Michael Allwright

Michael has 15 years experience in machine learning and comes from a background in mathematics. He has worked in the social sector starting “The Minerva Collective” a data sharing charity, and “Homeless In Focus”. In the research group he focuses on applying multi modal machine learning across the board to both internal and publicly available datasets relating to Parkinson’s Disease (PD) and Alzheimers Disease (AD).

Forefront Group: FOREFRONT BIOINFORMATICS & STATISTIC RESEARCH GROUP

Supervisors:

Dr Boris Gueennewig and Prof. Simon Lewis

Expertise:

  • Machine Learning
  • Strategy
  • Big Data

Affiliate Organisations:

USYD

Neurodegeneration of interest:

PT, AD, Depression

Specific Skills:

  • Machine Learning – python
  • Classification models
  • Neural networks
  • Clinical data
  • Clustering

Project - Multi Modal Machine Learning Applied to Parkinson’s and Alzheimers combining genetic, imaging, lifestyle and clinical data to determine new biomarkers and risk groups

Disease area:

PD, AD

Research Project Description

Intro and Overview: PD and ND, disease impact

  • PD and ND, disease symptomology and biology, sub-types and mechanisms
  • PD and ND clinical trial landscape
  • PD and ND biomarkers – need and status
  • Data types used in PD/ND biomarkers – multimodal
  • ML in general – advantages, requirements
  • ML in ND, precision medicine, opportunities
  • Current barriers to ML in ND
  • Multimodal Data in ND PPMI – taking it further (possibly split into other chapters)
  • Meta data analysis and development of clinical sub-types
  • Genetic data analysis
  • MRI data analysis
  • Dimensionality Reduction – WGCNA, UMAP, tSNE etc.
  • Multi-modal data fusion techniques (tbd)
  • Putting it all together
  • WES Data for PD Subtypes – comparing 15 clinical tests with WES to determine PD progression
  • UK Biobank ML on 500,000 participants contracting AD/PD – importance of APOE4 and other SNPs – predictive modelling and SHAP