and for detecting whether their cancer was getting worse (63.6% vs 81.1%, p=0.05). Overall, significantly fewer respondents trusted AI to help doctors predict how long they will live compared to other potential AI functions (46.1%; pairwise Bonferroni corrected p<0.05). Most felt that patients should be allowed to decline having AI involved in treatment decisions (85.3%), communication with their healthcare team (82.4%), cancer diagnosis (81.7%), and research (72.9%).

Conclusions: In this ongoing survey, patients with both hematologic and solid tumor malignancies generally supported the use of AI in several aspects of cancer care. On the other hand, they had reservations about its ability to predict life expectancy and felt that AI-supported care would be less personal. The majority also wanted to be able to opt out of its use for various aspects of their care.  Interestingly, patients with blood cancer were less trusting of AI for certain aspects of their care, suggesting that cancer AI implementation strategies may need to account for cancer type.

Eric C. Blackstone, PhD, HEC-C: No relevant conflicts of interest to declare.