CoChat
Sign Up
Cited by 165Mobile Health and mHealth ApplicationsPhysical Activity and HealthInnovation in Digital Healthcare Systems

A Physical Activity and Diet Program Delivered by Artificially Intelligent Virtual Health Coach: Proof-of-Concept Study

Open access
Maher, Carol Ann, Davis, Courtney Rose, Curtis, Rachel Grace, Short, Camille Elizabeth, Murphy, Karen Joy·2020·JMIR mHealth and uHealth
DOI 10.2196/17558PMC PMC7382010

Published version of a preprint

Open this paper in CoChat — free account

Abstract

Background: Poor diet and physical inactivity are leading modifiable causes of death and disease. Advances in artificial intelligence technology present tantalizing opportunities for creating virtual health coaches capable of providing personalized support at scale. Objective: This proof of concept study aimed to test the feasibility (recruitment and retention) and preliminary efficacy of physical activity and Mediterranean-style dietary intervention (MedLiPal) delivered via artificially intelligent virtual health coach. Methods: This 12-week single-arm pre-post study took place in Adelaide, Australia, from March to August 2019. Participants were inactive community-dwelling adults aged 45 to 75 years, recruited through news stories, social media posts, and flyers. The program included access to an artificially intelligent chatbot, Paola, who guided participants through a computer-based individualized introductory session, weekly check-ins, and goal setting, and was available 24/7 to answer questions. Participants used a Garmin Vivofit4 tracker to monitor daily steps, a website with educational materials and recipes, and a printed diet and activity log sheet. Primary outcomes included feasibility (based on recruitment and retention) and preliminary efficacy for changing physical activity and diet. Secondary outcomes were body composition (based on height, weight, and waist circumference) and blood pressure. Results: Over 4 weeks, 99 potential participants registered expressions of interest, with 81 of those screened meeting eligibility criteria. Participants completed a mean of 109.8 (95% CI 1.9-217.7) more minutes of physical activity at week 12 compared with baseline. Mediterranean diet scores increased from a mean of 3.8 out of 14 at baseline, to 9.6 at 12 weeks (mean improvement 5.7 points, 95% CI 4.2-7.3). After 12 weeks, participants lost an average 1.3 kg (95% CI –0.1 to –2.5 kg) and 2.1 cm from their waist circumference (95% CI –3.5 to –0.7 cm). There were no significant changes in blood pressure. Feasibility was excellent in terms of recruitment, retention (90% at 12 weeks), and safety (no adverse events). Conclusions: An artificially intelligent virtual assistant-led lifestyle-modification intervention was feasible and achieved measurable improvements in physical activity, diet, and body composition at 12 weeks. Future research examining artificially intelligent interventions at scale, and for other health purposes, is warranted.

AI snapshot

Written from the paper’s full text.

TL;DR

This 12-week single-arm proof-of-concept study tested an artificially intelligent virtual health coach, Paola, delivering a physical activity and Mediterranean-style diet program to inactive adults aged 45–75. The intervention was feasible and was associated with improvements in physical activity, diet adherence, weight, and waist circumference, with no significant blood pressure changes.

Key findings

  • Feasibility was strong: 99 potential participants expressed interest within 4 weeks, 31 were enrolled, and retention was 90% at 12 weeks (28/31 attended); no adverse events occurred.
  • Physical activity increased by 109.8 minutes/week from baseline to week 12 (95% CI 1.9–217.7, P=.005).
  • Mediterranean diet adherence rose from 3.8/14 at baseline to 9.6/14 at 12 weeks, a mean improvement of 5.7 points (95% CI 4.2–7.3, P<.001).
  • Participants lost an average 1.3 kg (95% CI –2.5 to –0.1, P=.01) and 2.1 cm in waist circumference (95% CI –3.5 to –0.7, P=.003); blood pressure did not change significantly. Engagement averaged 6.9 of 11 weekly check-ins (64%), below the a priori 70% target.

