Code & Cure

What if your discharge instructions were written in a language you couldn’t read? For millions of patients, that’s not a hypothetical, but a safety risk. And at 2 a.m. in a busy hospital, translation isn’t just a convenience; it’s clinical care.

In this episode, we explore how AI can bridge the language gap in discharge instructions: what it does well, where it stumbles, and how to build workflows that support clinicians without slowing them down. We unpack what these instructions really include: condition education, medication details, warning signs, and follow-up steps, all of which need to be clear, accurate, and culturally appropriate.

We trace the evolution of translation tools, from early rule-based systems to today’s large language models (LLMs), unpacking the transformer breakthrough that made flexible, context-aware translation possible. While small, domain-specific models offer speed and predictability, LLMs excel at simplifying jargon and adjusting tone. But they bring risks like hallucinations and slower response times.

A recent study adds a real-world perspective by comparing human and AI translations across Spanish, Chinese, Somali, and Vietnamese. The takeaway? Quality tracks with data availability: strongest for high-resource languages like Spanish, and weaker where training data is sparse. We also explore critical nuances that AI may miss: cultural context, politeness norms, and the role of family in decision-making.

So what’s working now? A hybrid approach. Think pre-approved multilingual instruction libraries, AI models tuned for clinical language, and human oversight to ensure clarity, completeness, and cultural fit. For rare languages or off-hours, AI support with clear thresholds for interpreter review can extend access while maintaining safety.

If this topic hits home, follow the show, share with a colleague, and leave a review with your biggest question about AI and clinical communication. Your insights help shape safer, smarter care for everyone.

Reference: 

Accuracy of Artificial Intelligence vs Professionally Translated Discharge Instructions
Melissa Martos, et al. 
JAMA Network Open (2025)

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

What if your health data spoke in one calm voice instead of twenty buzzing ones? In this episode, we explore an AI “interpreter layer” that turns step counts, sleep stages, and alerts into fewer, smarter signals that nudge real behavior—without the anxiety spiral. Vasanth (AI researcher and cognitive scientist) and Laura (emergency physician) bring lab insight and frontline reality to a problem most dashboards ignore: humans have limited working memory, serial attention, and a knack for missing rare but important events. More data isn’t always better; often, it’s just louder. 

So what does “useful” look like? Clear summaries in plain language. Patterns stitched across streams—workouts linked to calmer moods, dinner timing tied to glucose swings. Personal baselines that ditch one-size-fits-all thresholds. Instead of a raw feed, imagine a tight weekly brief that surfaces the top two trends, why they matter, and one small experiment to try—aligned with your clinician. That’s the shift from charts to choices.

Trust and safety stay center stage. We unpack sensor accuracy, false arrhythmia flags, and the risk of AI hallucinations. The answer isn’t blind automation; it’s human-in-the-loop oversight, transparent provenance, and user controls to set goals, define “normal,” and mute the rest. We also show how primary care can ingest concise, standardized summaries instead of five pages of logs—making visits more focused and collaborative.

If you’re ready to trade a 24/7 body ticker for meaningful insights you can act on, this conversation offers a realistic blueprint. Subscribe, share with a friend drowning in metrics, and leave a review telling us the one metric you actually use—and the one you’d happily hide.

Reference: 

Do we need AI guardians to protect us from health information overload?
Arjun Mahajan and Stephen Gilbert
npj Digital Medicine (2025)

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

What happens when your health chatbot sounds helpful—but gets the facts wrong? In this episode, we explore how AI systems, especially large language models, can prioritize pleasing responses over truthful ones. Using the common confusion between Tylenol and acetaminophen, we reveal how a friendly tone can hide logical missteps and mislead users.

We unpack how these models are trained—from next-token prediction to human feedback—and why they tend to favor agreeable answers over rigorous reasoning. We spotlight a new study that puts models to the test with flawed medical prompts, showing how easily they comply with contradictions without hesitation.

We then test two potential fixes: smarter prompting that gives models room to say no, and fine-tuning that teaches them how to refuse bad questions. Both strategies improve accuracy—but they come with trade-offs like overfitting and reduced flexibility.

