Why Human Expertise Still Matters in Healthcare AI

Artificial Intelligence can pull structured data from a medical chart in seconds—but can it understand the "why" behind an oncology patient's changing treatment plan? True data quality will always require an experienced human in the loop. In this episode of Exploring AI in Oncology, we bust the myth that AI can completely automate clinical registries and reveal why data abstraction is actually a critical patient safety issue where a single missed data point can lead to dangerous over-radiation. Read the full accompanying article on Oncology Data Advisor: To dive into this massive paradigm shift, our host Dr. Waqas Haque sits down with Erin Tobaison, Senior Clinical Data Abstraction Manager at Carta Healthcare. Together, they discuss how AI is reshaping registry workflows while reinforcing the critical need for strong clinical judgment, specialty expertise, and institutional support. 🔹 The Safety Link: Why faulty data abstraction isn't just a compliance issue, but a direct threat to downstream clinical decision-making and patient safety. 🔹 Where AI Wins: How AI-assisted workflows cut down abstraction times from 30 minutes to 15 minutes for simple cases, and save hours on complex charts. 🔹 The Hybrid Model: Why expert human oversight is indispensable for interpreting ambiguous language, conflicting documentation, and nuanced pathology reports. 🔹 The Specialty Myth: Why abstractors are not interchangeable, and why moving an expert from oncology to a cardiovascular registry makes them a "novice" again. 🔹 The Learning Health Cycle: How monthly audits and investing in ongoing provider education ensure AI acts as an accelerator, not a replacement for clinical judgment.