Methods

  • Design: 12-week single-arm pre-post proof-of-concept study conducted in Adelaide, Australia, from March to August 2019.
  • Participants: 31 inactive community-dwelling adults aged 45–75 recruited through news stories, social media, and flyers; eligibility required not meeting Australian physical activity guidelines and not following a Mediterranean dietary pattern.
  • Intervention and analysis: participants used Paola, a natural language processing virtual health coach built with IBM Watson and deployed on Slack, plus a Garmin Vívofit4 tracker, MedLiPal website, and diet/activity log. Feasibility was analyzed descriptively; efficacy used unadjusted repeated measures ANOVA on an intention-to-treat basis with last observation carried forward, plus complete-case sensitivity analyses.

Limitations

  • Stated in the paper: the pre-post design lacked a randomized control comparator, and follow-up was limited to three months.
  • Evident from the text: the sample was small (n=31) and mostly women (21/31), primary outcomes were self-reported via the Active Australia Survey and a Mediterranean diet questionnaire, engagement fell below the a priori target (64% vs 70%), and 3 participants had minimal smartphone skills while technical problems with Slack or the software sometimes interrupted check-ins.

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

Integrity signals

Registry

Crossref · 10.2196/17558

Cites

35 of 35 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 27 cited sources checked · 0 supported · 0 flagged

  • Findingnone given

    No adverse events related to study participation were experienced.

    “No adverse events related to study participation were experienced.”
    Quote in the paperNot a substantial claim
  • Findingown data

    Overall, from baseline to 12 weeks, physical activity increased by 109.8 minutes (95% CI 1.9 to 217.7, P=.005).

    “Thus overall, from baseline to 12 weeks, physical activity increased by 109.8 minutes (95% CI 1.9 to 217.7, P=.005).”
    • Also stated in: § Results · plain
      “Participants completed a mean of 109.8 (95% CI 1.9-217.7) more minutes of physical activity at week 12 compared with baseline.”
    Quote in the paperSubstantial claim
  • Findingown data

    Engagement with the virtual health coach was slightly lower than the a priori threshold of ≥70% (actual completion rate 64%).

    “Engagement with the virtual health coach (based on completed weekly check-ins) was slightly lower than our a priori threshold of ≥70% (actual completion rate 64%).”
    Quote in the paperSubstantial claim
  • Findingown data

    Mean change in Mediterranean diet adherence from baseline to 12 weeks was 5.7 points (95% CI 4.2 to 7.3, P<.001).

    “(mean change baseline to 12 weeks 5.7, 95% CI 4.2 to 7.3, P<.001).”
    • Also stated in: § Results · plain
      “Mediterranean diet scores increased from a mean of 3.8 out of 14 at baseline, to 9.6 at 12 weeks (mean improvement 5.7 points, 95% CI 4.2-7.3).”
    Quote in the paperSubstantial claim
  • Findingown data

    Participants completed an average of 6.9 out of 11 possible weekly check-ins (64%, range 1-11).

    “Out of a maximum of 11 possible check-ins, participants completed an average of 6.9 (64%, range 1-11).”
    Quote in the paperNot a substantial claim
  • May combine two claims. 33
  • Classified as probably not a substantial claim. 15

  • May combine two claims.

    Findingown data

    From baseline to week 6, weekly physical activity increased by approximately one hour, and then by a further 50 minutes at 12 weeks.

    “From baseline to week 6, weekly physical activity increased by approximately one hour, and then by a further 50 minutes at 12 weeks.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Around 70% of participants completed check-ins in weeks two, three, four and 12, with engagement sliding to around 50% through weeks eight and nine, before rising again in the final weeks.

    “Around 70% of participants completed the check-ins in weeks two, three, four and 12, with engagement gradually sliding to around 50% through weeks eight and nine, before rising again in the final weeks of the program.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingcitation

    Using an identical search strategy, the authors updated the review in January 2020 and identified a further 11 studies; most recent studies were aimed at mental health self-management (n=5).

    “Using an identical search strategy, we updated this review in January 2020 and identified a further 11 studies. Similar to the previous review, most recent studies were aimed at mental health self-management (n=5) [10-12].”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [10] source

      Not checked in the free run.

