Sage Journals: Discover world-class research

Abstract

Introduction

The glucose management indicator (GMI) and time-in-range (TIR) are important glycemic parameters calculated from continuous glucose monitoring (CGM) values. KARAZ, a hybrid Internet of things—artificial intelligence platform, was developed in Saudi Arabia to help manage diabetes mellitus. The complex nature of self-care and behavior changes associated with diabetes mellitus requires breaking large behaviors into achievable ones and related incentives.

Aim

This study explored how tiny habits as a behavioral intervention and incentive system affect glycemic control among KARAZ Platform users with diabetes mellitus in various age groups in Saudi Arabia.

Methods

This retrospective study included KARAZ Platform users and analyzed the effects of behavioral interventions and incentives on GMI and TIR as glycemic control parameters.

Results

Of 296 active users, 118 (40%) and 148 (50%) maintained a desirable TIR and GMI, respectively. Adult females aged ≥ 26 years who consistently followed tiny habits and behavior changes exhibited a significant reduction in the GMI (5%). Intrinsic motivation through behavioral modification was more effective than external incentives for maintaining glucose control.

Conclusion

The findings highlight how behavioral interventions can impact GMI, suggesting their effectiveness in promoting better health behaviors and improving glycemic control in the Saudi Arabian context. Further research should investigate how these habits and behaviors can be maintained sustainably without relying on external incentives. Recommendations discussed how children with Type 1 diabetes mellitus would benefit from CGM connection to KARAZ Platform iteration and the integration of a comprehensive diabetes care program within the Saudi health system.

Keywords

Glucose management indicator time-in-range behavioral interventions incentives hybrid IoT-artificial intelligence

Introduction

The glucose management indicator (GMI) is among the glycemic control parameters calculated from average continuous glucose monitoring (CGM) values, which measure glucose in the interstitial fluid every 1–5 min. The GMI provides perspective on how hemoglobin A1c (HbA1c) is trending but in a shorter time window because it incorporates time spent within the target glucose range.¹ A randomized clinical trial (RCT) on adults with Type 2 diabetes mellitus (T2DM) who underwent CGM reported a 16% increase in time-in-range (TIR) over an eight-month follow-up period.² In contrast, a trial concluded the effectiveness of CGM in reducing times with hypoglycemia among children with type 1 diabetes mellitus (T1DM) in aged between 2 and 8 years.³ Moreover, a clinical trial among older adults with T1DM who used CGM devices reported a significant increase in TIR by 8% over 52 weeks and a reduction in hypoglycemia.⁴ In 2019, the international consensus on TIR recommended three key CGM measurements with cutoff points for actionable glycemic control: percentage of readings and time per day within the target glucose range (TIR), time below the target glucose range (TBR), and time above the target glucose range (TAR). Variations in targets based on individual profiles have been suggested for glycemic control assessment.⁵

KARAZ can be identified as a hybrid Internet of Things (IoT)—artificial intelligence (AI) platform in Saudi Arabia, combining IoT for data collection from CGM devices and AI for data analysis and personalized health management. A recent study conducted in the USA reported that AI-assisted care targeting at risk individuals with DM can decrease DM-related hospital admissions by up to 19% by helping patients manage glucose levels more effectively.⁶ Moreover, IoT are claimed to provide scalable healthcare solutions, making CGM technology more accessible, particularly in underserved areas. Furthermore, previous studies have revealed that IoT-enabled healthcare services, including CGM, could reduce healthcare costs by 25% and improve remote patient care access by 50%, particularly in rural locations.^7–10

Regarding personalized management plans, an RCT in Japan concluded that IoT devices connected to glucose monitors and activity trackers facilitated real-time feedback to both users and providers.¹¹ Moreover, IoT systems improved HbA1c levels by integrating body weight, blood pressure, and activities and empowered individuals with T2DM to better manage their behavior and conditions over a 52-week follow-up period than the control group.¹² Furthermore, Internet-based devices provide interactive communication with automated data upload and have been reported to significantly decrease HbA1c levels from 7.86 to 7.55 in the intervention group compared with 7.81–7.70 in the control group.¹¹

