Trust,perceived accuracy,and continued use of patient–facing symptom monitoring applications: Cross-sectional study

Abstract

Background

Patient-facing symptom monitoring applications are increasingly used in digital health to support preliminary symptom assessment and health-related decision-making. While perceived accuracy and trust are recognized as important determinants of mHealth adoption, their joint role in shaping users’ continued engagement with symptom monitoring applications remains insufficiently understood.

Objective

This study examined the relationships among perceived accuracy, trust, and continued use intention of symptom monitoring applications and assessed whether trust mediates the association between perceived accuracy and continued use intention.

Methods

A cross-sectional survey was conducted among 300 adult outpatients recruited from three affiliated hospitals in China. Participants evaluated a debranded symptom monitoring application scenario and completed validated measures of perceived accuracy, trust, continued use intention, and eHealth literacy. Multiple linear regression analyses were performed with adjustment for demographic and health-related covariates. Mediation analysis with bootstrap resampling was used to estimate indirect effects.

Results

Perceived accuracy was positively associated with continued use intention (B=0.52, p<.001) and trust (B=0.48, p<.001). Trust was strongly associated with continued use intention (B=0.47, p<.001) and showed a larger effect size than perceived accuracy in the full model. Mediation analysis indicated that trust partially mediated the relationship between perceived accuracy and continued use intention. Findings were consistent across sensitivity and subgroup analyses.

Conclusion

Perceived accuracy and trust jointly influence continued use intention of symptom monitoring applications, with trust acting as a key mechanism linking performance perceptions to sustained engagement. Promoting trust alongside technical performance may support more sustainable use of patient-facing digital health technologies.

Keywords

symptom monitoring applications mHealth trust perceived accuracy continued use intention digital health

1. Introduction

1.1. Symptom monitoring applications and the challenge of continued use

Symptom monitoring applications represent an emerging class of patient-facing health information systems designed to support early symptom assessment and guide decisions about seeking medical care. They promise low-friction, on-demand support for individuals seeking to interpret symptoms, decide whether to seek care, and track health status over time. These functions align with broader goals of improving access, reducing unnecessary utilization, and enabling self-management. At the population level, surveys show that health app use is widespread among smartphone owners, indicating a large potential user base for symptom monitoring tools.¹ At the same time, symptom monitoring applications are frequently positioned as informational aids, which places users’ perceptions at the center of adoption and ongoing engagement decisions.

However, a persistent challenge in digital health is not merely uptake but sustained engagement. The “law of attrition” in eHealth research highlights that discontinuation and drop-off are common, structurally important phenomena that can undermine both effectiveness and evaluation.² Real-world evidence across app-based health domains similarly demonstrates steep declines in retention over time, with many users disengaging shortly after installation. For example, panel-based analyses of popular mental health apps report very low short-term retention rates, suggesting how quickly app use can decay outside controlled trials.³ Large-scale digital health studies also emphasize that retention is a key determinant of whether mobile tools can generate meaningful health impact, and that retention varies substantially by context and design.⁴ For symptom monitoring applications, whose value often depends on repeated use to track symptom trajectories, detect changes, or refine recommendations, continued use is therefore not a secondary outcome but a central requirement for real-world benefit.

Sustained use is also complicated by ongoing concerns about the quality and safety of symptom app outputs. Audit and vignette-based evaluations suggest that diagnostic suggestions and triage advice can be inconsistent and, on average, less accurate than clinician judgment.⁵ In Semigran and colleagues’ BMJ audit of publicly available symptom checkers, correct diagnoses were frequently not ranked first and triage advice showed notable variability, with a tendency toward risk-averse recommendations.⁶ Subsequent studies and reviews have continued to document variability in performance across tools and conditions, reinforcing that accuracy is not a stable, uniform property from the user perspective.^7,8 Even when an app performs reasonably in certain scenarios, users typically encounter uncertainty. They may not know whether a recommendation is tailored, whether it reflects up-to-date evidence, or how the system behaves across symptom combinations and patient profiles. In this environment, continued use may hinge less on objective accuracy and more on perceived accuracy, users’ subjective sense that the app’s feedback fits their experience and seems dependable.

In this study, we focus specifically on patient-facing symptom monitoring applications presented through a standardized hypothetical scenario. These applications are conceptually related to, but distinct from, one-time symptom checker tools used primarily for triage. While symptom checkers are typically designed for single-episode decision support, symptom monitoring applications emphasize repeated use, allowing users to reassess symptoms and seek guidance over time.

1.2. Trust and perceived accuracy in mHealth

In mHealth, trust is widely recognized as a psychological mechanism that shapes whether users rely on an app’s guidance, how they interpret its recommendations, and whether they remain engaged over time. A useful starting point comes from the human factors literature on trust in automation, which defines trust as a determinant of appropriate reliance, that is when and how people decide to depend on automated advice under uncertainty.⁹ Lee and See argue that trust becomes especially influential when system behavior is complex and users cannot fully verify performance, making trust a practical shortcut for reliance decisions.¹⁰ On this foundation, Hoff and Bashir synthesize evidence that trust is not static; it varies across dispositional, situational, and learned components, and it is continually updated through experience with system performance and context.¹¹ These perspectives are highly relevant to symptom monitoring applications, where users typically cannot directly evaluate clinical correctness and must decide whether to continue using the tool based on imperfect feedback loops.

In digital healthcare, trust is also a measurable construct with clear behavioral consequences. A recent systematic review of instruments and empirical findings on trust in digital healthcare highlights that trust is consistently linked to downstream outcomes such as intention to use, acceptance, and perceived usefulness, while being shaped by factors including perceived data accuracy, privacy and security concerns, and the extent of human oversight or professional endorsement.¹² This aligns with mHealth-specific work that treats trust not merely as a general attitude but as a multi-faceted judgment about whether a mobile system is dependable, safe, and aligned with the user’s interests.

While trust captures the willingness to rely, the construct of perceived accuracy captures a more specific belief. It is the users’ subjective assessment that an app’s outputs are correct, credible, or fit their lived health experience. In symptom monitoring contexts, perceived accuracy matters because objective performance is often opaque to users, and feedback may be probabilistic, non-diagnostic, or presented as general guidance. Consequently, users may form judgments based on cues such as coherence with their symptoms, consistency with prior knowledge or clinician advice, and the perceived professionalism of the interface. This is consistent with broader mHealth trust research emphasizing that trustworthiness is inferred from a bundle of cues.¹³ For example, the mobile health app trustworthiness (mHAT) checklist was developed to capture end users’ views on what makes an mHealth app trustworthy, including characteristics related to information quality and reliability alongside usability and transparency elements.¹⁴

Perceived accuracy and trust are conceptually distinct but strongly connected. Perceived accuracy can be viewed as a central performance cue that feeds into trust formation. If users repeatedly feel that an app’s feedback is plausible and helpful, trust is reinforced; if outputs appear inconsistent or misleading, trust erodes. Theoretical models of technology use and post-adoption behavior further support this linkage. In information systems research, continuance frameworks posit that users update beliefs after initial use and that these post-adoption evaluations drive continued intention.¹⁵ In mHealth, this matters because the pathway from initial trial to sustained use typically depends on whether perceived benefits and performance match expectations. It is a process in which perceived accuracy is a natural evaluative signal, and trust becomes a mechanism translating those evaluations into willingness to keep using the tool.

