1st Winter School
ChatGPT & AI Apps for Medical Doctors

Snow-covered Mount Olympus framed by black pine branches

University of the Aegean · Winter School on Mount Olympus

1st Winter School “ChatGPT & AI Apps for Medical Doctors” (AIMedWINTER 2027)

Saturday 13 – Sunday 14 February 2027 · Litochoro, Pieria, at the foot of Mount Olympus, Greece

Venue
Naval Museum Hall, Municipality of Dion-Olympos, Litochoro
Hours
09:00–14:00 on both days
Attendance
Blended: in person or online
Fees
€300 in person · €200 online
Who it is for
Students and graduates of Medicine & Dentistry
Certification
Certificate of attendance · 0.5 ECTS
Register for the 1st Winter School of the University of the Aegean

What you will learn in two days

10 teaching units, hands-on practice with real clinical scenarios

Patient Electronic Health Records from ehealth.gov.gr (e-prescriptions, medical interventions, lab tests, etc.).

Full customisation of ChatGPT 6 and appropriate prompts for Advanced Clinical Reasoning Tasks.

Multimodal medical data: interpreting complex medical signals (ECGs, EEGs, etc.) and automatically summarising a patient record.

Drug-drug interactions with ChatGPT and biomedical databases (PubMed, DrugBank, etc.).

Med Gemini, Claude and dozens of AI apps for doctors (Perplexity, MedGemma, NotebookLM, etc.).

Agentic AI in Medicine: the ChatGPT Atlas AI Agent and intelligent agents that act autonomously.

Ethics & GDPR: the AI Act, protection of sensitive patient data and anonymisation techniques.

Interview with the Professor and a trainee doctor

Instructors

Panagiotis Symeonidis, Associate Professor, University of the Aegean, Scientific Coordinator
Biography

Panagiotis Symeonidis is Associate Professor at the School of Information and Communication Systems Engineering of the Aegean University, Greece from November 2020. Before, he was Assistant Professor at the Faculty of Computer Science (scientific sector INF/01) of the Free University of Bolzano, Italy, from November 2016 till November 2020. Before moving to Bolzano, he worked for 8 years as Adjunct Assistant Professor at the Department of Informatics of the Aristotle University of Thessaloniki, Greece. He received a B.Sc. degree in Applied Informatics from University of Macedonia at Thessaloniki in 1996. He also received a M.Sc. degree in Information Systems from the same university in 2004. He received his Ph.D. in Web Mining and Information Retrieval for Personalization from the Department of Informatics of the Aristotle University of Thessaloniki in 2008. His research interests include personalized health, AI and Medical Data Management. He is co-author of 3 international books, 2 Greek books, 6 book chapters, 32 journal publications and 46 conference/workshop publications. His published papers have received more than 3600 citations and has an h-index = 31, according to Google Scholar.

Christos Andras, Laboratory Teaching Staff, International Hellenic University
Biography

Christos Andras received a bachelor’s in applied informatics from Macedonia University, Greece in 1996 and PhD in Sociology of Internet Technology from the same university in 2009 and MSc in Network and Complexity, Aristotle University Thessaloniki in 2022. Lately he has worked as a Software Developer and System Administrator at the Greek Language Centre, Greece. Currently, he is Specialized and Laboratory Teaching Staff in Department of Industrial Engineering and Management - International Hellenic University, specialized in "Social Information Systems". He is teaching databases and information systems. He has developed commercial applications such as (MyBusiness E.R.P, www.epsiloncomp.gr), information system for the management of psychiatric clinics etc. He is specialized in programming in .NET environment and SQL SERVER databases.

The venue: Litochoro Naval Museum, at the foot of Mount Olympus

Classes take place in the hall of the Naval Museum of the Municipality of Dion-Olympos, in the centre of Litochoro. The hall has a projector, a screen and a microphone at every seat. Coffee, cookies and water are served during the breaks.

Detailed Programme — 10 Teaching Units (2 days × 5 hours)

Each day runs 09:00–14:00, with five 50-minute teaching hours, a coffee break and a short break.

