Samuel Kaski A

Machine learning on magnetoencephalography data yields generalizable low-dimensional neural fingerprints that distinguish individuals across task conditions

The paper uses lnBRRR to create low-dimensional neural fingerprints from MEG data that distinguish individuals across tasks, showing they are generalizable and intrinsic.

HAPNEST: efficient, large-scale generation and evaluation of synthetic datasets for genotypes and phenotypes

HAPNEST generates and evaluates large-scale synthetic genotype and phenotype datasets for polygenic risk score analysis.

The influence of inter-regional delays in generating large-scale brain networks of phase synchronization

This study compares Biophysical Network Models (BNMs) with different delay specifications to determine their role in generating alpha-band phase synchronization networks in MEG data.

Gene-Gene Interaction Detection with Deep Learning

This paper introduces a deep learning method to detect gene-gene interactions by considering all SNPs and complex interactions, outperforming existing methods and identifying novel interactions in a cholesterol study.

Improving Neural Networks for Genotype-Phenotype Prediction Using Published Summary Statistics

This paper introduces a regularization method called main effect prior (MEP) to improve genotype-phenotype prediction using GWAS summary statistics, showing improved performance on small datasets.

Biophysical network models of phase-synchronization in MEG resting-state

This study investigates how biophysical network models can replicate MEG-derived phase-synchronization networks, finding that conduction delays do not improve the match but structural connections and Wilson-Cowan dynamics do.

Block HSIC Lasso: model-free biomarker detection for ultra-high dimensional data

The paper introduces block HSIC Lasso, a nonlinear feature selection method for ultra-high dimensional data, showing it retains more biological information than existing techniques in synthetic and real genomic data.

Modelling GxE with historical weather information improves genomic prediction in new environments

The paper shows that using genomic data, machine learning, and historical weather data can predict crop yield in new environments, highlighting non-linear GxE effects and soil-rain interactions.

Genomes of a major nosocomial pathogen Enterococcus faecium are shaped by adaptive evolution of the chromosome and plasmidome

This study analyzes the genomes of Enterococcus faecium strains to understand their adaptive evolution, identifying plasmids as key indicators of host specificity through population-wide genomic analysis.

Resolving outbreak dynamics using Approximate Bayesian Computation for stochastic birth-death models

This paper uses Approximate Bayesian Computation to improve parameter estimation in stochastic birth-death models for disease outbreaks, showing better identifiability of key parameters like R and infectious population size.

MediSyn: uncertainty-aware visualization of multiple biomedical datasets to support drug treatment selection

MediSyn synthesizes biomedical datasets to support drug treatment selection by visualizing drug-target relations and uncertainties.