Empirical mHealth evidence also supports the central role of trust for continued use. Vaghefi and Tulu’s longitudinal qualitative work on continued use decisions suggests that users’ post-adoption choices are shaped by perceived effectiveness and evolving judgments about whether the app is worth relying on in real life.¹⁶ Furthermore, broader reviews of mHealth acceptance frameworks suggest that trust frequently appears as an extended construct alongside perceived usefulness, risk, and digital literacy, indicating its recurring importance in explaining adoption and sustained engagement.^17,18 These findings converge with emerging evidence from AI-enabled consumer health apps, where qualitative studies of direct-to-consumer AI-mHealth tools highlight trust as a salient theme in how end users evaluate such apps, including concerns about reliability, transparency, and the boundaries of appropriate reliance.¹⁹ Therefore, focusing on trust and perceived accuracy provides a theoretically grounded approach to explaining why users may persist with or abandon symptom monitoring applications, and it offers actionable implications for design such as clearer communication of uncertainty, transparency of data sources, and interface cues that support users’ confidence without encouraging overreliance.

1.3. Study objectives

This study aims to advance understanding of why users continue to use symptom monitoring applications beyond initial adoption, based on the challenges of sustained engagement in symptom monitoring applications and the theoretical relevance of trust and perceived accuracy in mHealth. Although prior research has examined adoption, acceptance, and usability of mHealth applications, fewer studies have explicitly focused on continued use as a distinct behavioral outcome in the context of symptom monitoring. Moreover, existing work often treats trust or accuracy-related perceptions as background variables.

The primary objective of this study is to examine how perceived accuracy and trust jointly shape users’ continued use intention of symptom monitoring applications. Specifically, the study seeks to (1) assess the association between perceived accuracy and trust, (2) evaluate the independent and combined effects of perceived accuracy and trust on continued use intention, and (3) test whether trust functions as a mediating mechanism linking perceived accuracy to continued use. By focusing on perceived accuracy, the study reflects the reality that most users cannot directly verify the correctness of app outputs and instead rely on subjective judgments when deciding whether to keep using a symptom monitoring application. In addition, this study explores whether prior experience with symptom monitoring applications is associated with differences in these relationships. Continued use decisions may differ between users who have previously interacted with such apps and those who are considering use based on hypothetical or limited exposure. Examining this potential heterogeneity allows for a deeper understanding of trust formation and reliance decisions across different stages of user experience.

1.4. Hypotheses

This study proposes a set of hypotheses linking perceived accuracy, trust, and continued use intention in the context of symptom monitoring applications. These hypotheses are grounded in the premise that users often face uncertainty when interpreting app-generated health feedback and therefore rely on subjective evaluations.

(1) Perceived accuracy reflects users’ judgments about whether the information and feedback provided by a symptom monitoring app appear correct, credible, and consistent with their health experience. Perceptions of information quality and accuracy have been identified as important antecedents of trust.²⁰ Perceived accuracy reflects users’ evaluation of the reliability of app-generated information. When users perceive the system outputs as accurate, they are more likely to develop trust in the application.

H1. Perceived accuracy of symptom monitoring applications is positively associated with user trust.

(2) Trust has been shown to play a critical role in shaping reliance and continued engagement with digital systems. In mHealth contexts, trust has been linked to acceptance, intention to use, and willingness to rely on app-based recommendations.²¹ From a behavioral perspective, trust reduces perceived risk and cognitive effort, making continued use more likely once an app has been tried.²² Higher trust is expected to increase users’ intention to continue use the system.

H2. User trust in symptom monitoring applications is positively associated with continued use intention.

(3) Perceived accuracy may influence continued use intention both directly and indirectly through trust. Users who believe that an app provides accurate and reliable feedback may be more inclined to continue using it, even if trust is not explicitly articulated.²³ Post-adoption models of information systems use suggest that users update their intentions based on performance-related evaluations after initial exposure, making perceived accuracy a plausible direct predictor of continued use.²⁴

H3. Perceived accuracy of symptom monitoring applications is positively associated with continued use intention.

(4) Accordingly, perceived accuracy may influence continued use both directly and indirectly through trust, consistent with a mediation mechanism.²⁵ In this view, trust functions as a psychological pathway through which accuracy-related perceptions are translated into sustained engagement.²⁶ Such a mechanism aligns with both trust calibration perspectives in human–automation interaction and continuance models in information systems research.

H4. Trust mediates the association between perceived accuracy and continued use intention of symptom monitoring applications.

(5) Finally, users’ prior experience with symptom monitoring applications may shape how trust relates to continued use.²⁷ Individuals with prior experience may have more concrete reference points for evaluating app feedback, potentially strengthening the link between trust and continued use decisions.²⁸ In contrast, users without prior experience may rely more heavily on general attitudes or hypothetical judgments. Therefore, this study explores whether prior use experience is associated with differences in the strength of the trust–continued use relationship.

H5. The positive association between trust and continued use intention is stronger among users with prior experience using symptom monitoring applications than among those without prior experience.

2. Methods

2.1. Study design

This study employed a multicenter cross-sectional survey design to examine the associations among perceived accuracy, trust, and continued use intention of symptom monitoring applications. A cross-sectional approach was chosen because the primary aim was to investigate users’ perceptions and behavioral intentions at a single point in time. This design is commonly used in mHealth research to assess user perceptions, attitudes, and intentions related to mobile health technologies.

Data were collected between October 2025 and November 2025 from adult outpatient populations across three affiliated hospitals within the same university medical system (Zhejiang Chinese Medical University) in China. Conducting the survey across multiple clinical sites allowed for recruitment of a heterogeneous sample of smartphone users with diverse health conditions and prior experiences, thereby enhancing the generalizability of the findings within the clinical outpatient setting. All participants completed the same standardized questionnaire under identical study procedures, regardless of recruitment site. The study focused on continued use intention as the primary outcome, reflecting the importance of sustained engagement for the effectiveness of symptom monitoring applications. Perceived accuracy and trust were specified a priori as key explanatory variables, based on existing theory in mHealth, trust in automation, and post-adoption technology use. Given the observational nature of the study, no experimental manipulation or random assignment was involved, and causal inferences were not drawn. An overview of the study design is shown in Figure 1. The questionnaire is provided in Supplementary Material 2.

Figure 1.

Overview of the study design and analytical workflow. The figure summarizes the cross-sectional, multi-center survey design used to examine perceived accuracy, trust, and continued use intention of a debranded symptom monitoring application, along with the key measures and analytical approach.

2.2. Participants and recruitment

Participants were recruited from the outpatient clinics of three tertiary hospitals affiliated with Zhejiang Chinese Medical University. These hospitals represent comprehensive, high-volume clinical settings that serve diverse patient populations, including general medicine and specialty outpatient services. Recruiting participants across multiple affiliated hospitals enabled a multicenter sampling approach while maintaining consistency in institutional context and study procedures. Eligible participants met the following inclusion criteria: (1) aged 18 years or older; (2) ownership and regular use of a smartphone; (3) ability to read and understand Chinese; and (4) capacity to provide informed consent. Participants were not required to be current users of symptom monitoring applications; both individuals with and without prior experience using such apps were eligible, allowing examination of continued use intentions across different experience levels. Individuals were excluded if they were unable to complete the questionnaire independently or declined to participate.

