Physician Researcher · Practical Medical AI

Nariman
Naderi.

I translate clinical questions into rigorous medical AI research and practical evaluation workflows.

Current work: medical LLM reliability, clinical guideline synthesis, and medical imaging—grounded in gastroenterology and real healthcare constraints.

View publications → Explore projects

CLINIC_TO_SYSTEM CURRENT

Research workflow

From a clinical question to a tested workflow with explicit failure boundaries.

01 Clinical problem Grounded in care
02 Rigorous evaluation Reliability first
03 Practical system Built for workflow

01 / Clinical base: Gastroenterology & Liver Research Center, Taleghani Hospital
02 / Research focus: LLM reliability · guideline synthesis · medical imaging
03 / Implementation: Python workflows · evaluation pipelines · local prototypes

About

Clinical insight, rigorous evaluation, practical systems

My role is to connect the clinical question with the research design, technical workflow, and failure analysis needed to evaluate a medical AI system responsibly.

I am a physician researcher at the Gastroenterology and Liver Research Center, Taleghani Hospital. My work focuses on medical AI in gastroenterology and extends to clinical large language models, guideline synthesis, computer vision, and imaging-dataset evaluation.

I earned my MD from Shahid Beheshti University of Medical Sciences in 2025 and have worked on AI-in-medicine research since 2022, including two first-author studies on medical LLM reliability and prompt evaluation.

I am most interested in systems that are useful under clinical constraints: traceable enough to audit, rigorous enough to study, and practical enough to fit real healthcare workflows.

Methods Python · LLM evaluation · retrieval-augmented generation · medical imaging · data workflows · lightweight clinical tools

Research

Current themes

Four connected themes centered on reliability, evidence synthesis, medical imaging, and clinical relevance.

01 / reliability

LLM confidence and uncertainty

Evaluating how medical large language models express confidence, uncertainty, and calibration in clinical reasoning tasks.

Reliability / calibration

02 / synthesis

Clinical guideline RAG

Using retrieval-augmented generation to synthesize overlapping or conflicting clinical guidelines into structured outputs.

Guidelines / synthesis

03 / vision

Vision-language models

Studying when general-purpose vision-language models can compete with trained models in medical image tasks.

Endoscopy / VLMs

04 / data

Imaging dataset quality

Assessing medical imaging datasets for clinical relevance, bias, redundancy, and real-world applicability.

Dataset realism / bias

Canonical scholarly works

First-author publications

Active research themes

Domain clusters · GI · LLM · imaging

Publications

Selected research outputs

Google Scholar →

2026

Peer-reviewed journal article · co-first author

Across generations, sizes, and types, large language models poorly report self-confidence in gastroenterology clinical reasoning tasks.

npj Gut and Liver · Co-first author.

View DOI ↗

2026

Peer-reviewed journal article · guideline RAG

Using large language models to integrate international IBD guidelines: A retrieval-augmented generation approach.

Colorectal Disease · Co-author.

View DOI ↗

2025

Peer-reviewed journal article · vision-language models

Vision language models versus machine learning models performance on polyp detection and classification in colonoscopy images.

Scientific Reports · Co-author.

View DOI ↗

2025

Conference paper · first author

Evaluating prompt engineering techniques for accuracy and confidence elicitation in medical LLMs.

EXTRAAMAS · Lecture Notes in Computer Science · First author.

View DOI ↗

2025

Preprint · dataset review

State of abdominal CT datasets: A critical review of bias, clinical relevance, and real-world applicability.

arXiv · Co-author.

View arXiv ↗

Projects

Selected projects

Two public research repositories connecting peer-reviewed medical AI studies with inspectable code, processed data, and evaluation workflows.

Open research code + data

LLM confidence analysis repository

Code, processed data, and analysis notebooks supporting the 2026 npj Gut and Liver study of self-reported confidence across 300 gastroenterology questions and 48 paired model entries.

Role: First author · statistical analysis · manuscript development
Methods: Python notebooks · calibration · Brier score · ROC-AUC

Public code, processed data, and citation metadata

View code and data ↗

Open research code

Medical LLM prompt-engineering study

Code and prompt templates supporting the 2025 Springer study of 156 evaluated configurations across five models, three temperatures, multiple prompt strategies, and confidence formats.

Role: First author · study implementation · evaluation
Methods: Batch APIs · structured outputs · prompting · calibration

Public code, prompt templates, and citation metadata

View code and prompts ↗

Insights

Short notes on medical AI systems

Brief field notes connecting research questions, implementation choices, and clinical constraints.

Guideline RAG

Research collaboration in medical AI

I welcome collaboration on medical LLM evaluation, clinical guideline synthesis, and medical imaging research. Email is the best way to start.

Email me → Google Scholar

GitHub ORCID LinkedIn X / Twitter

Nariman
Naderi.

Clinical insight, rigorous evaluation, practical systems

Current themes

LLM confidence and uncertainty

Clinical guideline RAG

Vision-language models

Imaging dataset quality

Selected research outputs

Across generations, sizes, and types, large language models poorly report self-confidence in gastroenterology clinical reasoning tasks.

Using large language models to integrate international IBD guidelines: A retrieval-augmented generation approach.

Vision language models versus machine learning models performance on polyp detection and classification in colonoscopy images.

Evaluating prompt engineering techniques for accuracy and confidence elicitation in medical LLMs.

State of abdominal CT datasets: A critical review of bias, clinical relevance, and real-world applicability.

Selected projects

LLM confidence analysis repository

Medical LLM prompt-engineering study

Short notes on medical AI systems

Making medical guidelines usable with RAG

The practical future of medical AI

Medical imaging AI beyond the model

Research collaboration in medical AI

NarimanNaderi.

Clinical insight, rigorous evaluation, practical systems

Current themes

LLM confidence and uncertainty

Clinical guideline RAG

Vision-language models

Imaging dataset quality

Selected research outputs

Across generations, sizes, and types, large language models poorly report self-confidence in gastroenterology clinical reasoning tasks.

Using large language models to integrate international IBD guidelines: A retrieval-augmented generation approach.

Vision language models versus machine learning models performance on polyp detection and classification in colonoscopy images.

Evaluating prompt engineering techniques for accuracy and confidence elicitation in medical LLMs.

State of abdominal CT datasets: A critical review of bias, clinical relevance, and real-world applicability.

Selected projects

LLM confidence analysis repository

Medical LLM prompt-engineering study

Short notes on medical AI systems

Making medical guidelines usable with RAG

The practical future of medical AI

Medical imaging AI beyond the model

Research collaboration in medical AI

Nariman
Naderi.