Web-app tool to support technology use decision making for individuals with AD/ADRD and their caregivers

Awardee Organization(s): University of Washington
Principal Investigator(s): Clara Berridge, PhD, MSW
Official Project Title: Talking Tech With Dementia Care Dyads: Improving a Self-Administered Tool to Support Informed Decision
AITC Partner: PennAITech
Website(s): https://socialwork.uw.edu/faculty/professors/clara-berridge

AITC Consortium; Basic Information; PennAITech Awardees
Abstract: The proposed project is to enhance the Let’s Talk Tech (LTT) intervention that is delivered as a web application. LTT is the first of its kind self-administered tool to help families meaningfully engage people living with mild dementia in digital technology use planning to enable optimal use to support dementia caregiving at home. Let’s Talk Tech is an education and communication intervention that supports decision making and planning for technology use. It includes the following components: education accessible to people living with mild AD/ADRD and care partners about multiple technologies, facilitation of dyadic communication, and documentation of the person living with dementia’s preferences. LTT has demonstrated in a pilot promising preliminary feasibility and efficacy on targeted measures for informed shared decision making about technologies. This project will implement what was learned from the pilot study about ways to further expand its reach, relevance, and sharing with the entire care network. Aim 1 is to enhance Let’s Talk Tech to achieve wider relevance and equitable access with 4 new features. Aim 2 is to implement EHR integration and patient-controlled sharing of LTT’s preference summaries, and Aim 3 is to assess clinician acceptability of viewing dyad’s LTT preferences in a test instance the EHR. This will expand the intervention’s reach and functionality, employ standards to promote interoperable sharing of documented preferences, and further test and iteratively improve LTT.

View Resource

Xcessiv

Xcessiv is an open-source, web-based application developed using Python and Javascript for automating and visualizing the model selection process, hyperparameter tuning, and feature extraction in machine learning. It provides a user-friendly interface for managing and executing experiments across multiple algorithms and datasets. Xcessiv employs models from the Scikit-learn package, supports parallel hyperparameter searches using Bayesian optimization, and enables easy management and comparison of hundreds of different model-hyperparameter combinations, easy stack ensemble creation, and automated ensemble construction. It can also export created stacked ensembles as a standalone Python file to support multiple levels of stacking.

Link: https://xcessiv.readthedocs.io/en/stable/

View Resource