Recruitment took place in outpatient waiting areas at each participating hospital. Potential participants were approached by trained research staff and briefly informed about the purpose of the study, emphasizing that participation was voluntary and unrelated to their medical care. Interested individuals were invited to scan a QR code using their smartphones, which directed them to the online survey platform. No financial incentives were provided for participation. The target sample size of approximately 300 participants was determined based on statistical and feasibility considerations. Prior guidelines suggest that samples exceeding 200 are generally sufficient for stable estimation in regression models with multiple predictors,²⁹ and that samples of 200–400 are adequate to detect medium-sized mediation effects.³⁰ Recruitment continued until the target sample size was reached, with efforts made to achieve a balanced distribution of participants across the three hospital sites. All participants completed the survey anonymously, and no personally identifiable information was collected.

2.3. Procedure

The study was conducted using a standardized survey procedure across all three participating hospitals. Data collection took place in outpatient waiting areas, where eligible participants were invited to complete a self-administered questionnaire on their own smartphones. The survey was implemented using a secure online survey platform and required approximately 8–10 minutes to complete. Upon accessing the survey link, participants were first presented with an electronic informed consent form describing the purpose of the study, the voluntary nature of participation, expected time commitment, and data confidentiality. Participants were informed that their responses would be collected anonymously and would not affect their medical care in any way. Only participants who provided informed consent were able to proceed to the questionnaire.

To minimize brand-related bias and prior attitudes toward specific commercial products, the study employed a debranded app description. Participants were asked to consider a hypothetical symptom monitoring application described in general terms, without reference to any specific brand, developer, or platform. The description outlined core functionalities commonly found in symptom monitoring applications, such as entering symptoms, receiving possible condition suggestions, and obtaining general guidance on whether to seek medical care. The description explicitly stated that the app was intended to provide informational support. The hypothetical application used in this study was designed to represent a generalized symptom monitoring application with features that may support repeated use.

Following the app description, participants were presented with three short vignette scenarios designed to reflect typical use situations of symptom monitoring applications. Each vignette described a brief health-related scenario in which an individual experiences symptom and uses the app to obtain feedback. The vignettes varied in symptom type and context to enhance realism and encourage participants to consider a range of plausible situations. After each vignette, participants were asked to rate their perceptions of the app’s accuracy, their level of trust in the app, and their intention to continue using the app in similar situations. After completing the vignette-based questions, participants were asked to respond to a series of standardized questionnaire items assessing perceived accuracy, trust, and continued use intention at a general level. Additional questions captured demographic information, health-related background, and prior experience with symptom monitoring applications. An attention-check item was embedded within the questionnaire to assess response quality.

Participants were first exposed to three vignette scenarios to provide a structured and standardized context for evaluating the hypothetical application. These vignette-based items captured scenario-specific perceptions of accuracy, trust, and willingness to use the app in different situations. For the main analyses, the primary constructs were derived from the standardized multi-item scales presented after the vignette section. These measures reflect participants’ overall evaluation of the application.

2.4. Measures

All study measures were assessed using a structured self-report questionnaire. Unless otherwise specified, items were rated on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Higher scores indicated higher levels of the corresponding construct. The primary outcome and explanatory variables used in regression and mediation analyses were based on the global measures.

(1) Perceived Accuracy. Perceived accuracy refers to participants’ subjective evaluation of how accurate, reliable, and appropriate the app’s feedback appeared. This construct was measured using four items adapted from prior mHealth and information quality research, with wording adapted to the symptom monitoring application context.^31,32 Items assessed the extent to which participants perceived the app’s outputs as correct, credible, and consistent with their expectations. Sample items included: “The app provides accurate health-related feedback” and “The information provided by the app seems reliable.” Item scores were averaged to create a composite perceived accuracy score, with higher values indicating greater perceived accuracy.

(2) Trust. Trust in the symptom monitoring application was measured using five items reflecting users’ confidence in the app’s dependability, integrity, and suitability for health-related decision support. The items were informed by established frameworks in trust in automation and digital health trust research, adapted to the mHealth context. Sample items included: “I trust the app to provide dependable health information” and “I feel confident relying on the app’s feedback when assessing my symptoms.” Responses were averaged to form an overall trust score, with higher scores indicating greater trust in the app.

(3) Continued Use. Continued use was operationalized primarily as continued use intention, reflecting participants’ willingness to keep using the symptom monitoring application in the future. This construct was measured using four items adapted from prior studies on mHealth continuance and post-adoption behavior. Sample items included: “I intend to continue using this app in the future” and “I would consider using this app again when experiencing similar symptoms.” Item responses were averaged to create a composite continued use intention score, with higher scores indicating stronger intention to continue using the app. In addition, participants who reported prior experience using symptom monitoring applications were asked to indicate their frequency of past use on a 7-point scale. This variable was used descriptively and as a supplementary behavioral indicator. The outcome variable reflects anticipated continued use intention in response to a hypothetical application scenario.

(4) Covariates. Several covariates were included to account for potential confounding factors. Demographic variables included age, gender, and education level. Health-related background variables included self-reported presence of chronic conditions. Prior experience with symptom monitoring applications was assessed using a binary item yes or no. To account for differences in digital capability, eHealth literacy was measured using a brief three-item scale adapted from existing eHealth literacy instruments, capturing participants’ perceived ability to find, understand, and use digital health information. Recruitment site was recorded and included as a covariate in multivariable analyses to account for potential site-level differences.

2.5. Statistical analysis

Statistical analyses were conducted to examine the relationships among perceived accuracy, trust, and continued use intention of symptom monitoring applications. All statistical analyses were conducted using Python, including pandas for data management and statsmodels for regression and mediation analyses. Statistical significance was set at a two-tailed P value < .05.

Descriptive statistics were computed for all study variables, including means, standard deviations, and frequency distributions, to summarize participant characteristics and key constructs. Internal consistency of multi-item scales was assessed using Cronbach’s alpha, with values of .70 or higher considered acceptable.³³ Composite scores for perceived accuracy, trust, and continued use intention were calculated by averaging the corresponding item responses.

Multiple linear regression analyses were performed to test the direct associations specified in Hypotheses 1–3. Trust was regressed on perceived accuracy to test H1. Continued use intention was regressed on trust and perceived accuracy to test H2 and H3. All regression models adjusted for relevant covariates, including age, gender, education level, presence of chronic conditions, prior experience with symptom monitoring applications, eHealth literacy, and recruitment site. Variance inflation factors were examined to assess multicollinearity.

Mediation analysis was conducted to test Hypothesis 4, which proposed that trust mediates the association between perceived accuracy and continued use intention. The mediation effect was examined using a bootstrapping approach with 5000 resamples to estimate the indirect effect and its 95% confidence interval. An indirect effect was considered statistically significant if the confidence interval did not include zero. Both direct and indirect effects were reported to assess partial mediation.

Additionally, to explore Hypothesis 5, which examined whether prior experience with symptom monitoring applications was associated with differences in the trust–continued use relationship, interaction analyses were conducted by including an interaction term between trust and prior app experience in the regression model.