Behavioral modifications have been applied in various contexts, including diabetes management, to encourage the adoption of healthier behaviors. However, the “tiny habits” framework has shown promise in increasing gratitude and psychological well-being.¹³ The tiny habits framework suggests that behavioral modifications can be facilitated by breaking down desired behaviors into small, manageable actions, attaching these actions to existing routines in life and celebrating successes.^14,15 However, its application in diabetes control remains limited. The information—motivational—behavioral (IMB) skills model, which emphasizes information, motivation, and behavioral skills, has been validated as a comprehensive framework for diabetes self-care.¹⁶ Community-based diabetes management, which incorporates “tiny habits” principles, has been shown to be effective when patients are knowledgeable and self-efficacious.¹⁷ However, various studies have also indicated that sustainable behavior changes in diabetes management are complex and influenced by various factors, including social support, education, and personal motivation.^16,18,19 Literature suggests that a multifaceted approach, such as the IMB model, is more appropriate for addressing the complex nature of diabetes self-care behaviors.^{13,16,17,20,21}

In addition, to a previous study on the technology acceptance model among patients with T1DM within Saudi arabia,²² we could not identify any publications addressing the impact of IoT on glycemic control within the country. The significance of these studies is providing evidence for the effects of behavioral modifications and incentives on both the GMI and TIR, which can inform the development of more personalized interventions to improve glycemic control.²³

This study evaluated the impact of the “tiny habits” framework as a behavioral change framework on glycemic control, including TIR and GMI. Furthermore, this study analyzed the effects of the KARAZ incentive system on GMI, considering diverse age groups, sexes, and glucose impairment types among KARAZ users.

Methodology

This retrospective analytic study evaluated the effects of KARAZ Platform use on glycemic management and associated behavioral factors.

Participants

The study included active KARAZ Platform users who met the following inclusion criteria: used either device-based CGM or manual blood glucose monitoring systems associated with the KARAZ Platform for at least 90 days, demonstrated familiarity with the KARAZ Platform's self-management tools, and benefited from the KARAZ “Cherries” virtual and badge incentive system. The exclusion criteria were as follows: pregnant women with DM and individuals who reported comorbid conditions that could interfere with DM management or their ability to use the KARAZ Platform effectively.

Statistical analysis

All statistical analyses were performed using Statistical Package for the Social Sciences (version 16). Categorical variables are presented as frequencies and percentages. Continuous variables with nonnormal distribution are presented as medians and interquartile ranges (IQRs). The chi-square test was used to evaluate significant relationships between dependent and outcome variables. p-values < .05 were used to denote statistical significance. Correlation analysis was performed to explore the relationship between selected tiny habits, namely, “hits” and “misses,” and TIR. Furthermore, the relationship between the KARAZ incentive system, namely, “badges” and “cherries,” and TIR was explored. A regression model was applied to control for confounding factors and identify significant predictors of GMI. These predictors included “hits” and “misses” and “badges” and “cherries.” To ensure regression model stability, a current sample size of 296 observations was included, and three of the identified significant predictors within the study objective were included in the model. “Badges” and sleeping were excluded from the model because they exhibited low reporting values. Moreover, only a small proportion (3.7%) of users benefited from “badges.”

The GMI was calculated using the following equation¹:

GMI (%) = 3.31 + 0.02392 \times [\frac{mean glucose in mg}{dL}] .

Standardized categories and definitions of glycemic indices were adopted from the Advanced Technologies & Treatments for Diabetes Congress International Consensus statement in 2019. The TIR metric is available using CGM and is generally defined as the percentage of time glucose is within a predetermined target range. TIR may provide a more accurate assessment of blood glucose stability and hypoglycemia.^1,24 For most adults with DM, the desirable target GMI value is an HbA1c of <7%, which corresponds to a mean glucose level of approximately 154 mg/dL. Moreover, for the recommended glucose target range of 70–180 mg/dL, the identified percentage value of TIR was >70%.²⁵ These values are widely recognized as indicators of effective glucose management and are aligned with the standard HbA1c targets designed to reduce the risk of DM-related complications.^1,25–27 Notably, discrepancy between GMI and HbA1c in approximately 22% of individuals at ≥1% can be observed.^28,29 The TBR and TAR were discussed in another publication.