    • cites [11] source

      Not checked in the free run.

    • cites [12] source

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    Weekly total moderate and vigorous physical activity increased from baseline to week 12 by 109.8 minutes (95% CI 1.9 to 217.7), with overall model F(2,29)=6.45, P=.005.

    “| Weekly total MVPAb minutes (min) | 206.1 (166.5) | 266.8 (207.2) | 315.9 (261.7) | 6.45, .005 | 60.8 (–21.7 to 143.3) | 109.8 (1.9 to 217.7) |”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Waist circumference decreased from baseline to week 6 by 1.0 cm (95% CI -1.8 to -0.1) and to week 12 by 2.1 cm (95% CI -3.5 to -0.7), overall F(2,29)=7.13, P=.003.

    “| Waist circumference (cm) | 96.0 (16.0) | 95.1 (15.8) | 93.9 (15.8) | 7.13,.003 | –1.0 (–1.8 to –0.1) | –2.1 (–3.5 to –0.7) |”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    Of 31 participants enrolled, 3 had minimal smartphone skills and were reliant on partners or children to use their phones as part of the study.

    “Of the 31 participants meeting all eligibility criteria and enrolled in the study, it became clear during baseline appointments that 3 participants had minimal smartphone skills (eg, did not know how to send a text message or download an app). These participants were reliant on their partners or children to use their phones as part of the study.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    The first 38 potential participants completed the eligibility screening interview, of which 31 (82%) were eligible.

    “The first 38 potential participants completed the eligibility screening interview, of which 31 (82%) were eligible.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Weight decreased from baseline to week 6 by 1.1 kg (95% CI -2.0 to -0.3) and to week 12 by 1.3 kg (95% CI -2.5 to -0.1), overall F(2,29)=5.41, P=.01.

    “| Weight (kg) | 83.6 (19.0) | 82.4 (18.3) | 82.3 (18.1) | 5.41,.01 | –1.1 (–2.0 to –0.3) | –1.3 (–2.5 to –0.1) |”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    A total of 99 potential participants formally expressed their interest in the study within four weeks of recruitment opening, at which point expressions of interest were closed.

    “A total of 99 potential participants formally expressed their interest in the study within four weeks of recruitment opening, at which point expressions of interest were closed.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Waist circumference decreased by 1 cm from baseline to week 6, and then another 1 cm to week 12, leading to an overall loss of -2.1 cm (95% CI -3.5 to -0.7, P=.003).

    “Similarly, waist circumference decreased by 1 cm from baseline to week 6, and then another 1 cm to week 12, leading to an overall loss of –2.1cm (95% CI –3.5 to –0.7, P=.003).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    On average, participants lost 1.1 kg from baseline to week 6, and then lost a further 0.2 kg to week 12, resulting in an overall average loss of 1.3 kg (95% CI -2.5 to -0.7, P=.01).

    “On average, participants lost 1.1 kg from baseline to week 6, and then lost a further 0.2 kg to week 12, resulting in an overall average loss of 1.3 kg (95% CI –2·5 to –0·7, P=.01).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Mediterranean diet adherence score increased from baseline to week 6 by 6.0 points (95% CI 4.3 to 7.7) and to week 12 by 5.7 points (95% CI 4.2 to 7.3), overall F(2,29)=44.56, P<.001.

    “| Mediterranean diet adherence score (out of 14) | 3.8 (1.8) | 9.8 (3.7) | 9.6 (3.1) | 44.56, <.001 | 6.0 (4.3 to 7.7) | 5.7 (4.2 to 7.3) |”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    Participants were sequentially booked for baseline assessments until the study quota of 30 participants was filled, with n=31 ultimately enrolled.

    “Participants were sequentially booked for baseline assessments until the study quota of 30 participants was filled, with n=31 ultimately enrolled.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Mediterranean diet adherence increased markedly from baseline to week 6 and then was approximately maintained from week 6 to week 12.

    “Mediterranean diet adherence increased markedly from baseline to week 6 and then was approximately maintained at this level from week 6 to week 12”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The most significant drop off with the virtual health coach occurred between the baseline appointment and week two, with around a quarter of participants failing to complete the first weekly check-in.