3. Results

3.1. Participant characteristics

A total of 300 participants were included in the final analysis. Participants had a mean age of 38.0 years (SD = 11.6), reflecting a predominantly working-age outpatient population. Slightly more than half of the participants were female (54.3%), while 45.7% were male. Approximately 28.0% of participants reported having at least one chronic health condition, and 50.3% of participants reported having used such apps before. Participants were recruited from three affiliated hospitals, with a relatively balanced distribution across recruitment sites. Detailed demographic and clinical characteristics of the participants are presented in Table 1.

Table 1.

Participant characteristics (N=300).

Characteristic	Value
Age, mean (SD), years	38.0 (11.6)
Gender, n (%)
Male	137 (45.7)
Female	163 (54.3)
Education level, n (%)
High school or below	76 (25.3)
College	68 (22.7)
Bachelor’s degree	115 (38.3)
Master’s degree or above	41 (13.7)
Chronic condition (self-reported), n (%)	84 (28.0)
Prior use of symptom monitoring applications, n (%)
Yes	151 (50.3)
No	149 (49.7)
Recruitment site, n (%)
Hospital 1	113 (37.7)
Hospital 2	94 (31.3)
Hospital 3	93 (31.0)

^aHospital 1–3 refer to three tertiary hospitals affiliated with the same university medical system.

3.2. Descriptive statistics and correlations

Descriptive statistics and bivariate Pearson correlation coefficients among the core study constructs are presented in Table 2. Overall, participants reported moderate levels of perceived accuracy (mean 4.08, SD 1.57), trust (mean 4.18, SD 1.63), continued use intention (mean 4.19, SD 1.81), and eHealth literacy (mean 4.07, SD 1.56), all slightly above the midpoint of the 7-point Likert scale. Bivariate correlations indicated that perceived accuracy was positively associated with trust (r = 0.48, p < .001) and continued use intention (r = 0.50, p < .001). Trust showed an even stronger positive correlation with continued use intention (r = 0.57, p < .001), suggesting that higher trust in symptom monitoring applications was closely related to greater intention to continue using them. In addition, eHealth literacy exhibited modest but statistically significant positive correlations with perceived accuracy (r = 0.24, p < .001), trust (r = 0.16, p < .01), and continued use intention (r = 0.19, p < .01). These associations provide preliminary support for the hypothesized relationships and justify subsequent multivariable regression and mediation analyses. Additional correlations involving demographic and experience variables are provided in Supplementary Table S1.

Table 2.

Descriptive statistics and correlations (N=300).

#	Variable	Mean	SD	1	2	3	4
1	Perceived accuracy	4.08	1.57	—
2	Trust	4.18	1.63	0.48***	—
3	Continued use intention	4.19	1.81	0.50***	0.57***	—
4	eHealth literacy	4.07	1.56	0.24***	0.16**	0.19**	—

^aValues in columns 1–4 are Pearson correlation coefficients (r). p values:

^*p<.05;

^**p<.01;

^***p<.001.

3.3. Main regression results

3.3.1. Predictors of trust

Table 3-5 reports the results of the multiple linear regression analyses examining predictors of trust and continued use intention for symptom monitoring applications. As shown in Table 3, perceived accuracy was a strong and significant predictor of trust (B = 0.48, SE = 0.04, β = 0.48, p < .001), indicating that higher perceived accuracy of symptom monitoring applications was associated with greater trust in the apps. eHealth literacy was also positively associated with trust (B = 0.14, SE = 0.04, β = 0.16, p = .002). In contrast, age (B = 0.01, SE = 0.01, p = .42), gender (female vs male; B = 0.06, SE = 0.09, p = .50), education level (B = 0.05, SE = 0.04, p = .21), chronic condition status (B = 0.07, SE = 0.11, p = .53), and prior symptom monitoring application use (B = 0.05, SE = 0.09, p = .58) were not significantly associated with trust. The model explained 30% of the variance in trust (R² = 0.30).

Table 3.

Multiple linear regression analyses predicting trust and continued use intention. Outcome: Trust.

Predictor	B (SE)	β	P value
Perceived accuracy	0.48 (0.04)	0.48	<.001
Age	0.01 (0.01)	0.04	.42
Gender (female)	0.06 (0.09)	0.03	.50
Education level	0.05 (0.04)	0.06	.21
Chronic condition (yes)	0.07 (0.11)	0.03	.53
Prior app use (yes)	0.05 (0.09)	0.03	.58
eHealth literacy	0.14 (0.04)	0.16	.002
R²	0.30

Table 4.

Outcome: Continued use intention (Model 1).

Predictor	B (SE)	β	P value
Perceived accuracy	0.52 (0.05)	0.50	<.001
Age	0.01 (0.01)	0.05	.33
Gender (female)	0.08 (0.10)	0.04	.44
Education level	0.04 (0.04)	0.05	.29
Chronic condition (yes)	0.02 (0.12)	0.01	.87
Prior app use (yes)	0.01 (0.10)	0.00	.96
eHealth literacy	0.17 (0.05)	0.19	.001
R²	0.33

Table 5.

Outcome: Continued use intention (Model 2).

Predictor	B (SE)	β	P value
Perceived accuracy	0.27 (0.05)	0.26	<.001
Trust	0.47 (0.05)	0.57	<.001
Age	0.01 (0.01)	0.05	.29
Gender (female)	0.06 (0.09)	0.03	.51
Education level	0.03 (0.04)	0.04	.37
Chronic condition (yes)	0.01 (0.11)	0.00	.94
Prior app use (yes)	0.02 (0.09)	0.01	.84
eHealth literacy	0.10 (0.04)	0.11	.018
R²	0.46

^aUnstandardized coefficients (B) with standard errors (SE) are reported, along with standardized coefficients (β).

^bGender was coded as 0 = male, 1 = female.

^cBinary variables were coded as 0/1.

3.3.2. Predictors of continued use intention

In Table 4 (Model 1), perceived accuracy was a significant positive predictor of continued use intention (B = 0.52, SE = 0.05, β = 0.50, p < .001), after adjusting for all covariates. eHealth literacy also showed a significant positive association with continued use intention (B = 0.17, SE = 0.05, β = 0.19, p = .001). None of the demographic or health-related covariates—including age, gender, education level, chronic condition status, or prior app use—were significantly related to continued use intention (all p > .05). This model accounted for 33% of the variance in continued use intention (R² = 0.33).

In Table 5 (Model 2), trust was added to the model alongside perceived accuracy. Trust emerged as a strong predictor of continued use intention (B = 0.47, SE = 0.05, β = 0.57, p < .001). Although the coefficient for perceived accuracy was attenuated compared with Model 1, it remained statistically significant (B = 0.27, SE = 0.05, β = 0.26, p < .001), suggesting that perceived accuracy continued to exert a direct effect on continued use intention after accounting for trust. eHealth literacy remained a significant predictor in this model (B = 0.10, SE = 0.04, β = 0.11, p = .018), whereas all other covariates were not significant (all p > .05). The inclusion of trust substantially increased the explained variance, with R² rising to 0.46, indicating improved model fit. Overall, these findings demonstrate that perceived accuracy is a key determinant of both trust and continued use intention, and that trust plays a central role in explaining participants’ reported intention to use the application in a hypothetical evaluation context.

