On this episode of Tell Me Where It Hurts, Dr. Jay Anders, MD, Chief Medical Officer of Medicomp Systems, describes a heated conversation with a colleague who believes AI will replace physicians, even to the point of performing surgery without a clinician in charge. His guest, Niki Panich, MD, FCFP, chief medical officer of Penguin Ai, takes a more measured view. She compares a newly deployed AI model to a medical student who overwhelms their professors with too much information, then advances to resident and eventually fellow. She tells Dr. Anders that the model still requires physician oversight.
Dr. Panich’s own path into data science began in her clinic. As a family physician whose practice includes obstetrics, she saw high rates of low birth weight and preterm labor among her patients and suspected the answer was in her charts, though she lacked the skills to find it. As a result, she earned a master’s degree in data science and an MBA from the Stanford Graduate School of Business. She has since served as chief medical officer of XO Health, a health plan, and as a corporate medical director at Cleveland Clinic.
At Penguin Ai, Dr. Panich applies her clinical, data science, and health plan experience to the administrative tasks that take clinicians away from their patients. The company builds AI agents for work such as prior authorization, claims appeals, and chart summarization. Dr. Panich tells Dr. Anders that clients often find the most value in the data layer beneath those agents, which reconciles records, such as a physician who practices in two locations or appears as Niki in one record and Nicole in another.
“The output of any system, whether it’s a model, whether it’s an insight, is only as good as the data that comes in.”
Niki Panich, MD, FCFP
Chief Medical Officer, Penguin AI
Dr. Anders questions whether adding AI to every workflow may increase a clinician’s cognitive load. Dr. Panich explains that her team measures clicks, scrolling, and decision-making ability at the start and end of a clinic day to make sure the tools reduce that load rather than add to it. Beyond interface design, she recommends that organizations build trust through low-risk projects, such as identifying patients likely to screen with elevated hemoglobin A1C, before automating prior authorization from start to finish.
As more real-world data becomes available, including factors such as exercise, diet and transportation, Dr. Panich expects clinicians to be able to show patients how different choices may affect their health over the next five years. Asked what she would change with a magic wand, Dr. Panich says she would give every patient the same understanding of their health data that she has of her own.
Show Notes
Among the topics covered:
- Why AI recommendations still need human judgment: an ideal candidate for an injectable GLP-1 may never take it if they’re afraid of needles.
- Data wrangling makes up about 90% of data science, and the model comes after.
- AI chart summarization can turn a 45-page discharge summary into the key follow-up items a primary care physician needs.
- Deciding whether an AI agent is the right tool, or whether standard code will do.
- “Forward-deployed” clinicians work alongside technical teams to gather requirements and translate medical language for engineers.
- A patient 360 profile combines clinical, lab and demographic data that can be shared within and across institutions to avoid duplicate tests.
- Dr. Anders on why getting data to the right place in a usable format matters, when records still arrive as PDFs and faxes.
- And more…
Show Links
- Dr. Niki Panich on LinkedIn
- Penguin Ai
- Dr. Jay Anders on LinkedIn
- Medicomp Systems on LinkedIn
- Tell Me Where IT Hurts on Healthcare NOW Radio