Finally, we look ahead to the promise of “reasoning-aware” systems—AI tools that pause, question assumptions, and gently course-correct with clarifications like “Tylenol is acetaminophen.” It’s a roadmap for safer digital health assistants: empathetic, accurate, and ready to push back when needed.

If you’re building medical AI, practicing care, or just googling symptoms at 2 a.m., this episode offers practical insights into designing more trustworthy tools. Subscribe, share, and let us know—when should AI say no?

Reference: 

When helpfulness backfires: LLMs and the risk of false medical information due to sycophantic behavior
Shan Chen, et. al 
NPJ Nature Digital Medicine (2025)

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

What if digital assistants could triage symptoms, schedule appointments, and coordinate rides—all while doctors focus on the human side of care? That’s the promise of multi-agent AI in healthcare. In this episode, we explore how these intelligent teams of agents are transforming both clinical and operational workflows.

We begin by breaking down what an AI “agent” really is: not just a chatbot, but a goal-oriented system that can use tools, call APIs, and take real-world actions. You'll hear how agent teams are structured—with supervisors, shared workspaces, and collaborative checks—to ensure safety, usefulness, and accountability before any recommendation reaches a patient.

We also unpack the difference between clinical agents (like wearables that surface risks and suggest tests) and operational ones (verifying insurance or scheduling visits). Scoped access and role-based permissions keep data secure while enhancing efficiency.

Real-world examples bring it all to life. A spike in heart rate triggers a wearable agent to alert a clinician. Another agent gathers context. A third finds the nearest available lab slot. Even when agents disagree—say, over CT vs. ultrasound—a supervisory agent helps weigh the evidence, with the final call left to the human clinician.

We talk candidly about challenges: empathy from conversational agents can help with adherence but risks overreliance or emotional confusion. Guardrails like transparency, audit trails, and clear handoffs to humans are essential for trust and safety.

The big picture? AI agents aren’t replacing healthcare professionals—they’re extending their reach, improving responsiveness, and easing system burdens when thoughtfully deployed. When done right, it’s not man vs. machine—it’s care, better coordinated.

If you enjoyed this conversation, follow the show, share it with a colleague, and leave a quick review to help others discover it.

Reference: 

Coordinated AI agents for advancing healthcare
Michael Moritz et al. 
Nature Biomedical Engineering (2025)

What are AI agents, and what can they do for healthcare?
Carlos Pardo Martin et al
McKinsey Healthcare Blog (2025)

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

Artificial intelligence promises faster notes, smoother workflows, and smarter clinical decisions. But behind every seamless interaction lies an invisible cost—electricity, water, and carbon emissions that rarely enter the healthcare conversation.

In this episode, we trace what happens after you hit “enter” on a clinical prompt. From power-hungry GPUs to evaporative cooling systems in data centers, we uncover the hidden infrastructure fueling AI and how metrics like PUE translate convenience into environmental impact. A single prompt may only consume “a few drops,” but scaled across a hospital, it becomes a lake.

Blending insights from an AI researcher and an ER physician, we unpack the difference between training and serving costs, the overlooked impact of iterative prompting, and how everyday uses—charting, imaging, messaging—accumulate real-world carbon. Then we shift to what you can do now: swap out large models for leaner alternatives, trim excessive input context, and build smarter prompts that reduce compute without compromising care.

We also explore operational strategies: batch non-urgent tasks during off-peak hours, negotiate SLAs that trade slight latency for sustainability, and push vendors on the things that matter—like data center efficiency, water use, and renewables—not just performance scores.

Sustainable AI isn’t a dream—it’s a design choice. So where will you start: documentation, imaging, or patient messaging?

Reference: 

Sustainably Advancing Health AI: A Decision Framework to Mitigate the Energy, Emissions, and Cost of AI Implementation
Anu Ramachandran, et al.
NEJM Catalyst (2025)

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

What if the person who knows you best isn’t the best person to speak for you when it matters most?

We explore a study that tested just that—comparing the CPR preferences predicted by loved ones with those predicted by machine learning. The result? Algorithms got it right more often. That surprising outcome raises tough, important questions: Why do partners misjudge? And could AI really support life-and-death decisions when seconds count?