3.4. Mediation analysis

To examine whether trust mediated the association between perceived accuracy and continued use intention of symptom monitoring applications, a mediation analysis was conducted using a regression-based approach with bootstrap resampling. Perceived accuracy was significantly associated with trust (path a: B = 0.48, SE = 0.04, p < .001), and trust was significantly associated with continued use intention when controlling for perceived accuracy and covariates (path b: B = 0.47, SE = 0.05, p < .001). The total effect of perceived accuracy on continued use intention was significant (path c: B = 0.52, SE = 0.05, p < .001). When trust was included in the model, the direct effect of perceived accuracy on continued use intention was reduced but remained statistically significant (path c′: B = 0.27, SE = 0.05, p < .001), indicating partial mediation.

The indirect effect of perceived accuracy on continued use intention through trust was statistically significant, with a bootstrap estimate of B = 0.23 and a 95% confidence interval that did not include zero (95% CI: 0.17–0.30; Supplementary Table S1). This finding provides evidence that trust partially mediates the relationship between perceived accuracy and continued use intention. These results suggest that perceived accuracy influences participants’ reported intention to use the application both directly and indirectly through trust in the hypothetical scenario. The mediation model is illustrated in Figure 2. The indirect effect of perceived accuracy on continued use intention via trust was statistically significant, as indicated by a bootstrap 95% confidence interval that did not include zero. Detailed regression coefficients for the mediation model are provided in Supplementary Table S2.

Figure 2.

Mediation model of trust in the relationship between perceived accuracy and continued use intention.

3.5. Sensitivity and subgroup analyses

To assess the robustness of the main findings, a series of sensitivity and subgroup analyses were conducted. The results are summarized in Table 6. Across subgroups defined by prior symptom monitoring application use, the associations between perceived accuracy, trust, and continued use intention remained largely consistent. Among participants with prior app use, both perceived accuracy (B = 0.255, SE = 0.087, p = .004) and trust (B = 0.618, SE = 0.087, p < .001) were significant predictors of continued use intention, with the model explaining 51.0% of the variance. Similar patterns were observed among participants without prior app use, although the association between perceived accuracy and continued use intention was somewhat stronger in this subgroup (B = 0.348, SE = 0.087, p < .001), while the effect of trust remained robust (B = 0.333, SE = 0.080, p < .001).

Table 6.

Sensitivity and subgroup analyses for continued use intention.

Analysis	Subgroup/Specification	N	Perceived accuracy → Continued use (B [SE], p)	Trust → Continued use (B [SE], p)	R²
Subgroup	Prior symptom app use = Yes	151	0.255 [0.087], p=.004	0.618 [0.087], p<.001	0.510
Subgroup	Prior symptom app use = No	149	0.348 [0.087], p<.001	0.333 [0.080], p<.001	0.316
Subgroup	Chronic condition = Yes	84	0.180 [0.115], p=.123	0.686 [0.107], p<.001	0.547
Subgroup	Chronic condition = No	216	0.384 [0.074], p<.001	0.380 [0.071], p<.001	0.356
Sensitivity	Excluding recruitment site indicators	300	0.325 [0.061], p<.001	0.471 [0.058], p<.001	0.396
Sensitivity	Excluding eHealth literacy	300	0.335 [0.061], p<.001	0.473 [0.058], p<.001	0.395
Sensitivity	Robust SE (HC3) for primary model	300	0.323 [0.062], p<.001	0.471 [0.058], p<.001	0.396

^aUnstandardized coefficients (B) with standard errors (SE) are reported.

^bThe sensitivity models reestimated the primary regression under alternative specifications.

^cSubgroup models were estimated within each subgroup using the same set of covariates unless otherwise specified.

Subgroup analyses by chronic condition status showed that trust was consistently and strongly associated with continued use intention in both groups. Among participants with chronic conditions, trust demonstrated a large effect size (B = 0.686, SE = 0.107, p < .001), whereas the association between perceived accuracy and continued use intention was not statistically significant (B = 0.180, SE = 0.115, p = .123). In contrast, among participants without chronic conditions, both perceived accuracy (B = 0.384, SE = 0.074, p < .001) and trust (B = 0.380, SE = 0.071, p < .001) were significant predictors. Sensitivity analyses further supported the robustness of the findings. Excluding recruitment site indicators did not materially alter the estimated effects of perceived accuracy (B = 0.325, SE = 0.061, p < .001) or trust (B = 0.471, SE = 0.058, p < .001). Similarly, removing eHealth literacy from the model produced nearly identical results. When robust standard errors (HC3) were applied, the key associations between perceived accuracy, trust, and continued use intention remained statistically significant, with minimal changes in coefficient estimates. Overall, these sensitivity and subgroup analyses indicate that the central findings of this study are stable across different participant subgroups and analytical specifications, reinforcing the robustness of the observed relationships between perceived accuracy, trust, and continued use intention of symptom monitoring applications in the hypothetical evaluation context.

4. Discussion

4.1. Principal findings

This study investigated how perceived accuracy and trust are associated with users’ continued use intention of symptom monitoring applications in a hypothetical scenario. The findings highlight that these constructs represent distinct but interrelated attitudinal components shaping how users evaluate such applications.

Perceived accuracy showed a clear positive association with continued use intention. After controlling for demographic characteristics, health status, prior app experience, and eHealth literacy, users who perceived the app’s symptom assessments as more accurate reported stronger intentions to continue using the app. This result highlights perceived accuracy as an important performance-related factor underlying long-term engagement, particularly in symptom monitoring contexts where users rely on algorithmic interpretations of personal health information.

Trust emerged as an even stronger predictor of continued use intention. When both perceived accuracy and trust were included in the regression model, trust remained robustly associated with continued use intention, while the coefficient for perceived accuracy was attenuated but remained statistically significant. This pattern indicates that trust captures an essential relational dimension of user reliance that extends beyond evaluations of technical performance alone.

Mediation analysis further clarified the relationship between perceived accuracy, trust, and continued use intention. Perceived accuracy was strongly associated with trust, and trust, in turn, was positively associated with continued use intention. The significant indirect effect, alongside a remaining direct effect of perceived accuracy, supports a partial mediation pattern. These findings suggest that perceived accuracy contributes to sustained use in two ways: directly, by shaping users’ cognitive evaluations of system performance, and indirectly, by fostering trust in the app as a dependable health support tool.

The robustness of these relationships was supported by sensitivity and subgroup analyses. Across alternative model specifications and user subgroups, trust consistently demonstrated a strong association with continued use intention, while the direction and magnitude of the effects of perceived accuracy remained stable. This consistency strengthens confidence in the central role of trust as a mechanism linking performance perceptions to sustained engagement with symptom monitoring applications. The results indicate that perceived accuracy and trust function in a complementary manner in shaping participants’ continued use intention of symptom monitoring applications in the evaluation context.

4.2. Trust as a mechanism linking perceived accuracy and continued use

The present study highlights trust as a central mechanism linking perceived accuracy to participants’ continued use intention of symptom monitoring applications in a hypothetical scenario. While prior mHealth research has consistently identified perceived accuracy and trust as important correlates of technology adoption and engagement,³⁴ our findings clarify how these factors interact in the specific context of symptom monitoring, where users rely on apps to interpret personal health information and support self-management decisions. Our results demonstrate that perceived accuracy is strongly associated with trust in symptom monitoring applications, which in turn substantially predicts users’ intentions to continue using these tools. This pattern suggests that users do not translate accuracy perceptions into continued engagement in a purely instrumental manner. Instead, perceived accuracy appears to function as a foundational signal that shapes users’ trust in the app, consistent with prior work on trust in automation showing that users rely on trust when deciding whether to depend on algorithmic systems.³⁵ This mechanism is particularly salient in symptom monitoring contexts, where uncertainty is inherent and users often lack the expertise to directly verify the correctness of algorithmic outputs.