Data cleaning procedures included removing duplicates and anonymizing participant information by the KARAZ Platform's administrator. The 90-day platform use was selected to ensure the reliability of the GMI and TIR values. Variables were extracted using a custom-designed standardized Excel sheet and managed based on the following operational definitions: age reported in years; sex categorized as male, female, or not reported; KARAZ incentive system (“cherries” and “badges”) incentivizes healthy behaviors by awarding virtual “Cherries” for actions such as glucose monitoring, meal logging, and attending consultations. These points can be redeemed for tangible awards, such as discounts on healthy products, services, and well-being programs. Tiny habits were behavioral factors reported by the participants and included sleep duration (hours per day), insulin use adherence, and average number of daily “hits” (successful behaviors) and “misses” (unachieved targets).

Ethical considerations

Ensuring anonymization and maintaining privacy, KARAZ complies with Saudi government legal requirements for the following: data protection, data accuracy, and protecting users from harm. It complies with the regulations set by the Saudi Food and Drug Association (SFDA) and Ministry of Health for telemedicine and remote monitoring.³⁰ KARAZ Platform management ensures that all devices, such as CGM systems, meet SFDA standards. Furthermore, KARAZ is pursuing international certifications (ISO 14971:2019—MEDICAL DEVICES APPLICATION OF RISK MANAGEMENT TO MEDICAL DEVICES, ISO / IEC 27001:2013 Information Security Management Systems, ISO 13485:2016 Medical Devices—Quality Management System) and ISO 27001 for information security, ensuring quality management and data protection. Users of the platform obligated to consent approval on data collection and use for research purposes according to the KARAZ privacy policy https://karaz.app/privacy-policy/.

Ethical approval was obtained from the Institutional Review Board according to the following application number: UJ-REC-262 and under the following title: Impact of the KARAZ platform's behavioral interventions and incentives on diabetic glycemic control in Saudi Arabia. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement checklist was used to ensure the compliance of quality of reporting combined cohort and cross-sectional studies.

Results

The study analyzed data from 296 participants residing in Saudi Arabia; these participants met the inclusion criteria. Most participants 218 (73.6%) were identified as having lifestyle glucose impairment or T2DM, whereas less than one-fifth were identified as having T1DM. Table 1 presents the detailed demographic and clinical profiles of the participants.

Table 1.

Profile of KARAZ beneficiaries (n = 296).

Variable	Values
Age
≤4	8 (2.7%)
5–20	58 (19.6)
21–26	63 (21.3)
>26	54 (18.2%)
Not reported	113 (38.2%)
Sex
Male	55 (18.6%)
Female	159 (53.7%)
Not reported	82 (27.2%)
Number of continuous glucose monitoring days	Median (90_IQR = 0)
Glucose impairment type identified by KARAZ users
T1DM	52 (17.6%)
Lifestyle glucose impairment/T2DM	218 (73.6%)
Not reported/not sure	26 (8.8%)
Methods of glucose monitoring
Manual-BGM	15 (5.1%)
Device based-CGM	281 (94.9%)

Glycemic control indicators

Regarding glycemic control, approximately half of the participants 148 (50%) achieved the desired GMI goal of <7%. For TIR, 118 (40%) participants maintained glucose levels within the target range of 70–180 mg/dL (Table 2).

Table 2.

Description of glycemic control indicators for KARAZ beneficiaries (n = 296).

Variable	Values
Average glucose reading per day
Desirable <180 mg/dl	15 (5.1%)
Not desirable ≥180 mg/dl	281 (94.9%)
Glucose management indicator
Desirable goal <7%	148 (50%)
Poor control ≥7%	148 (50%)
Targeted value of TIR
≤70%	178 (60.1%)
>70%	118 (39.9%)
Time in range	Median (63.4, IQR = 29)

Behavioral patterns and incentives

Only 59 (20%) participants reported their average daily sleep duration, of whom 47 (15.9%) reported sleeping at least 8 h per day. Insulin use was reported by 40 (13.5%) participants, with usage being highest among individuals with T1DM, particularly in the 4-year-old and 21–26-year-old age groups. The differences in insulin use across other glucose impairment types and age groups were statistically significant (Table 3). The participants reported a median of 1.73 successful behavior “hits” per day (IQR = 99.8) and a median of 99.6 behavior “misses” per day (IQR = 79.6). The mean number of “cherries” earned per day was 5.1 ± 2.31, and only 11 (3.7%) participants benefited from earning badges. The highest number of badges earned was 39, whereas the lowest was only one.

