If you asked someone half a century ago what they thought about riding in an autopilot car, there would be a plethora of uninviting and discouraging responses like, “Impossible!” “Future ex-worst decision,” “Epic fail,” and so on. Yet, today we live in a world where the Apple® iPhone uses face recognition to unlock your phone, autopilot car systems can tell a red traffic light apart from a green light and drive you around without supervision (or minimal supervision), and retailers are getting ready to use voice and body language pattern recognition for customer satisfaction assessments. Well, now that artificial intelligence (AI), data science, and computer vision are rapidly emerging as the holy trinity of the future, ever wonder how far they are from medicine and hematology? Past is prologue, and in hematology, we are programmed to catch the trends quickly, adopt new technologies seamlessly, and strive for improvement habitually.
The Scientific Program session “AI, Data Science, Computer Vision and the Hematology Laboratory of the Future” (with virtual and in-person live Q&A), taking place Sunday, December 12, will showcase the work of three speakers. Chaired by Dr. Scott J. Rodig, these talks are arranged to perfection to introduce you to the deployment of computer science and AI in the hematology laboratory, ranging from the basic discovery to the translational, as well as the real-world aspects of these tools. Dr. Yinyin Yuan will review her team’s work on geospatial organization of the tumor microenvironment for pathological analysis of tumor samples. Next, Dr. Metin Gurcan will discuss how to use AI on large datasets to improve pathological diagnosis. Lastly, Dr. David Jaye will spell out current practice patterns and the merits and pitfalls of incorporating these technologies into medicine.
We are beginning to see that our ever-growing bubble of hematology has a lot to gain from AI, and this new session acquainting us with AI and hematology is especially timely. When I heard of it for the first time, I envisioned a super fancy automated cell counter in the hematology laboratory, only to then realize it will be even fancier, more accurate, and quicker than I thought. While the title of this session points to the future, I can’t help but wonder if the implementation of these technologies in our field is not too far from the present. That is why there is some knowledge to glean for all of us proud lifelong learners — early trainees as well as seasoned pathologists and clinicians.
Now, the future of AI will dwell not only in diagnosis but also in prediction tools, prognosis assessments, response to treatment patterns, and possibly treatment decisions. As this field evolves, we will only know it best if we grow with it — in other words, learning by being a part of it!
We have all witnessed the face-lift that next-generation sequencing has brought to diagnosis in hematology. Yet we remain aware that the marrow phenotype remains the final determinant. Augmenting the latter with application of AI into analyzing the microscopic images of cells in cytomorphology or the cell population in bidimensional flow cytometry plots, takes us atop the summit of precision in diagnosis. This further segues into deep learning algorithms for disease classification and subtyping, practice and response patterns, in silico clinical trials, and wearable devices for longitudinal monitoring of disease phenotypes. My inner Captain Obvious urges me to state that all this happens above the level of human observer error — what computers are best known for.
And if AI is of existing or emerging interest to you, let me also direct you to the Scientific Workshop on Artificial Intelligence in Diagnostic Hematology (available on demand) and oral abstract session “Emerging Diagnostic Tools and Techniques: Artificial Intelligence and Comprehensive Sequencing Improve Diagnostics and Prognostication in Hematology” being presented on Saturday, December 11, and other poster abstracts included within disease groups including “Patients Diagnosed With Myelodysplastic Syndromes Display Lower Bone Mass: Assessment Using Bone Marrow Biopsies Manually and with Artificial Intelligence” (Monday, December 13, at 6:00 p.m.) “Artificial Intelligence and Flow Cytometry to Classify Plasma Cells Dyscrasias” (Monday, December 13, at 6:00 p.m.), and “Artificial Intelligence-Based Predictive Models for Acute Myeloid Leukemia” (Monday, December 13, at 6:00 p.m.).
In sum, the opportunities are boundless. The question is, how far do we want to go?
Dr. Jain indicated no relevant conflicts of interest.