# SPEEKO for Elderspeak: A Self-Monitoring App to Improve Communication and Reduce Behavioral Symptoms in Care for Persons with Alzheimer's Disease and Other Dementias

> **NIH NIH R21** · UNIVERSITY OF KANSAS MEDICAL CENTER · 2020 · $229,500

## Abstract

PROJECT ABSTRACT
 The growing population of persons with dementia presents challenges to providing high quality nursing
home (NH) care. Care is complicated by behavioral and psychosocial symptoms of dementia (BPSD) such as
aggression, vocal outbursts, wandering, and withdrawal that occur as persons with dementia lose cognitive
and communication abilities and cannot verbally express unmet physical and psychosocial needs. NH staff
use “elderspeak,” speech similar to baby talk that is demeaning to residents. Residents react to elderspeak
with BPSD resulting in increased use of psychotropic medication to control BPSD, higher staffing needs, more
staff stress and turnover, and increased costs. The Changing Talk (CHAT) training program educates staff
about elderspeak and better communication strategies and is proven to reduce staff elderspeak and BPSD in
NH residents. To increase dissemination, this effective classroom-based training has been transformed to
online internet modules that improve access for busy staff including those in rural and small NHs and increase
ease of dissemination across long term care service settings. To further enhance training effects, SPEEKO for
Elderspeak, an automated performance-based app that uses natural language processing has been developed
to provide immediate reinforcement of training to staff at the point-of care. This study will build on app proof of
concept to establish feasibility of use in the clinical setting and will test preliminary efficacy of the app for
augmenting the effect of training on reducing elderspeak communication in dementia care.
 This R21 will conduct a clinical trial in response to PAR-18-179; Research on Informal and Formal
Caregiving for Alzheimer’s Disease by testing a technology-based, natural language processing app to support
caregivers in addressing unique challenges of providing advanced dementia care. We will demonstrate
feasibility of use at the point-of-care with five certified nursing assistants, and further validate the app’s
accuracy by comparing app outputs reporting diminutive use to results from transcription and psycholinguistic
coding. Hypothesized app effects on augmenting reductions in staff elderspeak diminutive use after training
will be tested in a cluster randomized trial comparing immediate and delayed feedback with 60 CNAs in 6
nursing homes. If effective in boosting training effects, this readily scalable app can be disseminated across
long term care service and support settings to increase communication training effects that will reduce resident
BPSD and psychotropic medication use and provide periodic reinforcement for communication training.
 The goal is to further improve and sustain effects of communication training in nursing staff that will
reduce BPSD and inappropriate use of psychoactive medications. This nonpharmacological approach uses
available technology to improve dementia care and aligns with the top priority goal of the National Partnership
to Impr...

## Key facts

- **NIH application ID:** 9993195
- **Project number:** 5R21AG065029-02
- **Recipient organization:** UNIVERSITY OF KANSAS MEDICAL CENTER
- **Principal Investigator:** Kristine N. Williams
- **Activity code:** R21 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2020
- **Award amount:** $229,500
- **Award type:** 5
- **Project period:** 2019-08-15 → 2023-04-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/9993195

## Citation

> US National Institutes of Health, RePORTER application 9993195, SPEEKO for Elderspeak: A Self-Monitoring App to Improve Communication and Reduce Behavioral Symptoms in Care for Persons with Alzheimer's Disease and Other Dementias (5R21AG065029-02). Retrieved via AI Analytics 2026-05-25 from https://api.ai-analytics.org/grant/nih/9993195. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*