Move from ChatGPT to a working system
Set up Codex, define safe boundaries and turn your instructions into reusable AGENTS.md files and skills.
Practical online training · first edition
AI for medical research
From ChatGPT to Codex: scientific evidence, clinical data and reproducible analyses.
José Daniel Subiela
MD, PhD · Consultant uro-oncologist and clinical professor
Jesús López
Applied AI, data and instructor
600 € · refundable deposit of 100 €
Reserve for €100
30–31 October · 15 seats · deposit refundable until 16 October.
Clinical practice notebook IA · 01
01 · Evidencia
Summarize a paper without losing clinical judgement.
02 · Datos
Organize scattered information and build a useful collection sheet.
03 · Decisión
Turn a real need into a workflow you can repeat.
Immediate application
You will leave with methods, not theory.
Interpret evidence systematically
Move from a paper to a traceable critical reading, an evidence matrix and a synthesis that another physician can review.
Describe your clinical data
Build a collection sheet, harmonize sources and obtain auditable descriptive statistics from a practice dataset.
Understand applied models
Work through survival analysis and surgical learning curves, with explicit validation, bias and interpretation criteria.
Who it is for
For people who turn medical questions into knowledge.
You do not need to code: you need a question, judgement in your field and the willingness to work with a verifiable method.
01
Clinicians
Practising physicians and residents who want to turn clinical questions into structured, reviewable work.
02
Clinical and biomedical researchers
People who review literature, design studies, build registries or prepare data for analysis and publication.
03
Postgraduate students and PhD candidates
Master's students, PhD candidates and early-career academics developing theses, papers or research projects.
Cases from clinical research and practice
Explore five real examples of what you will learn to build.
Five demonstrations José built from his own clinical and research work. In the course you build your own with the same method.
Featured demonstration
From a paper to a reviewable evidence matrix
José frames the clinical question, Jesús builds the workflow in Codex, and both review what the system can and cannot conclude.
Video in preparation01:24
clinical-review.logTRACE · ON
01 · QuestionPICO estructurada
02 · Extraction12 variables
03 · Review3 alertas
PopulationMIBC · cT2–T4a
OutcomeOverall survival
LimitationSelection bias
Human validation before accepting conclusions
Network meta-analysis
Compare perioperative treatments in muscle-invasive bladder cancer
Three relevant regimens have no single direct trial that answers how they compare in efficacy and safety.
- Deliverable
- Interactive indirect comparisons, treatment rankings, safety overview and trial table.
- Validation
- The artifact separates efficacy, safety and the limitations of indirect evidence.
Course program
Two live sessions, four working blocks.
Two 3-hour sessions (Fri 30 & Sat 31 Oct), each with explanation, guided practice and a reusable deliverable. The small group ensures everyone leaves with their own working routine.
Recorded onboarding from 23 Oct, plus daily office hours (26–30 Oct) to unblock setup and questions. All sessions recorded for participants.
Session 1 3 h
your Codex working system and scientific evidence
Live online
Session 2 3 h
clinical data and applied models
Live online
01
Your Codex working system
Move from ChatGPT conversations to a Codex workspace with safe boundaries, reusable instructions and skills.
- 1.1 What AI can do and where it must stop 25 min
- 1.2 Prepare a safe environment with Codex 35 min
- 1.3 Turn your method into reusable instructions and skills 35 min
02
AI-assisted scientific evidence
Use AI to structure questions, read papers critically and produce traceable evidence syntheses.
- 2.1 Turn a clinical question into a source pack 35 min
- 2.2 Read a paper critically with AI 40 min
- 2.3 Create an evidence matrix and useful synthesis 40 min
03
Clinical databases and descriptive statistics
Turn clinical objectives into operational variables, auditable datasets and descriptive analyses physicians can interpret.
- 3.1 Define operational clinical variables 35 min
- 3.2 Build a data collection sheet with AI 40 min
- 3.3 Harmonize data and obtain descriptive statistics 40 min
04
Survival models and learning curves
Understand two applied medical analyses through real cases, with explicit validation, bias and interpretation criteria.
- 4.1 Frame an applied medical modelling question 30 min
- 4.2 Build and interpret a survival analysis 45 min
- 4.3 Analyse a surgical learning curve 45 min
A collaboration tested on real medical data
Years building medical ML models together, now accelerated with Codex.
José and Jesús were validating models with clinical data before today's AI. The course turns that into a method you can apply without losing traceability or clinical judgement.
Teaching team
José Daniel Subiela
MD, PhD · Consultant uro-oncologist and clinical professor
Consultant uro-oncologist and kidney transplant surgeon at Hospital Universitario Ramón y Cajal, and clinical professor at Universidad de Alcalá. His research combines real-world clinical data, biostatistics and artificial intelligence, including survival models, XGBoost and causal inference.
Jesús López
Applied AI, data and instructor
Founder of datons and LinkedIn Learning instructor. Years turning medical data into reproducible analyses, tools and models.
First edition
A format built for practice.
- Format
- Two live online sessions · 6 h total
- Live sessions
- Fri 30 & Sat 31 October 2026 (3 h each)
- Onboarding
- Step-by-step install videos, available 23 October
- Office hours
- 1 hour a day to unblock setup and questions, 26–30 October
- Recordings
- All sessions recorded and included
- Group
- Maximum 15 participants
- Reservation
- €100 deposit, refundable until 16 October; €500 balance due by then
- Price
- 600 €
- Reservation
- 100 € Refundable and deducted from the final price
Registration
Reserve your seat
Reserve with a €100 deposit, refundable until 16 October and deducted from the €600 price; the balance is due by that date to keep your seat. We work only with synthetic or publication-derived data, never real patient data.