    “Interestingly, the most significant drop off with the virtual health coach occurred between the baseline appointment and week two of the program (ie, the first check-in participants were supposed to complete on their own), with around a quarter of participants failing to complete the first weekly check-in.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    At the 6-week assessment, several participants reported difficulty consuming the recommended number of food group servings due to feeling satiated.

    “At the 6-week assessment, several participants reported difficultly consuming the recommended number of food group servings (due to feeling satiated).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingnone given

    Technical problems with Slack or the software interrupted attempts to check in, which generally were rectified after week 1.

    “In some cases, technical problems with Slack or the software interrupted attempts to check in, which generally were rectified after week 1.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Three participants who enrolled had very low smartphone literacy at baseline, which impacted their ability to use the program as intended.

    “Three participants who enrolled in the study were found to have very low smartphone literacy at the baseline session, which impacted their ability to use the program as intended.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    The study attracted mostly women, who were similar to national averages for socioeconomic advantage.

    “Our study attracted mostly women, who were similar to national averages for socioeconomic advantage (national average 1000) [28].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [28] Australian Bureau of Statistics (ABS)ABS2033.0.55.001 - Census of Population and Housing: Socio-economic Indexes for Areas (SEIFA)20162019-09-20https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/2033.0.55.0012016?OpenDocumOpe

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Feasibility was excellent in terms of recruitment, retention, and safety.

    “Feasibility was excellent in terms of recruitment, retention (90% at 12 weeks), and safety (no adverse events).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    An artificially intelligent virtual assistant-led lifestyle-modification intervention was feasible and achieved measurable improvements in physical activity, diet, and body composition at 12 weeks.

    “An artificially intelligent virtual assistant-led lifestyle-modification intervention was feasible and achieved measurable improvements in physical activity, diet, and body composition at 12 weeks.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The observed improvements in physical activity, diet, and body composition are attributable to the MedLiPal intervention rather than to secular trends, regression to the mean, or other factors.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    In highly developed countries, over ninety percent of Australian adults fail the recommended daily intake of vegetables and fruit, and approximately two-thirds do not meet guidelines of 30 min of physical activity most days.

    “In highly developed countries, these unfavorable health behaviors are pervasive; for example, over ninety percent of Australian adults fail the recommended daily intake of vegetables and fruit, and approximately two-thirds do not meet guidelines of 30 min of physical activity most days [2,3].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [2] 4364.0.55.003 - Australian Health Survey: Updated Results, 2011-2012. Daily Intake of Fruit and Vegetables20132019-09-01Australian Bureau of Statisticshttps://www.abs.gov.au/ausstats/abs@.nsf/Lookup/C549D4433F6B74D7CA257B8200179569?opendocument

      Not checked in the free run.

    • cites [3] 4364.0.55.001 - Australian Health Survey: First Results, 2011-1220122019-08-31Australian Bureau of Statisticshttps://www.abs.gov.au/ausstats/abs@.nsf/Lookup/4364.0.55.001main+features12011-12

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Laranjo and colleagues' 2018 systematic review of studies evaluating natural language processing virtual assistants in healthcare identified just 14 studies, most of which focused on mental health (n=6) or clinician decision making (n=4).

    “Laranjo and colleagues’ 2018 systematic review of studies evaluating natural language processing virtual assistants in healthcare identified just 14 studies, most of which focused on mental health (n=6) or clinician decision making (n=4) [9].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [9] source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Weight-loss virtual assistant Lark led to significant weight loss, comparable in magnitude to that achieved in in-person lifestyle interventions.

    “For example, weight-loss virtual assistant “Lark” led to significant weight loss, comparable in magnitude to that achieved in in-person lifestyle interventions.[6]”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [6] source

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductnone given

    The target sample size was set at n=30, which was considered adequate to answer feasibility research questions.

    “The target sample size was set at n=30, which was considered adequate to answer feasibility research questions.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Perceived relevance and the inclusion of social support are associated with increased effect sizes in digital behavior change interventions.

