CoChat

Toward a Smartphone-Based and Conversational Agent–Delivered Just-in-Time Adaptive Holistic Lifestyle Intervention for Older Adults Affected by Cognitive Decline: Two-Week Proof-of-Concept Study

Open access

The DOI matches its Crossref record; 72 of the 72 references it cites were found in a registry or an index; 13 of the 27 sources its claims cite were checked for support.

Abstract

Background: Dementia is projected to impact 152 million people by 2050, making it one of the most pressing global health challenges. The neurodegenerative process initiates well before clinical symptoms manifest, advancing from subjective cognitive decline (SCD) to mild cognitive impairment (MCI) and ultimately to dementia. Despite the growing prevalence, awareness of dementia prevention is limited, and many individuals express a desire to cease living upon diagnosis. Lifestyle interventions can mitigate cognitive decline, but there is a need for effective, scalable approaches to deliver these interventions to older adults. Digital health interventions, such as app-based just-in-time adaptive interventions, offer a promising solution, but their application in cognitively impaired older populations remains underexplored. Objective: This formative study evaluated the plausibility, acceptability, and adherence to a smartphone-based just-in-time adaptive digital lifestyle intervention delivered by a rule-based conversational agent (CA) among older adults with SCD or MCI. The primary focus was on adherence to the CA-initiated conversational turns (measured objectively via interaction logs), and secondary objectives included perceptions of technology acceptance, working alliance with the CA, self-reported adherence to the suggested health-promoting activity, and feedback for future improvements (through a questionnaire and short interview). Methods: This monocentric study investigated 15 participants (mean age 70.3, SD 5.01; 10 female and 5 male participants) with SCD (n=12) or MCI (n=3). Participants used the study app that delivered daily health-promoting activities through a CA over 2 weeks. Participants received notifications to engage in 7 health-related activities, and adherence to the activities was self-reported. Post intervention, participants rated their experience with the app and assessed their working alliance with the CA through the 6-item session alliance inventory. Data on smartphone use, demographic information, and cognitive performance (via Montreal Cognitive Assessment) were collected during a preintervention visit. Results: Participants rated the study app positively, especially regarding ease of use and a subset of the working alliance. Adherence to the CA-initiated conversational turn was measured at an average of 81% across 14 days. In total, 27% (mean 4.07, SD 2.27) of participants indicated being vulnerable, and 100% then responded with their state of receptivity, of which 83% (mean 3.14, SD 1.61) were receptive to completing the activity, and 69% (mean 2.86, SD 1.70) self-reported adherence to the activity. There was no significant decline in adherence across the study period. Qualitative results support these findings and present two emerging themes: app enjoyment and enhancing engagement. Conclusions: This study demonstrates that smartphone-based just-in-time adaptive interventions are feasible and generally well-accepted by older adults with SCD or MCI. However, the findings underscore the need for robust technological infrastructure and potential personal assistance to optimize adherence. Future interventions could benefit from integrating wearables to improve real-time engagement and accurately monitor adherence, ultimately supporting healthy aging and cognitive health in older populations.

Overview

CoChat Summary

## TL;DR

This 2-week single-arm proof-of-concept study found that a smartphone-based, rule-based conversational agent could deliver a holistic just-in-time adaptive intervention to older adults with subjective cognitive decline (SCD) or mild cognitive impairment (MCI) with generally good engagement and acceptability. The authors frame it as feasible and promising, but early-stage and limited by self-report data, a small sample, and technical/hardware issues.

