1. Alzheimer's treatment is stagnating with minimal progress → key focus on amyloid protein but no effective drugs have been made (still effects 47 million people worldwide) + big issue is cost of final drug beyond just success in clincical trials

   • Billions of dollars put in already, 47 million people impacted, anti-amyloid treatments have minimal success

   Approach: DNA variants that heightened Alzheimers were found to be connected to the innate immune system that triggers inflammation in response to pathogens → patients taking drugs that block molecular triggers for inflammation have 50-70% lower chance of getting Alzheimers

   • analyze new targeted immune genes and variants to help inform Alzheimer's neurogenetics therapies (more info) (dataset)

2. Feature selection for ALS → using preexisting genomic + transcriptomic data to narrow features for reducing underfitting issues (dataset)

   1. 5k cases per year, 1 in 50k ppl, $16k-$200k per person → $80 million - $1 billion saved in hospital costs, no known treatments
   2. Approach: using gene expression matrices → run autoencoder/ANN-based framework to determine statistically significant genes → use for DTE/DGE analysis to find specific genes to target within brain
      1. Can also use principle component analysis (PCA) or non-linear techniques like t-distributed stochastic neighbor embedding (tSNE) and Uniform Manifold Approximation and Projection (UMAP)
   3. Similarly can be done for brain cancer (dataset) - 30k cases, $450k-$1 million cost per person

3. Looking into MRI scans and finding specific tumors in glioblastoma - lots of research has been done for finding the presence of MGMT promoter methylation, so can look elsewhere

   1. Incidence rate of 3.19 per 100,000 persons in the United States and a median age of 64 years (code - individuals have attempted this, but not fully solved the problem)
   2. Approach: Deep learning model for classification —> could try some niche models and use from other papers who've worked on Alzheimer's classification to tailor towards our use case
   3. ^ Brain scans and screenings not for the actual brain, but the cells in the brain (ie. white blood cells and other innate immune cells to find new peripheral markers for alzheimers)

1. Looking at regional brain samples of biomarkers (i.e. iron) + finding correlation with genes to predict onset of Parkinsons (paper)
   1. dataset looks at different neurological data points from microarrays and other biomarkers (https://human.brain-map.org/)
   2. parkinsons costs 1.215B in direct and indirect for canada → leading cause of disability in the world + no definitive tesrting mechanism on biological level for parkinsons (holy grail is really the biomarkers that we can find) → more info on treatment and prevention

Focus: How can we leverage genomic/transcriptomic data on the brain to discover new targets for Alzheimer's disease + prediction with respect with its connection to the immune system?