We unpack the study’s approach in everyday terms: who was surveyed, what data fueled the models, and how three algorithms were trained using demographics, clinical records, and stated values. The twist? Basic details like age and sex turned out to be stronger predictors than deeply personal values or medical history. That finding sparks a deeper conversation about autonomy, identity, and the tension between individual dignity and data-driven generalizations.

We also dig into the practical side: advance directives, POLST forms, and the true role of a healthcare proxy. Rather than replacing human decision-makers, we imagine a partner-in-the-loop model—where AI offers guidance, not verdicts, and transparency is key. Because when emergencies hit, it's not just about having a plan—it's about making sure your voice is heard.

If this resonates, take one step today: name your proxy, talk to your doctor, and share your wishes. Then subscribe, send this episode to someone who needs it, and leave a review to help keep these critical conversations alive.

Reference: 

Machine Learning–Based Patient Preference Prediction: A Proof of Concept
Georg Starke, et al.
NEJM AI (2025)

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

What happens when an algorithm—not a doctor or a claims reviewer—denies your surgery? A single decision like that can trigger a much bigger conversation about how AI is reshaping access to care.

In this episode, we dive into Medicare’s WISER pilot and the complex world of prior authorization. What’s the goal? Reduce waste and streamline approvals. But where does it go wrong—and how can we fix it? With insights from AI researcher Vasan Sarati and emergency physician Laura Hagopian, we unpack how claims data trains decision-making models, why black-box algorithms erode clinician trust, and what real safeguards look like in practice.

We spotlight three high-volume services—skin and tissue substitutes, electrical nerve stimulators, and knee arthroscopy for osteoarthritis—and explore why these procedures made the list. Knee arthroscopy, in particular, becomes our case study: widely performed, weak evidence in most OA cases, but not without its exceptions. That tension reveals deeper risks: overreliance on flawed data, quick “human reviews,” and denials that feel rubber-stamped.

Then we imagine a better way. What if AI could argue with itself before deciding? Enter multi-agent models—a system where different specialized AIs represent the patient, the provider, and the payer. They debate function, evidence, policy, and risk—and their decisions come with plain-language justifications, escalation triggers, and audit trails. The goal: approvals that are not just faster, but fairer.

If you care about timely access, fewer roadblocks, and smarter AI guardrails in healthcare, this episode is for you. Subscribe, share it with a colleague, and tell us what you’d change about AI-driven prior auth. We’ll highlight listener ideas in an upcoming show.

Reference: 

Private health insurers use AI to approve or deny care. Soon Medicare will, too.
Lauren Sausser and Darius Tahir
NBC News (2025)

WISeR (Wasteful and Inappropriate Service Reduction) Model
Cemter for Medicare and Medicaid Services 

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

What if AI could turn chaotic clinical notes into clean, billable codes—without sacrificing accuracy or trust?

Every shift, emergency physicians face the same grind: time-crunched documentation, symptom-first note-taking, and the constant lure of the “unspecified” box just to move on. But what if a system could read between the lines—and suggest precise, payer-accepted codes grounded in real guidelines?

In this episode, we explore how retrieval-augmented generation (RAG) is reshaping medical coding. Laura, an emergency physician, shares what it’s really like to code in the middle of clinical chaos. Vasanth, an AI engineer, explains why standard large language models often hallucinate ICD-10 and CPT codes—and how RAG brings the conversation back to solid ground with verifiable sources, official codebooks, and audit-ready citations.

We unpack a recent study comparing clinician-assigned codes to RAG-augmented outputs on actual emergency department charts. The results? When reviewers didn’t know which was which, they often chose the AI-generated codes—ones that captured true clinical meaning, like “alcoholic gastritis without bleeding” instead of the vague “epigastric pain.”

Beyond accuracy, we dive into the ripple effects: cleaner claims, fewer denials, stronger datasets for research—and the essential guardrails that keep things safe and ethical, from privacy safeguards to human review and confidence scoring.

