Marco Chierici

This study evaluates generative models for transcriptomics data using statistical distances to assess their performance and convergence.

This paper introduces singleDeep, a deep learning pipeline for scRNA-Seq data analysis that improves phenotype prediction and provides insights into gene importance and cell types.

This paper introduces Signature Distance, a statistical distance that improves upon energy distance for comparing empirical distributions in high-dimensional biological data, showing greater sensitivity to density changes and enabling model-free data expansion.

The paper identifies potentially harmful genetic variants in euploid miscarriage embryos and highlights specific genes linked to miscarriage risk.

The paper identifies widespread transcription initiation at microsatellites across species and develops deep learning models to predict STR-associated transcription and its impact on disease variants.

The paper analyzes transcriptomics data availability and reusability between microarray and RNA-seq experiments published between 2009 and 2013, and provides recommendations to improve RNA-seq data sharing.

The paper introduces INF, a multi-omics framework that improves cancer classification by integrating gene expression, protein, and copy number data, achieving better accuracy and smaller feature sets than single-omics models.

The paper introduces a framework combining deep learning and radiomics for cancer prognosis using CT and PET imaging, showing improved accuracy over single-method approaches.

This study evaluates the reproducibility of AI algorithms in digital pathology using the DAPPER framework, analyzing accuracy and feature stability across multiple deep learning models and datasets.