One of humanity’s biggest problems is reducing human suffering. And we know there’s a lot of forms of suffering that impact our health, lifestyle, and general mental state. But when we look at the worst forms of suffering, its usually things that we know are destroying us but we can’t stop.

44 million individuals are afflicted with Alzheimer's, the most deadly neural disorder worldwide.

Scientists believe that a buildup of corrupted amyloid, an essential protein in neural growth and repair, is the leading risk factor for Alzheimer's; however, the extensive array of treatments and billions in R&D into amyloid has had the highest failure rate of any drug class in the space.

Revenue is projected to drop dramatically from 3 billion in 2022 to 500 million, and while we've found mild success in developing therapies, these are often expensive to access/manufacture.

Not much testing has been performed outside this space, with only 10% of grants going to genomics due to the complex neural system and lack of genomic understanding from researchers. Hundreds of gene and RNA targets are not explored, each of which could make an instrumental impact in the drug discovery process. This includes immune system parts, like the tumor necrosis factor, which has been shown to reduce inflammation and Alzheimer's onset.

Alzheimer's have no definitive testing mechanism that evaluates the quantitative biological markers responsible for the disease. Reliable biomarkers, which need to be abnormal, visible, and testable with high confidence, can be used by biotech to produce meaningful therapies and interventions. The only biomarkers which match these constraints fall at the genome level, whereas the logistical requirements need to be satisfied by brain imaging.

It's clear that learning more about the brain's processes at the neurobiological level will be critical for discovering new targets and therapies aimed at reducing the strain of neurodegenerative diseases like Alzheimer's. By using genomic data and MRI imaging, we aim to discover possible gene targets through a six-step process for future drug discovery methods in Alzheimer's and other neurological diseases, based on inflammation patterns detected in brain scans.

First, we detect different regions of the brain using a convolutional neural network image segmentation model through pre-passed MRI imaging data inputs. Then, we utilize an artificial neural network to map inflamed regions along with levels of inflammation.

Next, scatterplots of gene expression for various brain regions are superimposed to see the genes that are over-or underexpressed that further cause inflammation.

Afterward, tSNE visualized the high-dimensional gene expression data into noticeable clusters to select specific genes for further analysis, based on whether they caused inflammation.

Then, differential gene expression and transcript expression can find statistically significant features for targeting specific genes for neurological diseases, using DeSeq2 and Sleuth.

Through comparative analysis of inflammatory expression against the neurodegenerative profile, we found four genes with similar expressions (GRB2, MAPK1, PRKCG, and IL1B) in all brain areas.

Furthermore, in the hippocampal formation plots, we noticed a strong correlation with clustered genes and the ground-truth, indicating early onset of gene expression in this region directly correlating to AD. This was done for all brain regions so that insight can be found for its function.

Through feature selection done by tSNE, we discovered that this gene expression is dissimilar through different brain regions, showing the uniqueness of the hippocampal area.

Currently, industry, government, and universities are the biggest investors in research. Out of these, the government is most receptive to funding new research, with 27% of their grant money going to first-time grant awardees. After an initial build phase funded by the government, to offset maintenance costs, researchers will pay a licensing cost for the full service.

With our workflow and results, clinical researchers can run testing with modifications of the found genes through gene editing to see how it affects brain function. This open-source workflow can be interpretable in terms of other neurodegenerative diseases beyond Alzheimer’s, like Parkinson's. Key research and discovery bottlenecks need to be met with the ability to rapidly prototype new gene profiles and build models that accurately reflect neurodegenerative disease progression at different resolutions. Our study showcases the benefit of being able to realize this process in the case of connecting immune responses in the brain to the progression of Alzheimer's, which can play a key role in assessing, diagnosing, and even predicting new targets.

Our team is excited to continue this project further through the integration of MRI scans and applications into finding other targets for neurodegenerative diseases.