Hospital Medicine · Family Medicine · Clinical AI
Jeremy Tabernero, MD
Hospitalist at Essentia Health-St. Mary's Medical Center in Duluth, Minnesota. Family Medicine board-certified. Founder of JTMDAI — writing about clinical AI, patient education, and the work of inpatient medicine as it actually happens.
Selected work
Case studies on clinical AI at the bedside, editorial notes, and systems built under JTMDAI.
Case Study AMBIENT SCRIBES
Ambient AI scribes
What the technology actually is, what the trade-offs are, and where my line is currently drawn.
NOT USINGCLINICAL AI 2026
Patient Education PAH RESOURCE
Pulmonary Hypertension RN — PAH education library
A patient and caregiver education resource on pulmonary hypertension and PAH, with plain-language guides on symptoms, diagnosis, treatment, centers, and living with the disease.
PATIENT EDUCATIONPAH 2026
Literature Watch EVIDENCE
CMS coverage policy shifts for digital health and remote patient monitoring (RPM)
1. Post-discharge RPM becomes billable — A hospitalist can now set up 7-day post-discharge monitoring (e.g., CHF weight + vital signs) and bill for it, where the old 16-day…
HOSPITAL MEDICINEEVIDENCE 2026
Case Study DECISION SUPPORT
Clinical decision support
The narrow places where decision-support tools have earned a seat at the bedside.
CURRENTLY USINGCLINICAL AI 2026
Note VALIDATED ONLY
Why I only adopt clinically validated tools
There is nothing courageous about waiting for evidence. It is the baseline of how medicine has always advanced.
CLINICAL AINOTE 2026
Case Study EXPLAINABLE AI
Explainable AI in clinical monitoring
I fact-checked a viral video about a hallucination-proof ICU monitoring agent. The verdict is a case study in exactly the problem the video describes.
WATCHINGCLINICAL AI 2026
Literature Watch EVIDENCE
FDA AI/ML-enabled medical device clearances and SaMD guidance changes
1. Single-output tools can now avoid device regulation — A physician building a drug recommendation system can provide one answer (e.g., "Consider lisinopril") without listing…
HOSPITAL MEDICINEEVIDENCE 2026
AI & Algorithms FOUNDATIONS
Foundations: What Makes an Algorithm Work
Algorithms, complexity, and data structures explained plainly — and why validating one on real data is the whole game.
LEARNINGAI & ALGORITHMS 2026
JTMDAI HOBBY LAB
JTMDAI — my personal AI & automation lab
My personal AI & automation lab — a homepage for the things I build for myself after hours. A hobby, not a business or a clinical service.
HOBBYAI & AUTOMATION 2025
Case Study GEN-AI COMMS
Generative AI in clinical communication
Where generative models help patients understand their care, and where they replace the wrong part of the conversation.
EVALUATINGCLINICAL AI 2026
Literature Watch EVIDENCE
Acute kidney injury prevention in hospitalized adults
Clinical decision support gap: Most EHRs lack real-time nephrotoxic burden calculators. A CDS tool that flags cumulative nephrotoxin exposure (vanc + NSAID + contrast + ACEi +…
HOSPITAL MEDICINEEVIDENCE 2026
Case Study HOSPITAL AT HOME
Hospital at home
Where home-hospital programs make clinical sense, and where they are being deployed for the wrong reason.
WATCHINGCLINICAL AI 2026
Note AGAINST SCRIBES
The case against ambient AI scribes at the bedside
Ambient AI scribes are coming for the patient encounter, and the conversation about them has been almost entirely one-sided.
CLINICAL AINOTE 2026
Case Study PATIENT DATA
Patient data
Who owns the chart, who profits from it, and what that means for the patient I am about to see in room 412.
WATCHINGCLINICAL AI 2026
Literature Watch EVIDENCE
Hospital-acquired venous thromboembolism prophylaxis
The 40% problem is a systems gap, not a knowledge gap. Validated risk models exist (2024 International Consensus covers medical, surgical, pregnant, postpartum, cancer cohorts).…
HOSPITAL MEDICINEEVIDENCE 2026
AI & Algorithms GRAPHS & ML
Graphs & Machine Learning: Finding Structure and Making Predictions
Graph search and ML — from Dijkstra and PageRank to sepsis phenotypes and cardiac risk scores.