Saturday 13 February 2027

Patient Electronic Health Record and Medical Data Management

TimeTeaching UnitDetailed ContentMaterial
09:00–09:50 (Teaching Unit 1) Patient Electronic Health Record Management

a. Electronic Health Records Management from https://ehealth.gov.gr/

i. Diagnoses
ii. Medical interventions
iii. Prescriptions
iv. Medications
v. Drug side effects
vi. Laboratory/Microbiology tests
vii. Patient health progress notes
viii. Patient demographic data

b. Health data types (vital signs, bio-indexes and bio-markers (PSA, ECG, etc.), ECGs, CT scans, MRIs and other complex medical signals)

c. Case Study of a Boston Hospital's Data (MIMIC III Medical Dataset)
i. Description
ii. Data Types (Clinical data (EHRs), ECGs, Chest X-Rays, Progress Notes, etc.)
iii. Database schema
iv. Application demonstration for real-time analysis and visualisation of hospital data
v. How to filter patients based on a diagnosis code list (ICD-10 and ICD-9)

15 SAMPLE Slides
09:50–10:40 (Teaching Unit 2) Using ChatGPT for Medical Images and Other Complex Medical Signals

How to use ChatGPT for interpreting and evaluating complex medical signals, such as electrocardiograms (ECGs). Automated generation of reports based on combined diagnostics and extraction of reasoned medical conclusions.

  • Electrocardiogram (ECGs)
  • Other complex medical signals (EEG, ultrasound, audiogram, etc.)
  • Optimal prompts for clinical support with complex medical signals
Provided on site
10:40–11:10 · Coffee break (coffee, cookies, water)
11:10–12:00 (Teaching Unit 3) Ethical Issues of AI in Medicine and Protection of Sensitive Medical Data

a) Artificial Intelligence and Ethical Issues in Medicine
i. European legislation: AI Act 2024/1689, Medical Device Regulation (MDR) 2017/745
ii. Fundamental principles of biomedical ethics
iii. Transparency, Justice and Impartiality of AI
iv. Patient Consent in Clinical Decision Making

b) Protection of Patients' Sensitive Personal Data
i. European legislation: General Data Protection Regulation (GDPR 2016/679)
ii. Anonymization Techniques
iii. Date shifting
iv. Generalization of medical data interactions
v. K-Anonymity
vi. Adding Noise
vii. Loss of Information
viii. Differential Privacy

13 SAMPLE Slides
12:00–12:50 (Teaching Unit 4) Using ChatGPT for Summarising Polymorphic Medical Data

a. Using ChatGPT for the evaluation of a patient's Electronic Health Record (EHR)

i. Patient demographic data
ii. Admission/visit date and reason for admission/visit
iii. History of present illness and past medical history
iv. Medications the patient is taking, and allergies
v. Social and family history
vi. Physical clinical examination, laboratory results
vii. Assessment and therapeutic plan

b. Using ChatGPT to generate a detailed explanation of a therapeutic regimen for a patient.

i. A list of all medications the patient should continue to take after their visit, along with dosage and duration instructions.
ii. Specific instructions for the patient's home care, such as dietary guidelines, physical activities and other precautions to consider.
iii. Information about the patient's next appointments with doctors or other healthcare professionals to monitor their condition.

c. Demonstration of AI software that predicts the optimal insulin dose for a diabetic patient using a Deep Reinforcement Learning algorithm.

13 SAMPLE Slides
12:50–13:10 · Break
13:10–14:00 (Teaching Unit 5) ChatGPT for Identifying Adverse Drug Reactions

Use of ChatGPT for finding undesirable drug-drug interactions based on new medicine findings on the Web (e.g. DrugBank, PubMed)

a. Undesirable Drug-Drug Interaction Checker
b. Drug Allergy (Get drug allergy and cross-sensitivities info)
c. Build evidence-based tailored treatment plan
d. Empower medical doctors with reliable drug information
e. Example of a ChatGPT prompt: “Take the list of my patient's medications and then tell me if there are any undesirable drug-drug interactions that I should be aware of by using the latest medicine publications on PubMed platform.”
f. AI Software to apply PCA to perform dimensionality reduction and bring into surface latent associations between drugs and undesirable drug-drug interaction/side effects.
g. Demonstration of AI Software from https://storm.genie.stanford.edu/ for the automated generation of a review article based on PubMed publications.

14 SAMPLE Slides

Sunday 14 February 2027

AI Algorithms for Medical Data Assessment, Clinical Decision Support and LLMs

TimeTeaching UnitDetailed ContentMaterial
09:00–09:50 (Teaching Unit 6) AI Algorithms for Medical Data Assessment and Clinical Decision Support

a. Innovative AI Algorithms in Biomedicine (e.g. AlphaFold)
b. Linear Regression and Logistic Regression
c. Decision Tree Algorithm and Random Forest Algorithm
d. AI Algorithms for Drug Re-purposing (using existing drugs for new diseases, e.g. repurposing old drugs against COVID)
e. Deep Reinforcement Learning based on DeepMind algorithms by Google
f. Deep Neural Networks (MLPs, CNNs, RNNs, GCNs, etc.)
g. AI Algorithms and ChatGPT for Bioindex Generation (bioindexes generator)
h. Application of AI Algorithms for the combined assessment of clinical findings and genetic characteristics (DNA) of the patient.