Table 3.

The pattern of insulin use among various age group, gender and glucose impairment type (n = 296).

Variables	Reported Insulin user (n = 40)	Not reported insulin user (n = 256)	p-value
Age group			.002^a
≤4	3 (37.5%)	5 (62.5%)
5–20	9 (15.5%)	49 (84.5%)
21–26	16 (25.4%)	47 (74.6%)
>26	3 (5.6%)	51 (94.4%)
Not reported	9 (8%)	104 (92%)
Gender			.071
Male	6 (10.9%)	49 (89.1%)
Female	28 (17.6%)	131 (82.4%)
Not reported	6 (7.3%)	76 (92.7%)
Glucose impairment type			.014^a
T1DM	13 (25%)	39 (75%)
Lifestyle glucose impairment/T2DM	26 (11.9%)	192 (88.1%)
Not reported/not sure	1 (3.8%)	25 (96.2%)

Indicating significant difference in chi-square testing.

Age, sex, and glucose impairment patterns

The highest proportion of individuals with a desirable GMI (<7%) was observed among participants aged >26 years, with the differences between age groups being statistically significant (p < .018). Females exhibited a significantly higher proportion of individuals with a desirable GMI than males (p < .016), as shown in (Table 4).

Table 4.

The association between glucose management indicator value across age groups and gender.

Variables	Glucose management indicator (desirable control < 7%)	Glucose management indicator (poor control ≥ 7%)	p-value
Age group			.018^a
≤4	1 (12.5%)	7 (87.5%)
5–20	29 (50%)	29 (50%)
21–26	37 (58.7%)	26 (41.3%)
>26	33 (61.1%)	21 (38.9%)
Not reported	48 (42.5%)	65 (57.5%)
Gender			.016^a
Male	29 (52.7%)	26 (47.3%)
Female	89 (56%)	70 (44%)
Not reported	30 (36.6%)	52 (63.4%)

Indicating significant difference in chi-square testing.

The highest TIR (≥70% of the time) was reported among participants aged 21–26 years, followed by those aged >26 years, with differences across age groups being statistically significant (p < .004). TIR was equally high among males and females (p < .01), indicating good glycemic control compared with participants who did not report their sex. Although TIR was highest among individuals with T1DM, differences across glucose impairment groups were not statistically significant (Table 5).

Table 5.

The association between TIR value, age groups, gender and glucose impairment type.

Variables	TIR (cutoff) Target >70%	TIR (cutoff) ≤70%	p-value
Age group			.004^a
≤4	1 (12.5%)	7 (87.5%)
5–20	20 (34.5%)	38 (65.5%)
21–26	34 (54%)	29 (46%)
>26	28 (51.9%)	26 (48.1%)
Not reported	35 (31.9%)	78 (68.1%)
Gender			.01^a
Male	25 (45.5%)	30 (54.5%)
Female	72 (45.3%)	87 (54.7%)
Not reported	21 (25.7%)	61 (74.3%)
Glucose impairment type			.62
T1DM	22 (43.4%)	31 (56.6%)
Lifestyle glucose impairment/T2DM	88 (40.6%)	129 (59.4%)
Not reported/not sure	8 (30.8%)	18 (69.2%)

TIR: time-in-range; T1DM: Type 1 diabetes mellitus; T2DM: Type 2 diabetes mellitus.

Indicating significant difference in chi-square testing

Association between behaviors, incentives, and glycemic control

A Spearman's rank-order correlation analysis was performed to evaluate the relationship between TIR and behavioral/incentive metrics. A strong and statistically significant positive correlation was observed between TIR and both average daily “hits” (rs = 0.86, n = 296, p < .001) and total “badges” earned (rs = 0.54, n = 296, p < .001). Furthermore, TIR exhibited a strong positive correlation with the total number of “cherries” earned (rs = 0.91, n = 296, p < .001). These findings suggest that higher TIR values are associated with increased adherence to desired behaviors and greater engagement with the incentive system.