If documentation has ever pulled you away from patient care, this episode offers a hopeful shift. Learn where retrieval-based coding tools fit into your EHR workflow, how clinicians can stay in the loop, and which high-volume complaints to tackle first for maximum impact.

Subscribe for more deep dives into clinician-centered AI, share this with the colleague who always codes “unspecified,” and leave us your biggest documentation headache—we’ll decode it next.

Reference: 

Assessing Retrieval-Augmented Large Language Models for Medical Coding
Eyal Klang et al.
New England Journal of Medicine (NEJM) AI, 2025

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

What if we could glimpse our future health—not through guesswork, but through data-driven forecasts? A new AI model, codenamed “Delphi,” is redefining what it means to predict disease by learning from massive, population-scale medical histories. Built on transformer architecture, Delphi estimates the risk and timing of over a thousand possible diagnoses—offering a personalized view of what may lie ahead.

We start with familiar ground—cardiovascular risk scores—and explore how predictions only matter when they guide meaningful actions: improved blood pressure control, appropriate statin use, and lifestyle changes that truly bend the curve. But Delphi doesn’t stop at single conditions. It captures the real-world complexity of multimorbidity, mapping how diseases co-occur and unfold over time.

Delphi doesn’t “understand” biology—it recognizes patterns. Much like a weather forecast, it turns complex statistical relationships into calibrated probabilities. We break down how the model handles irregular patient histories, simultaneous diagnoses, and time-to-event forecasting—offering practical insights clinicians can use. We also explore how Delphi was validated across extensive UK and Danish datasets, and why “reliable” beats “flashy” in the real world of medicine.

One of Delphi’s most promising features? Generative timelines. By simulating possible health futures from partial records, the model creates synthetic patients—fueling research while protecting privacy.

At the core is a human question: would you want to know your likely diagnoses decades in advance? We unpack the emotional and ethical dimensions of predictive health—when foresight helps, when it overwhelms, and how to responsibly deliver these insights. If you care about AI in healthcare, predictive analytics, or the ethics of foreknowledge, this episode offers a grounded look at what’s here, what’s coming, and how to use it wisely.

Reference: 

Learning the natural history of human disease with generative transformers
Artem Shmatko et al. 
Nature, 2025

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/

Can a smile reveal the early signs of Parkinson’s disease?

New research suggests it can—and AI is making that detection possible. Scientists are training machine learning systems to spot subtle facial changes associated with Parkinson’s, particularly in how we smile. These early signs, often missed by the human eye, could hold the key to faster, more accessible diagnosis.

Parkinson’s typically presents with tremors, muscle rigidity, and slowed movement. But it also affects facial muscles, leading to “hypomimia”—a loss of expressiveness where smiles become slower, less intense, and less spontaneous. Using the Facial Action Coding System, researchers broke down these expressions into measurable muscle movements like the “lip corner puller” and “dimpler,” allowing AI to analyze them with clinical precision.

Interestingly, models trained specifically on smile-related features outperformed those using broader facial data, showing that a targeted approach may yield better diagnostic results. This innovation blends expert medical knowledge with AI—not as a mysterious black box, but as a transparent and focused tool for real-world screening.

While promising, the technology isn’t without challenges. False positives and issues with lighting, camera quality, and cultural differences in facial expressions highlight the need for more testing before widespread use. Still, in clinical settings, especially where neurologists are scarce, this tool could offer meaningful support.

Tune in to explore how artificial intelligence is helping decode the smallest of human expressions—and what that might mean for the future of neurological care.

References:

AI‑Enabled Parkinson’s Disease Screening Using Smile Videos
T. Adnan, et al.
NEJM AI, 2025 

Automated video-based assessment of facial bradykinesia in de-novo Parkinson’s disease
Michal Novotny et al.
npj, Nature Digital Medicine, 2022

Detection of hypomimia in patients with Parkinson’s disease via smile videos
G. Su, et al.
Annals of Translational Medicine, 2021

Analysis of facial expressions in parkinson's disease through video-based automatic methods
Andrea Bandini et al
Journal of Neuroscience Methods, 2017

Credits: 

Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/