## Key findings

  • Sample and feasibility: 17 participants were recruited; 2 were excluded because of technical/hardware issues, leaving 15 analyzable participants. The sample was 67% female (n=10), mean age 70.3 years (SD 5.01), and mean MoCA score 27.73 (SD 2.01); 12 had SCD and 3 had MCI.
  • Engagement with the conversational agent: Participants responded to the afternoon vulnerability prompt on average 81% of days over 14 days; 19% did not respond on average.
  • Behavioral pathway within the JITAI: On average, 27% (mean 4.07, SD 2.27) of participants reported vulnerability; among those vulnerable, 100% responded to the receptivity prompt; 83% (mean 3.14, SD 1.61) were receptive; and 69% (mean 2.86, SD 1.70) self-reported adherence to the suggested activity.
  • Acceptability: All mean technology acceptance and alliance ratings were above the neutral midpoint of 4 on the 7-point scales; Wilcoxon signed-rank tests found that perceived ease of use and two session alliance items were significantly higher than 4. Qualitative feedback produced two themes: app enjoyment and enhancing engagement.

## Methods

  • Design: 2-week, single-arm proof-of-concept study of a smartphone-based, rule-based conversational agent delivering a just-in-time adaptive intervention (JITAI) across 7 health domains.
  • Sample/data: Older adults aged 60-85 with SCD or MCI, recruited from the Bern Memory Clinic and local cognitive training groups in July 2024; final analyzable sample n=15. Data came from app interaction logs, a postintervention questionnaire, and short semistructured telephone interviews.
  • Analysis: Descriptive statistics for engagement and survey outcomes; Shapiro-Wilk tests and Wilcoxon signed-rank tests against neutral midpoint 4 for Likert-scale items; inductive thematic analysis of interview responses using MAXQDA.

## Limitations

  • The paper states this was a formative, discovery-oriented proof-of-concept study with no a priori benchmarks for adherence or acceptability, which limits interpretability and “go/no-go” decisions.
  • The study relied on self-reported data only for vulnerability, receptivity, and adherence; the authors note this may not scale to people with more advanced cognitive decline.
  • The intervention used a simple, low-burden rule-based CA with predefined response options and limited adaptability, which constrained personalization.
  • Evident from the text: the sample was small (n=15 analyzable), predominantly female, and drawn from a local clinical/cognitive-training setting, which limits generalizability. Technical/hardware problems also excluded 2 participants and affected notifications.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.2196/66885

Cites

72 of 72 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1. A decision tree indicating the JITAI logic for this 2-week single-arm proof-of-concept study among older adults with SCD or MCI, delivered by a rule-based CA through a smartphone app. CA: conversational agent; JITAI: just-in-time adaptive intervention; MCI: mild cognitive impairment; SCD: subjective cognitive decline.
Figure 1. A decision tree indicating the JITAI logic for this 2-week single-arm proof-of-concept study among older adults with SCD or MCI, delivered by a rule-based CA through a smartphone app. CA: conversational agent; JITAI: just-in-time adaptive intervention; MCI: mild cognitive impairment; SCD: subjective cognitive decline.CC BY · PubMed Central
Figure 2. The three conversational turns with the CA Elsa providing the JITAI in the 2-week single-arm proof-of-concept study among older adults with SCD or MCI. Translated from German. From left to right; first chat panel: determining the state of vulnerability (ie, “Have you completed an activity today?”); second chat panel: determining the state of receptivity (ie, if no: “Would you be able to complete the activity later?”); and third chat panel: determining adherence to the CA-suggested health-promoting activity (ie, “Did you complete this activity today?”). CA: conversational agent; JITAI: just-in-time adaptive intervention; MCI: mild cognitive impairment; SCD: subjective cognitive decline.
Figure 2. The three conversational turns with the CA Elsa providing the JITAI in the 2-week single-arm proof-of-concept study among older adults with SCD or MCI. Translated from German. From left to right; first chat panel: determining the state of vulnerability (ie, “Have you completed an activity today?”); second chat panel: determining the state of receptivity (ie, if no: “Would you be able to complete the activity later?”); and third chat panel: determining adherence to the CA-suggested health-promoting activity (ie, “Did you complete this activity today?”). CA: conversational agent; JITAI: just-in-time adaptive intervention; MCI: mild cognitive impairment; SCD: subjective cognitive decline.CC BY · PubMed Central
Figure 3. Daily response rates for afternoon vulnerability notifications to determine whether the participant has already completed the suggested activity or not, or if the participant does not respond. Graphed over the 2-week single-arm proof-of-concept study among older adults with subjective cognitive decline or mild cognitive impairment.
Figure 3. Daily response rates for afternoon vulnerability notifications to determine whether the participant has already completed the suggested activity or not, or if the participant does not respond. Graphed over the 2-week single-arm proof-of-concept study among older adults with subjective cognitive decline or mild cognitive impairment.CC BY · PubMed Central
Figure 4. Qualitative results of the 2-week proof-of-concept study conducted with a smartphone-based CA among older adults with SCD or MCI. This presents a thematic map of the main emerging themes, app enjoyment and enhancing engagement, from the qualitative interviews. Note: Numbers in circles represent the frequency count, or how many times the participant theme and its concept were mentioned by participants. Some participants mentioned multiple examples of a single theme during their interview, which then led to multiple frequency counts being identified during thematic analysis. CA: conversational agent; MCI: mild cognitive impairment; SCD: subjective cognitive decline.
Figure 4. Qualitative results of the 2-week proof-of-concept study conducted with a smartphone-based CA among older adults with SCD or MCI. This presents a thematic map of the main emerging themes, app enjoyment and enhancing engagement, from the qualitative interviews. Note: Numbers in circles represent the frequency count, or how many times the participant theme and its concept were mentioned by participants. Some participants mentioned multiple examples of a single theme during their interview, which then led to multiple frequency counts being identified during thematic analysis. CA: conversational agent; MCI: mild cognitive impairment; SCD: subjective cognitive decline.CC BY · PubMed Central

