DxSter Alzheimer's disease Algorithmic Diagnostic Tool

What is DxSter?

DxSter, the Alzheimer's disease Algorithmic Diagnostic Helper, is a codified algorithm that can diagnose normal cognition, MCI and Dementia which services as a valid alternative that reduces time, effort and biases associated with consensus diagnosis. Codifying the AlgDx algorithmic diagnosis project is an effort to translate the algorithm that calculates the physician and clinical diagnosis. The algorithm is designed to reduce the number of cases that need review in consensus conference. To accomplish this Dr. Meredith Wicklund from the University of Florida has been awarded a grant through the Ed and Ethel Moore Alzheimer’s Research Program in which CTS-IT has been provided funding to translate the AlgDx into Python and JavaScript software libraries. Click the button below to see how it works!

Check it out!

Publications

Duara, R., Loewenstein, D. A., Greig, M., Acevedo, A., Potter, E., Appel, J., … Potter, H. (2010). Reliability and Validity of an Algorithm for the Diagnosis of Normal Cognition, MCI and Dementia: Implications for Multi-Center Research Studies. The American Journal of Geriatric Psychiatry : Official Journal of the American Association for Geriatric Psychiatry, 18(4), 363–370.http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2844658/

Get the Code

Can DxSter help you? Click the button below to get access to the source code.

DxSter Code

FAQs

How does DxSter work?

When you go to the DxSter tool (found here https://ctsit.github.io/dxster/dxcalc.html), you will be asked to enter your Clinical dx as well as your Neuro Psychological dx. Based on these two variables, the DxCalc will generate an algorithmic diagnosis for the patient.

Where does the ALGDx Algorithm come from?

In 2009 Dr. Ranjan Duara et. al. from Mt. Sinai Medical Center in Miami Florida conducted a study to create an algorithm that could use a combination of the clinical diagnosis and the neuropsychological diagnosis to computationally arrive at a diagnosis with a high degree of accuracy. The traditional consensus diagnosis (ConsDx) of normal cognition, mild cognitive impairment (MCI), and dementia relies on the reconciliation of an informant-based report of cognitive and functional impairment by a physician (PhyDx), and a neuropsychological diagnosis (NPDx). As this procedure may be labor-intensive and influenced by the philosophy and biases of a clinician, the diagnostic algorithm (AlgDx) was developed to identify individuals as cognitively normal, with MCI or dementia.

What variables are used for calculating the diagnosis?

The AlgDx combines the PhyDx with the NPDx, using a computational algorithm that provides cognitive diagnoses, as defined by the National Alzheimer Coordinating Center/Uniform Data Set (NACC/UDS) nomenclature. Reliability of the AlgDx was assessed in 532 community- dwelling elderly subjects by its concordance with the ConsDx, and association with two biomarkers, medial temporal atrophy (MTA) scores of brain MRI scans, and ApoE-ε4 genotype.

What tools were used to create DxSter?

The Algorithmic diagnosis uses a combination of variables to compute the AlgDx. By collaborating with the authors of the algorithm we developed a rules engine that emulates the human interpretation of the physician and neuropsychologist diagnosis. Once the rules engine was developed we utilized the Python programing language to codify the rules engine and DxSter was born.

Can I take DxSter for a test drive?

Yes! The DxSter Alzheimer’s disease Algorithmic Diagnostic Helper is setup as a proof of concept web application and can be found at https://ctsit.github.io/dxster/dxcalc.html.

How do I get DxSter?

The code base is open source can be found at https://github.com/ctsit/dxster.

How do I use DxSter?

To use DxSter you will need to download the libraries and integrate them into your own tools. You will need to include Algorithmic diagnosis in your own data sets.

Examples

HTML & Javascript Implementation: DxSter Calculator

Go to DxCalc and view page source to see an example implementation of the dxster.js JavaScript AlgDx Module. This module is a port of the Python library. It implements the identical AlgDx truth table and AlgDx calc functions. dxcalc.html serves as a reference implementation of this functionality.

Bash & Python with CSV test data set script example

In the examples/bash folder in the GitHub repository are example files of a sample csv dataset called sample_data.csv that contains a subject ID, NPDx, and CDR score. The file calc_algdx_sample.sh will call Python and the dxster python module to loop through every row in your sample data and output the AlgDx corresponding diagnosis to stdout.

Test out your own data in the DxSter Uploader

If you like our DxSter Calc concept and want to go further with it, give it a test drive with your own data! Go to the DxSter Uploader and upload your own data set into the file uploader. Your results will populate with an option to print at the end.

How can you help?

DxSter has been a labor of love and we want to see it used in real world applications. We ask that you use the online calculator or download the code to use it to diagnose the cognitive state of study participants. After which please give us feedback on how to make the program better.

Cite Us

Please reference Clinical and Translational Science - Informatics and Technology group (CTS-IT) in any research report, journal, or publication that requires citation of authors' work. Recognition of CTS-IT resources you used to perform research is important for acquiring funding for the next generation of informatics services and our research and development activities in software development and information science.

At minimum, a citation should include: Clinical and Translational Science Informatics and Technology group at the University of Florida

Our suggested acknowledgement is (select one or more items within the braces, as appropriate): The authors acknowledge the Clinical and Translational Science Informatics and Technology group at the University of Florida for providing {Python software library, code, examples, calculator} resources that have contributed to the research results reported within this paper. URL: http://www.ncbi.nlm.nih.gov/pubmed/20306566

Open Source

Hack the code, improve the code.

Check out our Github page here.

Donate

This work has been supported by the Florida Department of Health's Ed and Ethel Moore Alzheimer's Disease Research Program grant number 66315-UF, the National Institute of Aging, and the 1Florida Alzheimer's Research Center grant number P50AG047266. To donate to the 1Florida ADRC go to http://1floridaadrc.org/donate.html.