    “Compared to more traditional knowledge transfer methods in digital health (eg, tailored modules), the conversational methods in virtual assistance may be perceived as more personally relevant and humanistic, which is critical given that perceived relevance and the inclusion of social support are associated with increased effect sizes in digital behavior change interventions [7,8].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [7] source

      Not checked in the free run.

    • cites [8] source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Preliminary evidence suggests that the intervention led to sizable improvements in physical activity, diet, and body composition across 12 weeks.

    “Preliminary evidence suggests that the intervention led to sizable improvements in physical activity, diet, and body composition across 12 weeks.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Missing data were imputed using the last observation carried forward.

    “Given that a small amount of data was missing (16% missing data at week 6, 10% at week 12 for primary outcomes), imputation was achieved using the last observation carried forward.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Lark only allowed users to respond to program-directed questions and did not allow users to ask questions.

    “However, Lark only allowed users to respond to program-directed questions and did not allow users to ask questions.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [6] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Sensitivity analyses using complete cases were conducted.

    “Sensitivity analyses using complete cases were also conducted.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Poor diet and physical inactivity are among the leading modifiable causes of death and disease and increase the risk of chronic health conditions such as type 2 diabetes, cardiovascular disease, obesity, cancers, depression, and anxiety.

    “Poor diet and physical inactivity are amongst the leading modifiable causes of death and disease and increase the risk of developing chronic health conditions such as type 2 diabetes, cardiovascular disease, obesity, cancers, depression, and anxiety [1].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [1] Remington, P, 2010, Chronic Disease Epidemiology and Control source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductown data

    Attendance for the 6- and 12-week assessments was 29 and 28, respectively.

    “Attendance for the 6- and 12-week assessments was 29 and 28, respectively (Figure 2).”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductnone given

    Efficacy was assessed using an unadjusted repeated measures ANOVA, conducted on an intention-to-treat principle.

    “Efficacy was assessed using an unadjusted repeated measures ANOVA, conducted on an intention-to-treat principle.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning

    The study had a pre-post design, which meant the study lacked a randomized control comparator.

    “Limitations include the pre-post design, which meant the study lacked a randomized control comparator, and follow-up is limited to three months.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The study confirmed that a virtual health coach diet and physical activity intervention is feasible for older adults and leads to sizable health behavior change and improvements in body composition.

    “This ground-breaking study confirmed that a virtual health coach diet and physical activity intervention is feasible for older adults and leads to sizable health behavior change and improvements in body composition.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Most older adults have sufficient skills to access a virtual health coach program and use it effectively to achieve sizeable behavior and health change.

    “Results suggest there is considerable enthusiasm amongst older adults for this style of intervention and confirm that most older adults have sufficient skills to access a virtual health coach program and use it effectively to achieve sizeable behavior and health change.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The study sample is representative of older adults generally.

  • May combine two claims.

    Study design and conductreasoning

    The study had a mostly hands-off approach, improving ecological validity compared with a study with a high level of human contact.

    “Other study strengths include its mostly hands-off approach (thus improving the ecological validity of the study compared with a study with a high level of human contact), the use of objective health measures, and the use of intention-to-treat analysis, which ensures that the intervention effects are not over-estimated.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisereasoning

    The virtual health coach used natural language processing, which allows users to converse using freely written language, representing a significant advancement over previous computer-tailoring approaches.

    “Our virtual health coach used natural language processing, which allows users to converse using freely written language, representing a significant advancement over previous computer-tailoring approaches, which typically rely on multiple choice.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The magnitude of change in physical activity was similar to or larger than that achieved in previous computer-tailoring interventions.

    “The magnitude of change in physical activity was similar to [36,37] or larger than [38,39] that achieved in previous computer-tailoring interventions.”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [36] source

      Not checked in the free run.

    • cites [37] source

      Not checked in the free run.

    • cites [38] source

      Not checked in the free run.

    • cites [39] source

      Not checked in the free run.

39 of 39

  • The paper relies on this without support.