The mediating role of trust observed in this study extends existing models of mHealth adoption by moving beyond direct associations between system characteristics and use intentions. This mediating role of trust aligns with recent evidence from medical AI research indicating that trust and perceived uncertainty jointly shape users’ reliance on algorithmic recommendations, beyond objective system performance alone.³⁶ Prior studies have shown that trust influences acceptance and reliance on digital health technologies, especially when users face information asymmetry and potential risks.^12,37 Our findings build on this work by empirically demonstrating that trust partially explains how perceived accuracy affects continued use intention, even after accounting for demographic factors, health status, prior app experience, and eHealth literacy. The persistence of a direct effect of perceived accuracy alongside the indirect effect through trust further suggests that accuracy perceptions may influence continued use both cognitively, by shaping evaluations of system performance, and relationally, by fostering trust. Specifically, the robustness of this mediation pattern across sensitivity and subgroup analyses highlights the general relevance of trust as a linking mechanism. Trust remained a strong predictor of continued use intention across different user groups and model specifications, reinforcing its central role in sustained engagement with symptom monitoring applications.³⁸ Therefore, these findings suggest that trust serves as a key psychological pathway through which users convert performance-related perceptions into ongoing use decisions in mHealth environments characterized by uncertainty and personal relevance.

The primary contribution of this study lies in clarifying the mechanism through which perceived accuracy is associated with continued use via trust. This contributes to existing mHealth and trust literature by moving beyond simple associations and providing an explicit mediation structure linking performance-related perceptions to behavioral intentions in a controlled scenario context. Additionally, an important consideration is the potential conceptual proximity between perceived accuracy and trust, particularly given that both constructs were measured using self-reported items at the same time point. It is possible that part of the observed association reflects a general positive evaluation of the application. However, perceived accuracy and trust are conceptually distinguishable. Perceived accuracy refers to users’ evaluation of the informational quality or correctness of the app’s outputs, whereas trust reflects a broader psychological willingness to rely on the system under conditions of uncertainty. In this sense, perceived accuracy can be understood as a performance-related judgment, while trust represents a relational and risk-oriented belief.

4.3. Implications for mHealth app design and use

The results have practical implications for the design, communication, and design implications of symptom monitoring applications. Because perceived accuracy is linked to continued use both directly and through trust, design strategies that strengthen performance perceptions and support trust formation are likely to be especially important for sustained engagement. These implications should be interpreted as hypothesis-generating and based on intention-level findings in a hypothetical evaluation context.

Perceived accuracy is not only a technical property but also a user-facing experience shaped by how outputs are presented and contextualized. Symptom monitoring applications can improve perceived accuracy by offering clearer input guidance and structured symptom entry, minimizing ambiguity in user-provided data, and presenting outputs in a way that aligns with users’ mental models of symptom assessment. Qualitative evidence suggests that users often evaluate symptom checker experiences through usability, coherence of questioning, and whether recommendations “make sense” in context, which may diverge from objective performance benchmarks. Design choices that reduce confusion during symptom input and that communicate the basis for recommendations can therefore support more stable performance perceptions in potential use.²⁸

Trust-oriented design should be treated as a primary pathway to sustained use. Recent evidence synthesizing factors that shape trust in medical AI highlights the centrality of transparency, explainability, and clarity around data use, in addition to performance-related cues.²⁰ In symptom monitoring applications, trust can be supported through bounded transparency that explains what the app can and cannot do, what data sources or rules are used (at an appropriate level of detail), and how uncertainty is handled. Broader trust scholarship also emphasizes that transparency and explainability are key determinants of trustworthiness perceptions, but that they must be designed to be usable and not overwhelming.^39,40 Practically, this supports interface patterns such as brief rationale statements, confidence ranges expressed in plain language, and clear escalation guidance when symptoms are severe or uncertain.

The mediation results also suggest a role for trust calibration strategies that help users rely on the app appropriately. Symptom monitoring applications should provide guidance that matches risk and uncertainty, including prompts to seek clinical care when red flags are present and reminders that app outputs do not replace diagnosis. Evidence from a multicenter randomized trial evaluating symptom checker use in acute care contexts suggests that integrating symptom checker outputs into care pathways is not straightforward and that perceived benefits may not translate into measurable improvements without thoughtful workflow integration.⁴¹ This implies that deployment strategies should consider where symptom monitoring fits in the patient journey, how outputs are meant to be used before visits, and whether clinicians have a feasible way to acknowledge or incorporate patient-generated app summaries.

Sustained use also depends on engagement and retention mechanisms that extend beyond momentary satisfaction. Large-scale analyses of adherence and retention in digital health studies underscore that retention is fragile and influenced by study and user factors, reinforcing the importance of designing for ongoing participation.⁴² For symptom monitoring applications, engagement strategies should be tied to meaningful user goals, reduce burden, and offer value that accumulates over time. Additionally, engagement design should remain consistent with trust calibration, thus reminders and persuasive elements should not encourage inappropriate reliance in situations where uncertainty is high.

Privacy and data governance are also tightly linked to trust, particularly for symptom monitoring where data are personal and potentially sensitive. User-centered evidence in mHealth shows that privacy-related behaviors and concerns are shaped by trust, autonomy, and policy transparency, and that clearer policies can improve perceived benefits while also increasing awareness of privacy risks.⁴³ Complementary qualitative work on direct-to-consumer AI-mHealth indicates that users care about what data are collected, how they might be used in the future, and whether guidance comes from credible health or regulatory organizations, suggesting that trust is strengthened by visible governance and accountability signals.¹⁹ Accordingly, symptom monitoring applications should implement privacy by design practices, communicate data use in plain language, and provide user controls for data access, sharing, and deletion consistent with broader digital health governance priorities.⁴⁴

Finally, the role of eHealth literacy in the models suggests that effective use may depend on users’ ability to interpret app information and act on recommendations. This supports onboarding and in-app education that explains how to enter symptoms, how to interpret outputs, and when to seek professional care. Such educational supports can reduce misunderstanding, strengthen trust calibration, and improve the likelihood that symptom monitoring applications are used safely and consistently across users with varying digital health skills.

4.4. Strength and limitations

This study focuses on continued use intention, addressing a less examined but practically important stage of mHealth engagement. By modeling trust as a mediating mechanism between perceived accuracy and continued use, the study advances understanding of how performance-related perceptions translate into sustained engagement with symptom monitoring applications. The use of a theory-driven analytical framework, combined with multiple regression, mediation analysis, and sensitivity checks, provides converging evidence for the stability of the observed relationships. Meanwhile, the recruitment strategy enhances ecological validity by capturing users who are actively engaged in health care contexts, where symptom monitoring applications are more likely to be used alongside clinical decision-making. Compared with convenience samples drawn from general online populations, this setting allows the findings to better reflect potential use conditions in which uncertainty, personal relevance, and health-related risk are emphasized. Additionally, the inclusion of relevant covariates, such as prior app experience and eHealth literacy, further strengthens confidence that the observed effects are not merely attributable to individual differences in digital skills or familiarity with health apps.

