Senior Design Projects

ECS193 A/B Winter & Spring 2019

Python Fuzzy-Logic Web App for Behavioral Data

Email **********
Name David K. Johnson, M.A., PhD.
Affiliation Neurology

Project's details

Project title Python Fuzzy-Logic Web App for Behavioral Data
Background Exponential growth in computing power has promoted radical advancement in biomedical research but has not been applied widely to behavioral research data. This project would develop a web app that applies a scikit-fuzzy PyPI Python code module ( already written by the faculty but in need of optimization ) to data mine and harmonize behavioral datasets to an archetypal data dictionary ( templated and exhaustive; also created and maintained by the faculty ). The newly developed app would create XML-based data keys that index and link dispirit data fields stored across datasets in multiple formats, according to DDI-3 standards. Ultimately, this app could be used by a variety of academic, government and clinical research institutions.
Description UC DAVIS USE CASE: The UC Davis Medical Center is home of an NIH Center of Excellence – the Alzheimer’s Disease Research Center in the Department of Neurology. We use paper and pencil neuropsychological tests to track dementia related cognitive changes in thousands of older Northern Californians. Keeping track of these data in standard analytic/academic software like SAS, SPSS and R is challenging. Not only are these data kept in a confederacy of formats, spread across platforms and servers, but also the data dictionaries that describe these data are most often incomplete ( e.g., SAS and SPSS have very poor interfaces for data documentation ). The incomplete documentation of the metadata jeopardizes the usability of these ( very ) valuable behavioral datasets. Archived databases are at greatest risk, when the instrumental value of the dataset may be lost or at best inferred/recreated by second generation users. This problem is manifold because there are hundreds of centers conducting behavioral research stored in very large databases, minimally-maintained and in precarious disarray.
Deliverable GOAL: To create an online utility where archived data dictionaries would be submitted to a fuzzy logic comparison process and results returned that are usable/interpretable by assessment experts who could then efficiently winnow through large behavioral databases to identify conceptually similar outcomes and define similarity/difference between clinical databases. The key to the usability of the proposed app is its ( 1 ) query inclusion of metadata from a very large and robustly documented meta-database of common behavioral measures/outcomes ( akin to Rosetta stone for behavioral measurement ) and ( 2 ) the return of results in DDI-3 compliant XML templates. This tool would create huge value added to existing behavioral datasets that could reanimate their clinical relevance by enhancing usability.
Skill set desirable Python
Phone number **********
Client time availability 30-60 min every two weeks
IP requirement Open source project
Attachment N/A
Selected No
Team members N/A
TA N/A