    A future definitive trial with a dedicated recruitment budget should be able to recruit approximately 400 participants over a year if 30 participants can be recruited in 6 weeks with no advertising budget.

    “We sought to recruit 30 participants to the study within 6 weeks, on the rationale that if this could be achieved with no advertising budget, a future definitive trial with a dedicated recruitment budget should be able to recruit approximately 400 participants over a year.”

  • The paper relies on this without support.

    The observed improvements in physical activity, diet, and body composition are attributable to the MedLiPal intervention rather than to secular trends, regression to the mean, or other factors.

    “An artificially intelligent virtual assistant-led lifestyle-modification intervention was feasible and achieved measurable improvements in physical activity, diet, and body composition at 12 weeks.”

  • The paper relies on this without support.

    The virtual health coach intervention is the first study to evaluate an advanced virtual assistant technology for delivering a personalized lifestyle intervention among older adults.

    “To our knowledge, this is the first study to evaluate the capacity of an advanced virtual assistant technology to deliver a personalized lifestyle intervention amongst older adults.”

  • The paper relies on this without support.

    The study sample is representative of older adults generally.

    “Results suggest there is considerable enthusiasm amongst older adults for this style of intervention and confirm that most older adults have sufficient skills to access a virtual health coach program and use it effectively to achieve sizeable behavior and health change.”
4 of 4

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Among non-RCT studies, Maher et al and Brar Prayaga et al reported highest compliance (5/9, 56%), followed by Masaki et al, Prochaska et al, and To et al (4/9, 44%).

    “Among the non-RCT studies, the studies by Maher et al [22] and Brar Prayaga et al [30] reported the highest compliance (5/9, 56%), followed closely by the studies by Masaki et al [25], Prochaska et al [31], and To et al [32] (4/9, 44%).”

    From Artificial Intelligence-Based Chatbots for Promoting Health Behavioral Changes: Systematic Review. (2023) · cites this paper as [22]

    The Methods (Power and Statistical Analyses; page unavailable) describe a single-group design, supporting the classification of Maher et al as a non-RCT study. The paper does not report compliance of 5/9 (56%) or compare its compliance with the other named studies; those parts of the citing sentence are not supported by this source. (LLM, conf=0.82)

    Experimental — accuracy not yet measured

  • In one of these studies, the engagement rates decreased gradually as the intervention progressed, and in the other study, the engagement rates decreased significantly by midprogram but increased to 70% in the final week, that is, the 12th week.

    “In one of these studies, the engagement rates decreased gradually as the intervention progressed [23], and in the other study, the engagement rates decreased significantly by midprogram but increased to 70% in the final week, that is, the 12th week [22].”

    From Artificial Intelligence-Based Chatbots for Promoting Health Behavioral Changes: Systematic Review. (2023) · cites this paper as [22]

    Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

    Experimental — accuracy not yet measured

  • The need for greater interactivity can also be associated with the fluctuations in user engagement found in 13% (2/15) of studies.

    “The need for greater interactivity can also be associated with the fluctuations in user engagement found in 13% (2/15) of studies [22,23].”

    From Artificial Intelligence-Based Chatbots for Promoting Health Behavioral Changes: Systematic Review. (2023) · cites this paper as [22]

    The Results section, “Engagement With the Virtual Health Coach” (page unavailable), reports that weekly check-in engagement fell from around 70% to 50% before rising again, supporting the reported fluctuation. The paper does not link that pattern to a need for greater interactivity; its Discussion instead considers first-check-in difficulties, smartphone literacy, and technical problems. As a single study, it also does not report the citing document’s 2/15-studies figure. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • In 20% (3/15) of studies, the humanistic yet nonhumanistic construct of AI chatbots provided a safe space for the users to discuss, share, and ask for information on sensitive issues.

    “Third, in 20% (3/15) of studies, the humanistic yet nonhumanistic construct of AI chatbots provided a safe space for the users to discuss, share, and ask for information on sensitive issues [5,22,23,35].”