Several limitations should be acknowledged. The cross-sectional design precludes causal inference regarding the relationships among perceived accuracy, trust, and continued use intention. Although the proposed mediation model is theoretically grounded and statistically supported, longitudinal or experimental designs are needed to establish temporal ordering and causal mechanisms more definitively. In addition, all measures relied on self-reported data, which may be subject to recall bias or social desirability effects. Future research could benefit from incorporating objective usage data or behavioral measures of continued engagement. The study focused on users’ intentions to continue using symptom monitoring applications. While intention is a well-established proximal predictor of behavior, discrepancies between intended and observed use may arise in practice due to contextual constraints, changing health needs, or evolving perceptions over time. Moreover, the study was conducted within a single national context, which may limit the generalizability of the findings to health systems, regulatory environments, or cultural settings with different norms around digital health use and trust. Finally, the use of a debranded app description and vignette-based scenario was intended to reduce brand-related bias and enhance internal validity. However, this approach may not fully capture the complexity of real-world interactions with specific symptom monitoring applications, where brand reputation, interface design, and prior experiences can shape perceptions and trust. In addition, the outcome reflects anticipated continuance and should therefore be interpreted as an intention-based evaluation under hypothetical conditions. Given the cross-sectional and self-reported nature of the data, the mediation results should be interpreted with caution. The observed relationships may partially reflect shared method variance or a general evaluative orientation toward the application. Future research using longitudinal designs, experimental manipulations, or behavioral usage data would be valuable in further disentangling these constructs and testing the proposed mechanism more rigorously.

5. Conclusions

This study examined how perceived accuracy and trust jointly influence continued use intention of symptom monitoring applications in a controlled scenario context. The findings indicate that perceived accuracy and trust play complementary roles in shaping sustained engagement, with trust acting as a key mechanism through which performance perceptions translate into continued use. By demonstrating a partial mediation effect of trust, the study highlights the importance of moving beyond accuracy-focused evaluations toward a more comprehensive understanding of user reliance in mHealth. Users appear to base their long-term engagement decisions not only on perceptions of technical performance but also on whether the app is perceived as trustworthy and dependable in managing personal health information. These findings have practical implications for the design and deployment of symptom monitoring applications. Efforts to improve sustained use should address both how accurately apps perform and how their outputs, limitations, and uncertainties are communicated to users in ways that support appropriate trust. Focusing on trust-aware design and deployment strategies may inform the design of symptom monitoring applications and their evaluation in user-centered digital health contexts.

Supplemental material

Supplemental material - Trust, perceived accuracy, and continued use of patient-facing symptom monitoring applications: Cross-sectional study

Supplemental material for Trust, perceived accuracy, and continued use of patient-facing symptom monitoring applications: Cross-sectional study by Qing Han, and Chenyang Zhao in Digital Health.

Supplemental material

Supplemental material - Trust, perceived accuracy, and continued use of patient-facing symptom monitoring applications: Cross-sectional study

Supplemental material for Trust, perceived accuracy, and continued use of patient-facing symptom monitoring applications: Cross-sectional study by Qing Han, and Chenyang Zhao in Digital Health.

Footnotes

Acknowledgments

The authors would like to sincerely thank all participants who generously contributed their time and perspectives to this study. Their willingness to share their experiences and views on symptom monitoring applications made this research possible. We are grateful for their participation and cooperation, which were essential to the successful completion of this work.

ORCID iD

Qing Han

Ethical considerations

The study protocol was reviewed and approved by the Institutional Review Board (IRB) of Zhejiang Chinese Medical University (Application Number: 20250926-3). The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki and relevant national regulations.

Consent to participate

All participants were informed about the purpose of the study, the voluntary nature of participation, and the anonymity of their responses. Electronic informed consent was obtained from all participants prior to survey completion. No personally identifiable information was collected, and all data were securely stored.

Author contributions

Q.H.: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. C.Z.: Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by National Social Science Fund of China (ID: 25FTQB009).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data and analysis code are available upon reasonable request. Any other identifying information related to the authors and/or their institutions, funders, approval committees, etc, that might compromise anonymity.*

Supplemental material

Supplemental material for this article is available online.

References

Krebs

Duncan

. Health App Use Among US Mobile Phone Owners: A National Survey. JMIR Mhealth Uhealth 2015; 3: e101. https://doi.org/10.2196/mhealth.4924

Eysenbach

. The Law of Attrition. J Med Internet Res 2005; 7: e11. https://doi.org/10.2196/jmir.7.1.e11

Baumel

Muench

Edan

, et al. Objective User Engagement With Mental Health Apps: Systematic Search and Panel-Based Usage Analysis. J Med Internet Res 2019; 21: e14567. https://doi.org/10.2196/14567

Pratap

Neto

Snyder

, et al. Indicators of retention in remote digital health studies: a cross-study evaluation of 100,000 participants. NPJ Digit Med 2020; 3: 21. https://doi.org/10.1038/s41746-020-0224-8

Chambers

Cantrell

Johnson

, et al. Digital and online symptom checkers and health assessment/triage services for urgent health problems: systematic review. BMJ Open 2019; 9: e027743. https://doi.org/10.1136/bmjopen-2018-027743

Semigran

Linder

Gidengil

, et al. Evaluation of symptom checkers for self diagnosis and triage: audit study. BMJ 2015; 351: h3480. https://doi.org/10.1136/bmj.h3480

Wallace

Chan

Chidambaram

, et al. The diagnostic and triage accuracy of digital and online symptom checker tools: a systematic review. NPJ Digit Med 2022; 5: 118. https://doi.org/10.1038/s41746-022-00667-w

Gilbert

Mehl

Baluch

, et al. How accurate are digital symptom assessment apps for suggesting conditions and urgency advice? A clinical vignettes comparison to GPs. BMJ Open 2020; 10: e040269. https://doi.org/10.1136/bmjopen-2020-040269

Sagona

Dai

Macis

, et al. Trust in AI-assisted health systems and AI’s trust in humans. npj Health Syst. 2025; 2, 10. https://doi.org/10.1038/s44401-025-00016-5

10.

Lee

See

. Trust in automation: designing for appropriate reliance. Hum Factors 2004; 46: 50–80. https://doi.org/10.1518/hfes.46.1.50_30392

11.

Hoff

Bashir

. Trust in automation: integrating empirical evidence on factors that influence trust. Hum Factors 2015; 57: 407–434. https://doi.org/10.1177/0018720814547570

12.

Catapan

Sazon

Zheng

, et al. A systematic review of consumers' and healthcare professionals' trust in digital healthcare. NPJ Digit Med 2025; 8: 115. https://doi.org/10.1038/s41746-025-01510-8

13.

Tejera

Gokhale

Gromisch

. Trustworthiness Features of Commercially Available Mobile Health (mHealth) Applications. In: 2025 IEEE 49th Annual Computers, Software, and Applications Conference (COMPSAC). : IEEE, 2025, pp. 1928–1933.

14.

van Haasteren

Vayena

Powell

. The Mobile Health App Trustworthiness Checklist: Usability Assessment. JMIR Mhealth Uhealth 2020; 8: e16844. https://doi.org/10.2196/16844

15.

Bhattacherjee

. Understanding Information Systems Continuance: An Expectation-Confirmation Model1. MIS Quarterly 2001; 25: 351–370. https://doi.org/10.2307/3250921

16.

