Gustav Olaf Yunus Laitinen-Fredriksson Lundström-Imanov

Research

Research Areas

Work at the intersection of machine learning methods, domain expertise, and real-world deployment constraints, across six core research themes.

biotech

Medical & Clinical AI

AI systems for radiology, pathology, fetal ultrasound, and clinical decision support. Human-AI collaboration studies, clinician-in-the-loop evaluation protocols, and federated learning for privacy-preserving clinical research. Current PhD focus at the Technical University of Denmark.

Related publications

article

Uncertainty-Calibrated Explainable AI for Fetal Ultrasound Plane Classification

article

Transparency-First Medical Language Models: Datasheets, Model Cards, and End-to-End Data Provenance for Clinical NLP

visibility

Explainability & Trustworthy AI

Uncertainty-calibrated explainability methods including GradCAM variants, LIME-style surrogates, and uncertainty-weighted activation maps. Transparency frameworks, model cards, data provenance, and governance for responsible AI deployment.

Related publications

article

Uncertainty-Calibrated Explainable AI for Fetal Ultrasound Plane Classification

article

Transparency-First Medical Language Models: Datasheets, Model Cards, and End-to-End Data Provenance for Clinical NLP

code_blocks

Bioinformatics & Molecular Systems

Multi-omics integration, large-scale proteomics pipelines, WGS/WES workflows, and ML-driven biomarker discovery. Current PhD focus at University of Luxembourg on systems and molecular biomedicine.

hub

Foundation Models & LLMs

Evaluation methodology and architecture research for large language models and multimodal systems. Patch-based time series transformers, transparent clinical NLP, and mechanistic analysis of catastrophic forgetting.

Related publications

article

PatchFormer: A Patch-Based Time Series Foundation Model with Hierarchical Masked Reconstruction and Cross-Domain Transfer Learning for Zero-Shot Multi-Horizon Forecasting

article

Mechanistic Analysis of Catastrophic Forgetting in Large Language Models During Continual Fine-tuning

map

Geospatial & Urban AI

GeoAI frameworks for urban mobility, spatio-temporal traffic modeling, climate-resilient housing, and autonomous site selection. Evidence-based urban policy and public-sector analytics.

Related publications

article

Spatiotemporal Heterogeneity of AI-Driven Traffic Flow Patterns and Land Use Interaction: A GeoAI-Based Analysis of Multimodal Urban Mobility

article

Urban Spatio-Temporal Foundation Models for Climate-Resilient Housing: Scaling Diffusion Transformers for Disaster Risk Prediction

calculate

Scientific Computing & Physics ML

Physics-informed neural networks with Bayesian uncertainty quantification for PDEs. Multi-fidelity frameworks, neuromorphic edge computing, and graph neural networks for combinatorial optimization.

Related publications

article

Multi-Fidelity Physics-Informed Neural Networks with Bayesian Uncertainty Quantification and Adaptive Residual Learning for Efficient Solution of Parametric Partial Differential Equations

article

Energy-Efficient Neuromorphic Computing for Edge AI: A Comprehensive Framework with Adaptive Spiking Neural Networks and Hardware-Aware Optimization