Regression model's assumptions and findings

The key assumptions were tested before applying the regression model. The residual plots exhibited homoscedasticity, confirming the assumption of constant variance. However, the assumption of linearity was violated based on the scatterplots of residuals against predictors. To address this issue, logistic regression analysis was performed on the binomial outcome GMI. Residual diagnostics for linearity revealed that the residuals for “cherries” were normally distributed, supported by a straight Q-Q plot and the Shapiro–Wilk test (p = .086). However, residuals for “misses” and “hits” deviated from normality. All predictors (“hits,” “misses,” and “cherries”) exhibited relatively low variance inflation factor values (3.6, 1.9, and 5, respectively), indicating no multicollinearity. The results indicate that each additional unit increase in the average “hits” per day was associated with a significant decrease in GMI by 5% (p < .001). In contrast, each additional unit increase in the average “misses” per day is associated with a significant increase in GMI by 6% (p < .001). However, the total number of “cherries” earned was not a significant predictor of GMI (Table 6).

Table 6.

Logistic regression identifying tiny habits and incentive related effect size on glucose management indicator.

Predictor	Exp (B)	p-Value
Average hits per day	0.95	.001^a
Average misses per day	1.06	.001^a
Average number of cherries earned	1.00	.75
Constant	0.75	.72

Indicating significant association.

Discussion

The largest proportion of KARAZ users in Saudi Arabia were classified as having lifestyle glucose impairment or T2DM 206 (73.3%).

Principal findings and comparison with previous studies

Of the 296 participants, 148 (50%) maintained the GMI goal of <7% over the 90-day use of the KARAZ Platform. The reported desirable GMI in this study was found to be highest among females and individuals aged >26 years. Considering that GMI offers insight into the trend of HbA1c, previous studies have considered it a surrogate for the targeted control value.^27,28 Our findings exhibited better glucose control over a short period among females aged >26 years. This is potentially due to more established and consistent management practices among this group. The findings are consistent with broader literature indicating that age and sex can play critical roles in DM management, with older adults and females often demonstrating greater interest in enrolling in chronic disease prevention programs.³¹ Females also have more stable glucose control, possibly due to higher insulin sensitivity and the protective effects of estrogen and receptor activation on adipose tissue and pancreatic beta cells.^32,33 Moreover, self-care behaviors, quality of life, and treatment adherence were identified as influencing factors.^34,35

This study exhibited disparities between lifestyle-related glucose impairments and/or T2DM and other glucose impairment categories, including T1DM. The findings highlight the need for targeted interventions and lifestyle modifications tailored specifically for individuals with T2DM. This aligns with that of previous research, which indicated that patients with T2DM often struggle with effective glycemic management due to insulin resistance and other metabolic factors, namely, beta cell dysfunction and obesity.^36,37 Furthermore, discrepancy between TIR and GMI was observed. This could be due to variations in the methods of glucose monitoring, device calibration, measurement duration, and high variability in glucose values.^1,38

The analysis revealed that the highest reported insulin use was among individuals with T1DM at the age of 4 years, comprising less than one-fifth of the sample. As shown in Table 3, despite the significant difference in insulin use observed across age groups and among glucose impairment patterns, the underrepresentation of T1DM within the study group made it challenging to conclude the direct impact of KAZAR Platform use on glycemic control compared with T2DM. The issue could be resolved using a randomized controlled design focused on children aged 2–8 years diagnosed with T1DM to assess TIR, GMI, and other glycemic control variables and evaluate the effectiveness of CGM devices integrated within the KARAZ Platform.

Effects of behavioral changes and incentive systems on TIR and GMI values

Many studies have considered TIR as a surrogate marker of long-term cardiovascular complications among individuals with DM.³⁹ Moreover, a review on TIR-driven CGM values identified an inverse relationship between TIR and diabetic nephropathy, diabetic peripheral neuropathy, and cardiac autonomic neuropathy.⁴⁰ The existing finding on the correlations of TIR with average “hits” per day and total number of “cherries” and “badges” earned by KARAZ users is statistically significant, suggesting a clear association between improved TIR and increased engagement in behaviors beneficial for glucose control among 40% of the study group. The application of logistic regression allowed controlling the confounding factors. Despite the current regression model limitations, the findings highlight how behavioral interventions, such as average “hits” per day habits and “misses,” can significantly impact GMI, suggesting their effectiveness in promoting better health behaviors and improving GMI in the Saudi Arabian context.

