The theme of this year’s meeting is 2025: A Hematologic Oncology Odyssey—Unleashing AI, Genomics, Targets, and Immune Engineering. Five talks will be presented SOHO 2025 on the topic of AI.
Artificial intelligence (AI) has seen a staggering influx of investment across nearly every sector, and hematologic oncology is no exception. While some may think AI is overhyped, others believe it is poised to revolutionize the way hematologic oncologists approach the diagnosis, treatment, and understanding of blood cancers.
Five talks at this year’s SOHO Annual Meeting address applications of AI and machine learning (ML) in the diagnosis, prognosis, treatment, and monitoring of several disease types, including myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPNs), and various lymphomas.
SOHO Insider reached out to several of the scheduled speakers, including Friday’s plenary presenter Charles Mullighan, MBBS (Hons), MSc, MD, FRS, to get their take on the present and future role of AI in the field of hematologic oncology.
AI highlights from SOHO 2025
Jonas Paludo, MD, whose session is titled “What Is the Potential of AI Remote Patient Monitoring for Lymphoma?” is an assistant professor of medicine and oncology at the Mayo Clinic in Rochester, Minnesota.
In his talk, Dr. Paludo argues that the integration of AI with remote patient monitoring can transform cancer care by enabling real-time, predictive surveillance that enhances early detection of complications, personalizes treatment, and may improve outcomes.
“AI-driven models can analyze multimodal data, from physiological signals to imaging and genomics, to anticipate adverse events such as cytokine release syndrome, offering a proactive approach to patient safety,” Dr. Paludo said. “This is particularly impactful for patients with lymphoma, who often require long-term surveillance and timely intervention for treatment-related complications.”
Dr. Paludo noted that AI-enabled platforms can synthesize data from wearable devices, electronic health records, and patient-reported outcomes to generate predictive insights, allowing clinicians to tailor follow-up strategies and reduce unnecessary hospital visits. However, he acknowledged that challenges remain in terms of validating AI models across diverse populations, ensuring data privacy, and integrating these tools into existing clinical workflows.
“As the field evolves, multidisciplinary collaboration will be essential to fully harness AI’s potential to improve outcomes for cancer patients,” he said.
Eric Hsi, MD, whose session is titled “AI Models for Diagnosis and Prognostication in MCL,” is chair of the Department of Laboratory Medicine and Pathology at Mayo Clinic.
“We are just beginning to realize the potential for AI-powered tools to assist in the diagnosis and management of patients with hematologic malignancy,” Dr. Hsi said. “Harnessing large datasets to unlock the hidden insights and efficiently deliver care is the goal, but doing this in a way the takes into account potential bias in our data will be of great importance.”
Dr. Hsi also noted that advances in scanning technology and the training of foundation models with large image sets are enabling the development of algorithms that can power proof-of-concept tools for AI-enabled diagnostics in the field of pathology.
Amin Turki, MD, PhD, whose session is titled “AI Agents Transforming Care in Bone Marrow Transplantation,” is director of the Computational Hematology Lab at the Institute for Artificial Intelligence in Medicine at the University Hospital Essen in Germany.
Dr. Turki’s talk highlights the role of AI agents in the context of cellular therapy and their transformative potential in the practical care of patients undergoing allogeneic hematopoietic stem cell transplantation (HSCT).
“To date, most AI models and tools remain experimental and, consequently, have not been formally transformed into medical devices; however, they have already produced a sustainable impact,” Dr. Turki said.
Specifically, Dr. Turki noted that the intricate patterns of viremia and the multifaceted interactions with the microbiota following allogeneic HSCT render them a compelling subject for AI-based approaches. He also cited research showing that continuous vital sign monitoring using wearable devices has been successfully leveraged to predict severe infections.
“The digital transformation of medicine is ongoing,” Dr. Turki observed. “Predictive AI models and data-driven classification systems that leverage big data have identified new disease patterns in cellular therapy but also revealed that progress remains challenging.”
While the role of AI in medicine is still in flux, implementation into practice is currently being enabled by university hospitals that are experimenting with in-house AI solutions using hospital exemption regulations, Dr. Turki noted.
He also observed that the training curriculum in hematologic oncology will need to begin incorporating AI-related topics going forward.
“The increasing impact of AI requires solid post-graduate education with a focus on the basic understanding and practical applications in hematology,” he said.
‘A soup-to-nuts approach’
Finally, SOHO Insider spoke with leukemia genomics expert Dr. Mullighan, who is the senior deputy director of the St. Jude Comprehensive Cancer Center, to get his take on AI’s impact.
“Some of the tools of AI, such as ML models, have actually been used for a long time, for instance, to understand genomic data,” Dr. Mullighan said. “Obviously, the transformation here has been in terms of the compute power and the development of approaches such as large language models. Now, we have this enormous amount of data, including genomic information that can be used to train the models.”
Diagnostics is a logical application domain for AI, he said.
“There is beautiful work that has been done by groups such as the Munich Leukemia Laboratory and others that have used AI to develop a soup-to-nuts approach to diagnose diseases quickly and accurately and reduce the amount of human interaction that is needed,” he said. “It’s very impressive.”
Citing the large data science initiative led by his colleague M. Madan Babu, PhD, FRSC, FMedSci, FRS, chief data scientist and senior vice president for data science at St. Jude, Dr. Mullighan highlighted ongoing efforts to extract meaningful insights from layered biological data.
“One thing that we are thinking about is how do we take various types of information, including genomic, transcriptomic, and proteomic information, and then distill that down?” he noted. “To pick our battles, if you like, and say, we think this subtype, or this protein, or this part of a protein is the right target for a focused effort to develop targeted therapies.”
“It’s a very exciting time,” he concluded.
Visit the SOHO 2025 meeting news page for more coverage from the meeting.