LEARNINGAI & ALGORITHMS 2026
Guardian One PERSONAL PROJECT
Guardian One — a multi-agent system for my own life
A multi-agent system I build for my own life — finance, scheduling, security, and smart-home tasks under one encrypted, audited setup. It manages my data, nobody else's.
HOBBYAGENTS 2025
Case Study FIVE REAL CLINICAL-AI
Five real clinical-AI deployments, and the line that separates proof from promise
A survey of five active clinical-AI technologies, sorted by what they've actually proven in trials, not what they claim in pitches.
EVALUATINGCLINICAL AI 2026
Literature Watch EVIDENCE
Inpatient hyperglycemia management (non-ICU)
Implementable protocol: 1. Trigger: 2+ BG readings 180 in 24h → auto-flag for scheduled insulin 2. Order set: basal (glargine/detemir QD or NPH BID) + prandial (lispro/aspart AC)…
HOSPITAL MEDICINEEVIDENCE 2026
Case Study VALUE-BASED CARE
Value-based care
What 'value-based' actually means at the bedside, and what it tends to mean in the slide deck.
WATCHINGCLINICAL AI 2026
Note AI AS TEACHING
AI as teaching, not transcription
If ambient scribes are the version of clinical AI I am most cautious about, patient-facing visualization is the version I am most excited about.
PATIENT EDUCATIONNOTE 2026
Case Study WORKFORCE SHIFTS
Workforce shifts
What is actually changing about who does which clinical work, and what that means for the patient at the bedside.
WATCHINGCLINICAL AI 2026
Literature Watch EVIDENCE
Safe transitions of care and 30-day readmission reduction
High-yield intervention bundle: 1. Risk stratify at admission (LACE+ or HOSPITAL score) → flag high-risk patients for enhanced discharge planning 2. Patient education (teach-back…
HOSPITAL MEDICINEEVIDENCE 2026
AI & Algorithms DEEP LEARNING
Deep Learning & Sequences: From Neural Nets to ChatGPT
Neural nets to transformers, plainly: the ideas and the verified milestones behind ChatGPT.
LEARNINGAI & ALGORITHMS 2026
Pantheon TINKERING
Project Pantheon — my autonomous-agent hobby project
My autonomous-agent hobby project — nights-and-weekends experiments in explainable decisions, reversible actions, and agents that learn from their own mistakes.
HOBBYAGENTS 2026
Literature Watch EVIDENCE
Sepsis-3 fluid resuscitation in the first 24 hours
Protocol-ready recommendation: 1. Sepsis-3 criteria met + hypoperfusion/shock → start crystalloid resuscitation immediately 2. Fluid choice: Lactated Ringer's or Plasma-Lyte…
HOSPITAL MEDICINEEVIDENCE 2026
AI & Algorithms APPLIED AI
Applied Systems: Recommenders, Cryptography, Scale, and Responsibility
Recommenders, crypto, scale, and AI ethics explained plainly, with real cases and healthcare stakes.
LEARNINGAI & ALGORITHMS 2026
AI & Algorithms NAIVE BAYES AND
Naive Bayes and Thinking in Probabilities
Bayes' rule as belief-updating — the spam filter and bedside diagnostic reasoning are the same math.
LEARNINGAI & ALGORITHMS 2026
AI & Algorithms LATENT SPACE: THE
Latent Space: The Map of Meaning
Embeddings place words, documents, and patients on a map where distance means similarity of meaning.
LEARNINGAI & ALGORITHMS 2026
AI & Algorithms GAUSSIAN SPLATTING AND
Gaussian Splatting and the Jacobian: From Bell Curves to 3-D Worlds
The bell curve you know from lab ranges, stacked in 3-D and projected through a Jacobian, becomes photorealistic scene capture — and the same math is how networks learn.
LEARNINGAI & ALGORITHMS 2026