Vaghefi

Tulu

. The Continued Use of Mobile Health Apps: Insights From a Longitudinal Study. JMIR Mhealth Uhealth 2019; 7: e12983. https://doi.org/10.2196/12983

17.

Nadal

Sas

Doherty

. Technology Acceptance in Mobile Health: Scoping Review of Definitions, Models, and Measurement. J Med Internet Res 2020; 22: e17256. https://doi.org/10.2196/17256

18.

Adnan

Irvine

Williams

, et al. Improving Acceptability of mHealth Apps-The Use of the Technology Acceptance Model to Assess the Acceptability of mHealth Apps: Systematic Review. J Med Internet Res 2025; 27: e66432. https://doi.org/10.2196/66432

19.

Ryan

Hogg

Kasun

, et al. Users' Perceptions and Trust in AI in Direct-to-Consumer mHealth: Qualitative Interview Study. JMIR Mhealth Uhealth 2025; 13: e64715. https://doi.org/10.2196/64715

20.

Shevtsova

Ahmed

Boot

IWA

, et al. Trust in and Acceptance of Artificial Intelligence Applications in Medicine: Mixed Methods Study. JMIR Hum Factors 2024; 11: e47031. https://doi.org/10.2196/47031

21.

Mohd Johari

Mohamad Ali

Mhd Salim

, et al. Factors driving the use of mobile health app: insights from a survey. Mhealth 2025; 11: 12. https://doi.org/10.21037/mhealth-24-44

22.

Honglin

Jianghua

Hui

. Quality factors affecting the continued use of mobile health apps in ethnic minority regions of Southwest China using PLS-SEM and ANN. Sci Rep 2024; 14: 25469. https://doi.org/10.1038/s41598-024-75410-4

23.

Stocchi

Michaelidou

Pourazad

, et al. Psychological drivers of the intention to continue using health apps amongst young users. European Journal of Marketing 2025; 59: 452–487. https://doi.org/10.1108/ejm-07-2024-0580

24.

Mirabootalebi

Holl

Meidani

, et al. Determinants of Nurses' Continuance Intention to Use Mobile Health Apps in Clinical Nursing Practice: Structural Equation Modeling to Extend the Expectation-Confirmation Model. JMIR nursing 2025; 8: e68048. https://doi.org/10.2196/68048

25.

Chen

Wang

Sun

. Exploring factors in AI-based healthcare services adoption: integrating cognitive, social and technical perspectives. Asia Pacific Journal of Marketing and Logistics 2025; 1–21. https://doi.org/10.1108/apjml-05-2025-0834

26.

Zhou

Yang

Tang

, et al. The mediation of trust on artificial intelligence anxiety and continuous adoption of artificial intelligence technology among primacy nurses: a cross-sectional study. BMC Nurs 2025; 24: 724. https://doi.org/10.1186/s12912-025-03406-0

27.

Wetzel

A-J

Klemmt

Müller

, et al. Only the anxious ones? Identifying characteristics of symptom checker app users: a cross-sectional survey. BMC Med Inform Decis Mak 2024; 24: 21. https://doi.org/10.1186/s12911-024-02430-5

28.

Müller

Klemmt

Koch

, et al. “That's just Future Medicine” - a qualitative study on users' experiences of symptom checker apps. BMC Med Ethics 2024; 25: 17. https://doi.org/10.1186/s12910-024-01011-5

29.

Green

. How Many Subjects Does It Take To Do A Regression Analysis. Multivariate Behavioral Research 1991; 26: 499–510. https://doi.org/10.1207/s15327906mbr2603_7

30.

Fritz

Mackinnon

. Required sample size to detect the mediated effect. Psychol Sci 2007; 18: 233–239. https://doi.org/10.1111/j.1467-9280.2007.01882.x

31.

Tam

Santos

Oliveira

. Exploring the influential factors of continuance intention to use mobile Apps: Extending the expectation confirmation model. Inf Syst Front 2020; 22: 243–257. https://doi.org/10.1007/s10796-018-9864-5

32.

The DeLone and McLean Model of Information Systems Success: A Ten-Year Update. Journal of Management Information Systems 2003; 19: 9–30.

33.

Tavakol

Dennick

. Making sense of Cronbach's alpha. Int J Med Educ 2011; 2: 53–55. https://doi.org/10.5116/ijme.4dfb.8dfd

34.

Stevens

Stetson

. Theory of trust and acceptance of artificial intelligence technology (TrAAIT): An instrument to assess clinician trust and acceptance of artificial intelligence. J Biomed Inform 2023; 148: 104550. https://doi.org/10.1016/j.jbi.2023.104550

35.

Dzindolet

Peterson

Pomranky

, et al. The role of trust in automation reliance. International Journal of Human-Computer Studies 2003; 58: 697–718. https://doi.org/10.1016/s1071-5819(03)00038-7

36.

Jacobs

Pradier

McCoy

, et al. How machine-learning recommendations influence clinician treatment selections: the example of the antidepressant selection. Transl Psychiatry 2021; 11: 108. https://doi.org/10.1038/s41398-021-01224-x

37.

Sawrikar

Mote

. Technology acceptance and trust: Overlooked considerations in young people's use of digital mental health interventions. Health Policy and Technology 2022; 11: 100686. https://doi.org/10.1016/j.hlpt.2022.100686

38.

Nie

Oldenburg

Cao

, et al. Continuous usage intention of mobile health services: model construction and validation. BMC Health Serv Res 2023; 23: 442. https://doi.org/10.1186/s12913-023-09393-9

39.

Afroogh

Akbari

Malone

, et al. Trust in AI: progress, challenges, and future directions. Humanit Soc Sci Commun 2024; 11: 1568. https://doi.org/10.1057/s41599-024-04044-8

40.

Huang

, et al. Developing trustworthy artificial intelligence: insights from research on interpersonal, human-automation, and human-AI trust. Front Psychol 2024; 15: 1382693. https://doi.org/10.3389/fpsyg.2024.1382693

41.

Schmieding

Kopka

Bolanaki

, et al. Impact of a Symptom Checker App on Patient-Physician Interaction Among Self-Referred Walk-In Patients in the Emergency Department: Multicenter, Parallel-Group, Randomized, Controlled Trial. J Med Internet Res 2025; 27: e64028. https://doi.org/10.2196/64028

42.

Cho

Olaye

Shandhi

MMH

, et al. Identification of key factors related to digital health observational study adherence and retention by data-driven approaches: an exploratory secondary analysis of two prospective longitudinal studies. Lancet Digit Health 2025; 7: e23–e34. https://doi.org/10.1016/S2589-7500(24)00219-X

43.

Shojaei

Vlahu-Gjorgievska

Chow

Y-W

. Enhancing privacy in mHealth Applications: A User-Centric model identifying key factors influencing Privacy-Related behaviours. Int J Med Inform 2025; 199: 105907. https://doi.org/10.1016/j.ijmedinf.2025.105907

44.

Alkhariji

Alhirabi

Alraja

, et al. Synthesising Privacy by Design Knowledge Toward Explainable Internet of Things Application Designing in Healthcare. ACM Trans. Multimedia Comput. Commun. Appl 2021; 17: 1–29. https://doi.org/10.1145/3434186

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.29 MB

0.27 MB