Our findings are consistent with those of previous studies, which concluded that regular physical activities have positive effects on insulin sensitivity and glucose control. Likewise, studies on habit formation have highlighted that developing behaviors such as consistent self-monitoring and meal tracking can improve long-term glycemic outcomes, often exceeding the benefits of external incentives.^41–43 Regarding incentives, the Trial to Incentivize Adherence for Diabetes study conducted in Singapore highlighted the benefits of process-based financial incentives for individuals with T2DM. By focusing on specific behaviors, such as regular glucose monitoring and physical activity, these incentives proved to be more effective at improving HbA1c levels than outcome-based incentives.⁴⁴

In contrast, the total number of “cherries” earned as incentives was not a significant predictor of glucose regulation. This finding supports contradicting results regarding the effectiveness of incentives in lifestyle modification programs, as one RCT study concluded a significant reduction in HbA1c levels among incentivized adults with DM.⁴⁵ However, the effectiveness of incentives varied depending on cultural values and the level of acceptance.⁴⁶

Finally, the impact of the KARAZ Platform after measuring HbA1c levels with active involvement of physicians and/or providers is worth evaluating considering the extended patient follow-up period of at least 52 weeks. This approach can help prevent long-term disease complications. For children aged 2–8 years diagnosed with T1DM, future evaluations should consider using a randomized controlled design. The aim would be to assess TIR, GMI, and other glycemic control variables and to evaluate the effectiveness of CGM devices integrated with the KARAZ Platform.

Limitations

This study has some limitations to consider. Primarily, there may be variability in participant engagement with the digital platform and its associated incentive system. Such variability may lead to inconsistent data regarding the not reported predictors of insulin resistance, such as BMI, mental health status, and dietary factors. Furthermore, the variable's reliance on self-reported data for specific behaviors, such as reported steps per day, exercise, reported insulin use, and sleeping time, may introduce systematic biases. Variations in the participants' familiarity and comfort with technology may hinder their ability to fully engage with the resources available, ultimately affecting the overall efficacy of the incentive system intervention. Furthermore, disparities in digital literacy among the participants could impact the effective use of both the incentive system and the digital tools provided. The regression model's limitations include the potential impact of nonnormal residuals for “misses” and “hits” as predictors. Furthermore, important predictors, namely, “badges” and “sleeping,” were excluded due to low reporting values, which may have introduced omitted variable bias. Regression findings should be considered with caution because they may limit the generalizability of the results. Finally, the exclusion of other components of the KARAZ Platform may restrict the model's ability to fully capture the incentive system's dynamics. Future studies with larger samples and additional predictors are recommended. Finally, the retrospective study design has limitations, such as the inability to establish causal relationships, limited control over data collection, and the presence of confounding factors. These limitations should be addressed in future iterations of IoT-based CGM platforms or applications, with a particular focus on enhancing the reporting of behavior change data, refining incentive systems, and strengthening the connection to glycemic control assessment methods. These improvements will contribute to a more reliable identification of causative relationships in future research.

Conclusion

Despite the listed limitations, our findings highlight how behavioral change reporting can significantly impact GMI, suggesting its effectiveness in promoting better health behaviors and improving glycemic control within the Saudi Arabian context and among KARAZ Platform users. Further longitudinal studies are recommended to investigate how these habits and behaviors can be maintained sustainably without relying on external incentives. Recommendations discussed how children with T1DM would benefit from CGM connection to the KARAZ Platform and system integration into a comprehensive diabetes care program within the Saudi health system. This research is novel and breaks new ground by exploring the influence of AI and IoT data on the connections between behavior changes and incentives and their impact on glycemic control.

Footnotes

Acknowledgments

The authors acknowledge the contribution of all study participants. Also, authors acknowledge the use of ChatGpt 4o in the writing and editing process of this manuscript. The tool provided valuable assistance in drafting, proofreading, and enhancing the overall readability of the text.