Verification

Claims the paper makes

8 Valid · 0 Not valid · 8 Not checked · 54 Own results and interpretations

30 more claims

Create a free account to see all 70 claims and to verify the 2 references still open.

See all claims — free account
  • It may state two separate claims as one. 21
  • Probably not a substantial claim. 19
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Self-check results appear here once an independent check has confirmed them.

Cited works that were retracted

Works this paper cites that were later retracted, from the retraction notices publishers deposit. Major when the notice came out before this paper was published.

1 of 1

Citations

Cites

1 retracted

Research works printed but not linked: 1

  1. Alzheimer’s Association, Alzheimers Dement, 17

Cited by

See all 5

Metadata

Authors
Brill, E., Vinay, R., Nißen, M., Joshi, P., Klöppel, S., & Kowatsch, T.
Affiliations
  • University of Bern (CH)
  • University of Zurich (CH)
  • Georgia State University (US)
Corresponding author
Vinay, Rasita
Venue
JMIR Formative Research
Published
2025-07-28
Volume
9
Pages
e66885-e66885
ISSN
2561-326X (ISSN-L)
OpenAlex
W4410287287
Cited by
5
References
72
Citations by year
2026: 5
FWCI
3.99
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR Formative Research (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Alexandria (UniSG) (University of St.Gallen) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Zurich Open Repository and Archive (University of Zurich) (submittedVersion, cc-by)
Topics
  • Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
  • Health Sciences › Health Professions › General Health Professions › Mobile Health and mHealth Applications
  • Health Sciences › Medicine › Psychiatry and Mental health › Dementia and Cognitive Impairment Research
Keywords
subjective cognitive decline, mild cognitive impairment, dementia prevention, just-in-time adaptive interventions, chatbots, rule-based chatbots, smartphone-based intervention, digital lifestyle intervention, older adults, intervention adherence, therapeutic relationship, technology acceptance, health-promoting activities, acceptability, digital health apps, Montreal Cognitive Assessment
MeSH terms
Smartphone, Proof of Concept Study, Aged, Aged, 80 and over, Female, Humans, Life Style, Male, Cognitive Dysfunction, Mobile Applications

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.