20 SAMPLE Slides
09:50–10:40 (Teaching Unit 7) Advanced Clinical Reasoning Using Google's Med Gemini

Example of a Clinical Case study given to Med Gemini:
"A 45-year-old male presents with a three month history of progressive shortness of breath, fatigue, and occasional chest pain. His medical history includes hypertension and diabetes. Physical examination reveals a heart rate of 95 bpm, blood pressure of 140/90 mmHg, and bilateral basal lung crackles. Laboratory tests show elevated B-type natriuretic peptide (BNP) levels and mildly reduced kidney function. What is the most likely diagnosis, and what further tests would you recommend to confirm it?"

This task requires the model to integrate the patient's symptoms, medical history, physical examination findings and laboratory results to suggest a likely diagnosis (such as heart failure) and recommend appropriate next steps for investigation (e.g. echocardiogram, chest X-ray).

a. Demonstration of several examples of complex clinical reasoning tasks made using Google's Med Gemini.
b. AI software to predict drug combinations for patients and explain the predictions using graph data (patient nodes, treatment node, drug node).

12 SAMPLE Slides
10:40–11:10 · Coffee break (coffee, cookies, water)
11:10–12:00 (Teaching Unit 8) Appropriate ChatGPT Prompts for Medical Doctors

a. What are ChatGPT prompts and why are they important?
b. Best practices for writing appropriate ChatGPT prompts (role-playing)
c. Setting constraints
d. ChatGPT prompts for after-visit remedy generation based on Word/Excel data

Example of a ChatGPT prompt:
"I have uploaded two files for you. Use the structure of the Word file, which documents the complete therapeutic regimen for one of my previous patients who had issues with blood sugar levels, liver, cholesterol/triglycerides, and vitamin D levels. Based on this file, generate a new detailed therapeutic regimen for a new patient of mine, who has a similar clinical condition, with clinical findings provided in the second uploaded Excel file."

e. Study of several different clinical cases along with the creation of adequate ChatGPT prompts
f. Clustering Algorithms
g. Software for clustering patients into groups with or without diabetes

15 SAMPLE Slides
12:00–12:50 (Teaching Unit 9) Artificial Intelligence Applications for Medical Doctors and the Internet of Medical Things (IoMT)

a. Precision Medicine and Personalised Medicine
b. Telemedicine and Telehealth Applications
c. Using "Speech to Text" applications to reduce the time spent on daily patient history documentation (e.g. OpenAI's Whisper software)
d. AI applications for carbohydrate calculation and prediction of additional insulin doses (Diabetes, GlucoCalc, Abbott’s FreeStyle LibreLink, etc.)

e. Digitised recording of daily vital signs (blood oxygen saturation, blood pressure, body temperature, heart and respiratory rate, ECG, Continuous Glucose Monitoring, etc.) for real-time monitoring and timely notification of doctors.
f. Wearables (smartwatches, rings, etc.): Samsung's Galaxy Ring, Medtronic glucose sensor for Continuous Glucose Monitoring (CGM)
g. Metrics for evaluating predictive models: Confusion Matrix, Precision, Recall, ROC curves, MAE and RMSE for Optimal Drug Dose Prediction
h. Demo of AI software that sends a patient's electronic health record to the ChatGPT API and gets a possible therapeutic prediction based on the given vital signs.

15 SAMPLE Slides
15 SAMPLE Slides
12:50–13:10 · Break
13:10–14:00 (Teaching Unit 10) ChatGPT Atlas, Agentic AI and Large Language Models (Transformers)

a) Configuring the ChatGPT Atlas AI Agent to act autonomously on the Web
i. Automatic form filling by the intelligent agent, such as e-prescription forms
ii. Operating the Atlas Agent in logged-in and logged-out modes
iii. Agentic AI in Medicine – Intelligent Agents in Medicine

b) Transformers
i. Attention and Self-Attention
ii. Large Language Models and ChatGPT
iii. Multi-modal Transformers and Google's Med Gemini
iv. ChatGPT and Retrieval Augmented Generation (RAG)
v. Explainability in Health

Example of a ChatGPT pre-prompt: "I have uploaded a Word file… Based on this template file, generate a new detailed therapeutic regimen for the patient under examination…"

Provided on site

Assessment and certification

Participants submit a composite assignment: the evaluation of an expanded patient medical record consisting of rich and polymorphic medical data (progress textual notes, CSV files, Word files, PDFs, ECGs, etc.) using AI techniques and ChatGPT.

On successful completion, a certificate of attendance with 0.5 ECTS is awarded (10 hours of attendance and 2.5 hours of assignment work).