    From Artificial Intelligence-Based Chatbots for Promoting Health Behavioral Changes: Systematic Review. (2023) · cites this paper as [22]

    The Introduction (page unavailable) says conversational virtual assistants may feel more humanistic, and Results > Other Feasibility Issues (page unavailable) reports that participants were encouraged to ask the coach for dietary advice. Neither passage establishes that the coach provided a safe space to discuss or share sensitive issues; searches for that experience found no supporting passage. The 3/15 figure describes the citing authors’ set of studies, rather than a result this individual study would be expected to report. (LLM, conf=0.79)

    Experimental — accuracy not yet measured

  • Maher et al reported an increase in physical activity by 109.8 minutes (P=.005) and a decrease in average weight and waist circumference by 1.3 kg (P=.01) and 2.1 cm (P=.003), respectively.

    “Maher et al [22] reported an increase in physical activity by 109.8 minutes (P=.005) and a decrease in the average weight and waist circumference by 1.3 kg (P=.01) and 2.1 cm (P=.003), respectively.”

    From Artificial Intelligence-Based Chatbots for Promoting Health Behavioral Changes: Systematic Review. (2023) · cites this paper as [22]

Self-check

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

Each contradiction between two passages listed here was confirmed by an independent check.

  • Average check-in completion percentage does not match the count

    Minor

    An average of 6.9 out of 11 check-ins is 62.7%, not 64%; rounding the reported average to one decimal place cannot account for the difference.

    Reported 64 · recomputed 62.727273

  • Study design described as both case-controlled and single-arm

    Minor

    The Methods calls the study case-controlled, but the abstract describes it as a single-arm pre-post study. A single-arm study has no control group.

  • Weight-loss confidence interval differs from Table 2

    Minor

    The Results text gives an upper confidence limit of -0.7 kg for the 12-week weight change, whereas Table 2 gives -0.1 kg.

3 of 3

Cites

Flagged first, then by citations

Research works printed but not linked: 4

  1. 2003, AIHW, Report Number: 1740242580
  2. Norton, K, 1996, Anthropometrica
  3. 2013, Canberra, NHRMC
  4. 2019, Canberra, AIHW, Cat. no. PHE 244

Cited by

Sorted by citations

Related

Authors
Maher, C. A., Davis, C. R., Curtis, R. G., Short, C. E., & Murphy, K. J.
Affiliations
  • University of South Australia (AU)
  • The University of Melbourne (AU)
Corresponding author
Maher, Carol Ann
ORCID
Venue
JMIR mHealth and uHealth
Publisher
JMIR Publications Inc.
Published
2020-07-10
Volume
8
Issue
7
Pages
e17558
ISSN
2291-5222 (ISSN-L)
DOI
10.2196/17558
PMC
PMC7382010
OpenAlex
W3021104312
Semantic Scholar
91d0e592da91fb71eb86f81c409d9529875b3c30
Cited by
165
References
35
Citations by year
2026: 38 · 2025: 51 · 2024: 22 · 2023: 29 · 2022: 12 · 2021: 11 · 2020: 2
FWCI
14.05
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR mhealth and uhealth (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Minerva Access (University of Melbourne) (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • UniSA Research Outputs Repository (University of South Australia) (submittedVersion, other-oa)
Funders
  • University of South Australia
  • Medical Research Council
  • National Health and Medical Research Council (1125913)
Topics
  • Health Sciences › Health Professions › General Health Professions › Mobile Health and mHealth Applications
  • Health Sciences › Medicine › Physiology › Physical Activity and Health
  • Health Sciences › Health Professions › Health Information Management › Innovation in Digital Healthcare Systems
Keywords
artificial intelligence, chatbots, physical activity, Mediterranean diet, lifestyle modification, body composition, weight loss, blood pressure, older adults, feasibility, recruitment and retention, wearable activity trackers, personalized health interventions, goal setting, community-dwelling adults, sedentary behavior, diet quality, waist circumference, body mass index, digital health interventions, proof of concept
MeSH terms
Proof of Concept Study, Adult, Aged, Artificial Intelligence, Australia, Blood Pressure, Body Composition, Diet, Female, Health Promotion, Humans, Male, Middle Aged, Exercise

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.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

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