Thursday, September 03, 2026
Meeting Time: 7:00 pm
Online event (via Zoom): Please register on meetup.com to obtain the Zoom link.
Lecture
Artificial intelligence is reshaping modern healthcare with promises of efficiency, accessibility, and insight at unprecedented scale. However, as AI systems continue to proliferate in clinical settings, critical questions emerge about their appropriate roles, limitations, and the potential consequences of misapplication.
This presentation provides an engineering perspective on evaluating AI tools for medical diagnostics. The presentation will distinguish between two fundamentally different AI architectures: probabilistic language models and deterministic task-specific systems.
Probabilistic systems function by predicting statistically likely outputs based on patterns in training data. They do not reason about physiology, evaluate evidence, or weigh competing possibilities.
In contrast, deterministic task-specific AI systems are designed to follow explicit rules and yield consistent results for identical inputs. These systems can be validated rigorously, tested repeatedly, and regulated because their behavior is predictable. A central theme will address why "usually right" is insufficient when individual patient outcomes are at stake.
Medicine is irreducibly multimodal, requiring integration of visual cues, physical examination findings, auditory information, laboratory data, imaging studies, waveform data, and clinical intuition developed over thousands of patient encounters.
Attendees will learn practical frameworks for critically evaluating AI accuracy claims, including understanding false positive and negative rates, recognizing the "external validation illusion," and identifying the accuracy paradox where a model can achieve high apparent accuracy while failing to detect the cases that matter most.
The presentation will conclude with guidance on identifying appropriate versus inappropriate applications of AI in healthcare, emphasizing why clinical judgment must remain central to patient care while acknowledging the genuine value AI can provide in specific, well-validated contexts.
Speaker Bio:
Dr. Milan Toma is an Associate Professor at the New York Institute of Technology College of Osteopathic Medicine (NYITCOM), specializing in clinical sciences. He holds a Ph.D. in Bioengineering from the Technical University of Lisbon.
His research spans computational biomechanics, medical imaging, and the application of machine learning and artificial intelligence in healthcare.
Dr. Toma authored two books on AI in medical diagnostics and has published dozens of peer-reviewed research studies.
His recent work includes AI applications in cardiovascular health, orthopedic patient classification, medical image diagnosis, and critical evaluation of AI capabilities in clinical settings.
Notes
There is no cost to attend this meeting, however, if you are a NYS Professional Engineer and would like to receive Professional Development Hours (PDHs) of continuing education credit, then payment of a $15 fee is required. PDHs will be granted based upon the actual duration of the lecture including any demos and Q&A. You must stay to the end to receive credit. You will also have to properly fill out an Evaluation Form to prove that you attended this lecture.
Click here to open the Evaluation Form. Simply fill it out and click on the “Submit” button.
See these detailed instructions on how to receive PDHs for this lecture.
