There is no turning a blind eye to AI at the 64th ASH Annual Meeting and Exposition in New Orleans. As you peruse the program, keep an eye out for the wide array of presentations highlighting creative ways of integrating artificial intelligence (AI) and machine learning into the practice of hematology. Similar to the excitement surrounding “smart” appliances and self-driving cars, imagine the impact of an expert consultation from a “smart” computer program that can quickly sift through one’s electronic medical records to devise the optimal, most evidence-based treatment plan. With an ever-expanding arsenal of “big data” available to the practicing hematologist, made possible by advances in precision medicine and molecular genetics, the role of pattern recognition by machine learning will be critical in the process of therapeutic decision-making. Merging robot-like accuracy with the human faculty of reasoning and spirit is the goal. Think Inspector Gadget.
In preparation for the start of the meeting, please keep an eye out for the Scientific Workshop “Leveraging Digital Technology to Modernize and Enhance Observational Studies.” Studies highlighting the use of machine learning and digital technology in the application of patient care will be explored in detail. A sneak peek: The workshop will include studies on the use of electronic medical data to assess risk of thrombosis and bleeding in medical inpatients, gauging true performance status via wearable devices in patients with lymphoma, and monitoring patients post-allogeneic transplantation with an AI-enabled EKG algorithm to predict arrythmias.
A follow-up Special-Interest Session on Monday titled “AI and Machine Learning: A New Frontier in Hematology” will further expand on how machine learning and deep learning algorithms actually work, including the process of validation, the implicit biases one should be aware of when interpreting output data, and the roadblocks to implementation of this technology. Other important issues, including regulatory and oversight considerations and the extent of appropriate clinical use of this technology, will be discussed as well.
As if that wasn’t enough, there will be a plethora of oral and poster presentations that delve into innovative approaches incorporating AI technology into general hematology practice. So, what does the future hold? Well, there is no memory capacity on this future hard drive. Here’s a snapshot:
How about utilizing an algorithm that incorporates specific imaging data on a positron emission tomography scan in conjunction with laboratory and clinical data to identify patients with relapsed and/or refractory diffuse large B-cell lymphoma at risk of complications from chimeric antigen receptor T-cell therapy? How about an AI algorithm that can predict a disease entity solely from whole-genome and transcriptome sequencing data? How about discerning certain genomic translocations on bone marrow samples without immunophenotyping or genomic assays and relying on morphologic features alone? How about going beyond the traditional acute myeloid leukemia (AML) European Leukemia Net risk stratification and using AI to capture prognostically relevant genetic abnormalities in deciding whether or not to proceed with an allogeneic transplant in first complete remission in patients with AML? How about using a machine-learning cloud-based algorithm to analyze cells on a peripheral blood smear rather than manual review?
While the specific ins and outs of how AI and machine learning work is beyond the scope of this piece, clearly the output of information received is only as good as the input provided. That’s why new molecular tools for precision diagnostics in hematology are so important and why this is the focus of an oral abstract session. Presentations will focus on liquid biopsy–based next-generation sequencing of circulating tumor DNA fragments in hematologic neoplasms, a new single cell proteogenomic platform that can simultaneously detect and analyze single nucleotide variants, copy number variations, and proteomics data, as well as the real-time application of treatment selection based on functional drug testing of a patient’s cell sample.
In an age of self-driving technology, wearable medical devices, “smart appliances” and “real” conversations with Alexa and Siri, the future application for advances in hematologic patient care is bright. With that said, keep an eye out for AI.
Dr. Hermel indicated no relevant conflicts of interest.
About the Author: Dr. David Hermel (@D_Hermel) is a clinical instructor of medicine at Scripps Clinic in La Jolla, California, where he focuses on bone marrow transplantation and cellular therapy. He earned his undergraduate degree at UCLA before attending medical school at the University of Vermont, completing his internal medicine residency at University of Southern California, and finishing a hematology/oncology fellowship at Scripps Clinic. Hailing from Sherman Oaks, California, and currently residing in La Jolla, Dr. Hermel enjoys spending time with his wife and three daughters. He is a proud member of the ASH Communications Committee and excited to be an author for the ASH News Daily.
Dr. David Hermel will be joined by Dr. Jonathan Hermel, who is currently a resident physician in pediatrics at UC San Diego and Rady Children’s Hospital. He graduated from the University of California, Berkeley in 2016 with a BA in molecular and cell biology. After earning numerous distinguished accolades for his undergraduate performance, he continued his medical training at the Tulane University School of Medicine, where he was inducted into the Alpha Omega Alpha Honor Medical Society in his Junior year. Dr. Hermel is interested in medical education and pediatric leukemia research.