Dana Mitchell

RN, BSN, MSEd, CCRN  Β·  PhD Student, Nursing Science  Β·  University of Rochester

ICU nurse educator at Thompson Health. PhD student at the University of Rochester School of Nursing, UR HEART (Heart, Electrophysiology, and Arrhythmia Research Team) (Advisor: Dr. Carey). Six years of cardiac ICU practice at Strong Memorial β€” ECMO, LVAD, IABP, Impella, cardiac transplant. MSEd in Literacy, BSN from the University of Rochester. Building clinical AI tools alongside the research and the teaching.


Background

Clinical depth.
Research rigor.

Clinical: Six years in the Cardiac ICU at Strong Memorial Hospital (University of Rochester Medical Center) β€” one of the leading cardiac programs in the country. Managed ECMO, LVAD, IABP, Impella, CVVHD, and cardiac transplant patients. Currently ICU nurse educator at Thompson Health.

Academic: PhD student in Nursing Science at the University of Rochester School of Nursing, UR HEART (Heart, Electrophysiology, and Arrhythmia Research Team) (Advisor: Dr. Carey). BSN from the University of Rochester. MSEd in Literacy (birth–adult). BS in Childhood Education. Research focus: implementation science, alarm fatigue, qualitative methods, AI in clinical education.

Building: Using AI tools β€” Claude Code, Codex, OpenClaw, and locally hosted Ollama models β€” to build clinical education software. Self-hosted infrastructure on a home lab network: Raspberry Pi gateway, Linux servers, RTX 3080 workstation, Mac mini M4 for iOS development. Deployed tools used by real nurses and students, not prototypes. The MSEd is why the learning design is grounded in actual science. The ICU experience is why the clinical content is accurate.

"Clinical credibility and learning science in the same person β€” that combination is rare. The tools I build reflect both."


Clinical AI Tools

Tools built from
clinical necessity.

Every tool here came from a gap I experienced personally β€” at the bedside, in the classroom, or in my own research. No dev team. No vendor. Built with AI collaboration and deployed to people who actually use them.

🫁
GasLab β€” ABG Interpreter
Step-by-step ABG interpretation with Henderson-Hasselbalch reasoning, Winter's formula, A-a gradient, and clinical guidance. iOS app with challenge mode coming 2026.
● abg.danamitchell.icu ⬑ iOS Coming
❀️
Cardiac Dysrhythmia Curriculum
Interactive rhythm instability presentation with live audience voting and a self-guided mode. AFib, CHB, VTach, junctional. Delivered to Thompson ICU staff, March 2026.
● Delivered β€” Thompson ICU
πŸ₯
Thompson ICU Education Hub
Centralized education portal: dysrhythmia curriculum, 21 preceptor guides, clinical resources. Guides are Thompson-specific (Pyxis, MOLST, Stryker LifePak). Phase 1 delivered; Phase 2 pending clinical review.
● thompson.danamitchell.icu
πŸ”¬
Sieve β€” Research Platform
Systematic/scoping review platform with AI-assisted screening and a secondary AI reviewer for human validation. Django + PostgreSQL. Used in my alarm fatigue scoping review.
● sieve.danamitchell.icu
πŸ’“
NightWatch ECG Trainer
Animated ECG rhythm strip generator and trainer β€” NSR, AFib, VTach, CHB, junctional, and more. Web version live, iOS (SwiftUI) Phase 2 complete.
● ecg.danamitchell.icu β—Ž iOS Phase 2
🧩
BrainRAG β€” Knowledge Base
RAG system for research: ingests PDFs/DOCX/notes, indexes with ChromaDB, answers queries with source citations. Session-based workspaces. Multi-LLM: Claude, Gemini, GPT-4o, or local Ollama.
● brain.danamitchell.icu β—Ž Refining
🧠
LANTERN β€” CCRN/ICU Prep
Adaptive 8-level learning engine grounded in ZPD, cognitive load theory, and spaced repetition. iOS in development.
β—Ž In Development
πŸ›‘οΈ
PI Prevention Tool
Clinical decision support for pressure injury prevention. Nurse-driven assessments and evidence-based protocols. Pending ICU staff review.
β—Ž In Development

Newsletter

Clinical Signal Radar

Weekly intelligence for critical care nurses β€” high-relevance research from NEJM, Journal of Nursing Scholarship, and critical care journals, translated into bedside-ready signals. Generated with AI, reviewed by a nurse educator.

Latest β€” Issue #7 Β· March 17–27, 2026
Workforce dynamics, whole blood in traumatic hemorrhage, digital burnout, servant leadership
Nursing Retention Whole Blood RCT Digital Burnout
Read Latest Issue β†’
View all issues β†’ Watch the AI generate one live β†’

Research

The evidence behind
the practice.

