nw  

GUI for identifying predictive variables of Alzheimer's disease in a large open dataset

Summary

Student research GUI on Alzheimer's with new ADNI genetic models.

Keywords

Python, Dash, SQLite, Machine learning, Deep learning, Alzheimer's Disease, ADNI

Objectives

This project is a continuation of a P5 and various data challenges conducted by Data Science students. Building on these solutions, the GUI (FHNW ADNI Explorer) provides an accessible way to explore the results obtained so far. It also reveals previously unexplored aspects of the ADNI data and lays the foundation for future integration with other brain-imaging and memory datasets, such as AIBL. The goal is to develop a visual analytics system that facilitates multi-modal machine learning. This system will help evaluate the contribution of each variable to Alzheimer's disease diagnosis, as well as tracking transitions from MCI (mild cognitive impairment) to AD or to healthy status.

Initial position

As life expectancy increases, the incidence of dementia-related conditions like mild cognitive impairment (MCI) and Alzheimer's Disease (AD) is rising. Although there is no cure for Alzheimer's, treatments can manage symptoms, improve quality of life, and potentially slow disease progression. For these treatments to be most effective, early diagnosis is essential. Open-source data initiatives such as the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing (AIBIL) provide extensive datasets that include cognitive tests, medical labs, brain imaging data, and newly added genetic variables. Analyzing these datasets could lead to significant advancements in understanding brain health, aging, and the impact of genetic factors.

Results

The GUI effectively integrates diverse datasets, including cognitive tests, medical labs, brain imaging data, and recently incorporated genetic variables from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and other sources. A key new addition is the inclusion of two machine learning model. The first model is working with the genetic variable telomere length and MRI imaging. The second model works the genetic variables pTau, polygenic hazard score (PHS) and cerebrospinal fluid (CSF). Both are analyzed separately by gender, enhancing the accuracy of the Alzheimer's disease and mild cognitive impairment prediction. The system allows users to explore the data interactively, uncovering significant patterns and correlations that influence the progression from mild cognitive impairment (MCI) to Alzheimer's Disease (AD). This updated tool provides crucial insights for advancing research in Alzheimer's prediction and risk assessment. The dashboard solution was validated through usability testing rounds with students that have worked on their own ADNI projects and are featured in the dashboard and some that have no knowledge of the ADNI data. Due to the sensitive ADNI participant data the GUI is not open to the public. Authorized users can find the GUI under https://iit.cs.technik.fhnw.ch/adni/.

GUI homepage

Homepage

Project data

Bachelor thesis
Project duration: 19.02.2024 – 16.08.2024
Effort in hours per student: 360h
Team size: 2

Customer
nw
 

Institute for Interactive Technologies, University of Applied Sciences and Arts Northwestern Switzerland
Bahnhofstrasse 5
CH-5210 Windisch

Project team

David Dedic (Data Science), Ifrah Gobdon (Computer Science)

Contact

Prof. Dr. Arzu Çöltekin: arzu.coltekin@fhnw.ch
Dr. Leticia Fernandez Moguel: leticia.fernandezmoguel@fhnw.ch

<< zurück