Rawshani Group
Gothenburg Cardiometabolic Research Group
Group
Vibha Gupta (vibha.gupta@gu.se), PhD, Engineer, post doc
Adam Piasecki, (adam.piasecki@vgregion.se), Consultant cardiologist, MD, PhD, post doc
Linnea Gustafsson (linnea.gustafsson@vgregion.se), Internal medicine specialist, MD, PhD student
Daniella Isaksén (daniella.isaksen@vgregion.se), Consultant cardiologist, MD, PhD student
George Lappas (george.lappas@gu.se), senior biostatistician
Johan Widing, MD cand, Master student
Pedram Sultanian, MD, PhD, post doc
Erik Andersson, MD cand, Researcher
Lukas Hilgendorf, MD, PhD student
Alfred Hjalmarsson, MD cand, Researcher
Arman Shahmari, MD cand, Researcher
Principal investigator (PI)
Araz Rawshani, MD, PhD
Professor of Cardiology, Consultat Cardiologist
Wallenberg Centre for Molecular and Translational Medicine, Institute of Medicine, University of Gothenburg; Department of Cardiology, Sahlgrenska University Hospital
Ongoing projects
AI Driven Metabolic Disease Atlas
We’ve created a fully AI-driven atlas for RCTs in metabolic diseases, starting with obesity drugs. The atlas uses an agentic framework with language models to extract and reason around trial results. It allows for on the demand meta-analyses of primary and secondary outcomes. The reasoning engine is currently being developed.
TrialNet - Efficient Inclusion of Participants to Randomized Clinical Trials
Our TrialNet project has now enrolled 2 randomized controlled trials (Balanced-HF [AstraZeneca], Prevent-HF [AstraZeneca]). TrialNet aims to use digital infrastructure and AI to identify patients for clinical trials with >90% precision from EHRs, and allows for immediate digital contact with patients.
Proteomic Signatures of Diabetes Across the Obesity Spectrum
Using UK Biobank data, we investigate how type 2 diabetes shapes the circulating proteome and metabolome across varying degrees of obesity. We explore distinct biomarker patterns by adiposity and links these to genetic risk for obesity and diabetes.
Deep Learning for the Detection of Coronary Artery Stenosis, Flow-Limiting Plaques and Vulnerable Plaques
Project using 18,000 coronary CT angiographies to create fully-automated interpretation of coronary CT scans.
Vibha Gupta, Lukas Hilgendorf, Erik Andersson
Funded by Novo Nordisk Foundation, Wallenberg Centre for Molecular and Translational Medicine, ALF VGR, Vetenskapsrådet.
Cardiometabolic Drug Discovery with Boltz‑2
Harnessing the power of Boltz‑2, a cutting-edge AI model for protein–ligand interaction and binding affinity prediction, to explore small molecules targeting diabetes and obesity. By combining structural precision with high-throughput virtual screening, we explore the discovery of next-generation therapies for cardiometabolic diseases.
Arman Shahmari.
Fully automated ECG interpretation of acute coronary syndromes
Phd cand: Lukas Hilgendorf, MD
Digital Cardiometabolic Graphs
We explore methods for automatic ontology creation for cardiometabolic disease databases. By combining expert curation with scalable automation for onotology creation and construction of knowledge graphs, the goal is to build a foundation for next-generation AI models that can reason across multi-omics for CMD.
Johan Widing
Agentic Clinical Decision Support
Our Cardiology Consultant agent is now live and being tested at 2 hospitals in Sweden.
Justus Råmunddal
Clinical prediction models for cardiovascular care
Due to new EU regulations (MDR, AI Act) these models are no longer available online.
SCARS-1
Predict 30-days survival after out-of-hospital cardiac arrest.
SCARS-2
Predict 30-days survival after in-of-hospital cardiac arrest.
SCARS-3
Predict long-term survival at discharge after OHCA.
The AI in Healthcare Conference
We hosted a fully booked AI in Healthcare Conference. A sequel is coming in Aug 2026.