UR HEART (Heart, Electrophysiology, and Arrhythmia Research Team), University of Rochester School of Nursing. Advisor: Dr. Carey. My research sits at the intersection of clinical nursing, implementation science, and AI.

Qualitative Study β€” Submitted March 2026
Nurses' Experiences with Alarm-Management Interventions in Adult Cardiac ICUs: A Qualitative Descriptive Study
Qualitative descriptive design (Sandelowski). Semi-structured interviews. Conventional content analysis (Hsieh & Shannon). CFIR implementation science framework (Damschroder et al.). Addresses a critical gap: 85–99% of ICU alarms are non-actionable, yet no studies have examined how nurses specifically experience implementing alarm-management interventions in cardiac ICUs. University of Rochester School of Nursing, NUR 518.
Scoping Review β€” Draft Complete
Interventions to Reduce Non-Actionable Physiologic Monitor Alarms in Adult Cardiac ICUs
PRISMA-ScR protocol. Systematic search across CINAHL, PubMed, Embase, Cochrane. Mapping what interventions exist, what populations were studied, and where the evidence gaps are β€” particularly in cardiac ICU settings. Conducted using Sieve, a custom AI-assisted screening platform.
Manuscript β€” In Preparation
The HAC Framework: Human-AI Collaboration in Clinical Nursing Education
Conceptual framework for how nurses can evaluate, adopt, and collaborate with AI tools in clinical practice. Grounded in literacy science, epistemology, and six years of ICU bedside experience.
Manuscript β€” In Preparation
Adaptive Learning in Critical Care: A Literacy-Science Approach to CCRN Preparation
Applies Vygotsky's ZPD, Baddeley's working memory model, and cognitive load theory (Sweller) to the design of adaptive clinical education tools. Draws on MSEd in Literacy and bedside clinical authority.

Speaking

Taking this
to the room.

Oct2026
From Chatbot to Colleague: Building Clinical AI Tools as a Bedside Nurse
AACN Mountain to Sound 2026 Β· Seattle, WA Β· National Conference
Upcoming Β· Two sessions confirmed
Sep2026
AI in Clinical Nursing Practice
AACN NYC Chapter Β· New York, NY Β· Chapter Presentation
Upcoming
Mar2026
Rhythm Instability in the Cardiac ICU
Thompson Health ICU Β· Canandaigua, NY Β· Staff Education
Delivered β€” with live AI-powered audience participation

Infrastructure

The home lab behind
the tools.

Everything I build runs on self-hosted infrastructure β€” a home lab network I designed and maintain. This isn't just about the software; understanding the full stack from hardware to deployment is part of how I evaluate AI systems and their clinical integration. All machines are linked via Tailscale mesh VPN β€” accessible worldwide, fully interconnected.

πŸ₯§
Prometheus β€” Raspberry Pi 5 Gateway
Always-on gateway server running OpenClaw (AI agent framework), Hailo-10H AI accelerator (2nd gen) with locally hosted Ollama models (Qwen 2.5, DeepSeek, Llama 3.2), an AI-integrated quiz/presentation system (QuizForge), and Navidrome music server. Manages automated backups and all service orchestration.
πŸ–₯️
Atlas β€” Linux Server
Primary deployment server. Docker-based hosting for Sieve, BrainRAG, Emily's Learning Studio, the Seattle demo infrastructure, and the Discourse community platform. Runs Pi-hole DNS and the Cloudflare tunnel for all public-facing services.
⚑
Laplace β€” AI Workstation (Custom Build)
Custom built. AMD Ryzen 9 9950X3D on ASUS ROG Crosshair Hero (AM5), 32GB DDR5, NVIDIA RTX 3080 Sakura Edition (10GB VRAM). Set up for local AI inference and coding (Ollama, Codex), image generation (Stable Diffusion), and music production β€” all running on the 3080. Tailscale networked for remote access across the lab.
🍎
Newton β€” Mac mini M4
iOS/macOS development machine. Xcode builds for GasLab, NightWatch ECG, and upcoming iOS apps. Also runs local Ollama models (Qwen 2.5 Coder 14B, GLM-4).
πŸ–₯️
Janus β€” Legacy Workstation (Custom Build)
First custom PC build, ~10 years old and still running strong. Plex media server. Currently being converted from Windows 10 to Fedora Linux to join the lab network as an additional compute node.
🎡
Orpheus β€” Raspberry Pi 5 DAC
Dedicated high-quality audio endpoint. Raspberry Pi 5 with DAC HAT for lossless music playback. Part of the self-hosted media stack alongside Navidrome on Prometheus.
πŸ“¦
M75 β€” ThinkCentre (Pending Deployment)
Lenovo ThinkCentre M75 β€” compact, capable, and about to be assigned a role in the lab. Deployment TBD.
πŸ’»
Hermes β€” Windows 11 Laptop
Mobile workstation for on-the-go development and remote access to the full lab network via Tailscale.