OnDemand: AI for Obesity Research and Clinical Practice-AI Uses in Clinical Care

Video Transcription

Video Summary

Dr. Nikhil Durandar, an obesity researcher and clinician, discussed practical AI uses in clinical care, especially in nutrition and obesity treatment. He explained basic AI concepts like machine learning, deep learning, NLP, and computer vision, then focused on three areas: food image recognition, clinical documentation/communication, and diagnosis/treatment support.<br /><br />He showed how AI can estimate calories and carbohydrates from food photos, sometimes with accuracy similar to humans. Examples included ChatGPT estimating pizza calories and apps like GoCarbs and SnapCalorie. He noted that AI works better with common packaged foods than with homemade, mixed, or culturally specific foods, and that users still need judgment and training.<br /><br />He also described NLP tools that can act as medical scribes, transcribe visits, support charting, handle admin tasks, and translate languages in real time. He stressed that hallucinations and errors mean human review remains essential.<br /><br />Finally, he highlighted AI research using patterns in voice to detect diabetes and CT-based models to predict muscle loss in cancer patients, suggesting similar methods could help monitor obesity treatments. His overall message: AI is not a replacement for clinicians, but a powerful supplement that can improve accuracy, efficiency, and personalization.

Asset Subtitle

By Nikhil V. Dhurandhar, PhD, FTOS

Keywords

artificial intelligence

obesity treatment

nutrition

food image recognition

clinical documentation

natural language processing

diagnosis support

machine learning