Consent for publication

This study utilized deidentified health data, ensuring that no personal identifiers were present and that individuals cannot be reidentified from the data. The publication of results is based on aggregated data and does not contain any information that could identify individual participants. All ethical standards and privacy of the Saudi national law have been followed. The Saudi Data and Artificial Intelligence Authority, under the Personal Data Protection Law, defines personal data as any information that can directly or indirectly identify an individual. This includes, but not limited to, specific identifiers such as: a person's name, National Identification Number (e.g., Iqama or Saudi National ID), contact details like email addresses, physical residential addresses or geographic location, IP addresses, and unique device identifiers. Additionally, during data extraction phase, KARAZ administration adhered to the General Data Protection Regulation guidelines on anonymization, particularly focusing on pseudonymization and data minimization principles. The anonymization process included the following steps:

Removal of identifiers: Personal identifiers such as names, addresses, contact information, and other direct identifiers were completely removed from the dataset.

Pseudonymization: Unique codes replaced sensitive attributes like patient ID numbers and replaces with user ID code, ensuring that no direct link between the code and the individual's identity.

Generalization: Certain data fields, such as age or location, were generalized to reduce details (e.g., age groups instead of exact age, or counntry as Saudi Arabia instead of specific addresses).

Suppression: Outlier data that could potentially lead to reidentification was either removed or suppressed. For example, outlier readings were excluded.

The possibility of reidentification through data combination has been minimized using the following strategies: (1) Removal of linkage variables: Any variables that could be linked to external datasets. (2) Risk reduction: No dataset released publicly or shared externally contains sufficient information to allow reidentification when combined with other datasets. In conclusion, the dataset has been anonymized using rigorous standards, and the likelihood of reidentification is negligible.

Consent to participate

Users of the platform obligated to consent approval on data collection and use for research purposes according to the KARAZ privacy policy ().

Declaration of conflicting interests

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

Ethical consideration

Ensuring anonymization and maintaining privacy, KARAZ complies with Saudi Government legal requirements for the following: data protection, data accuracy, and protecting users from harm. It complies with the regulations set by the Saudi Food and Drug Association (SFDA) and Ministry of Health for telemedicine and remote monitoring. KARAZ Platform management ensures that all devices, such as CGM systems, meet SFDA standards. Furthermore, KARAZ is pursuing international certifications (ISO 14971:2019—MEDICAL DEVICES APPLICATION OF RISK MANAGEMENT TO MEDICAL DEVICES, ISO / IEC 27001:2013 Information Security Management Systems, ISO 13485:2016 Medical Devices—Quality Management System) and ISO 27001 for information security, ensuring quality management and data protection. The STROBE statement checklist was used to ensure the compliance of quality of reporting combined cohort and cross-sectional studies.

Ethical approval

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Sulafa T Alqutub

Faisal Aljehani

References

Bergenstal

Beck

, et al. Glucose management indicator (GMI): a new term for estimating A1C from continuous glucose monitoring. Diabetes Care 2018; 41: 2275–2280.

Bailey

Calhoun

Chao

, et al. With or without residual C-peptide, patients with type 2 diabetes realize glycemic benefits from real-time continuous glucose monitoring. Diabetes Technol Therap 2022; 24: 281–284.

Laffel

Harrington

Hanono

, et al. A randomized clinical trial assessing continuous glucose monitoring (CGM) use with standardized education with or without a family behavioral intervention compared with fingerstick blood glucose monitoring in very young children with type 1 diabetes. Diabetes Care 2021; 44: 464–472.

Miller

Kanapka

Rickels

, et al. Benefit of continuous glucose monitoring in reducing hypoglycemia is sustained through 12 months of use among older adults with type 1 diabetes. Diabetes Technol Therap 2022; 24: 424–434.

Battelino

Danne

Bergenstal

, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care 2019; 42: 1593–1603.

Raldow

Raja

Villaflores

, et al. Proactive care management of AI-identified at-risk patients decreases preventable admissions. Am J Managed Care 2024; 30: 548–554.

WeForum. Scaling smart solutions with AI in health: unlocking impact on high-potential use cases, https://www3.weforum.org/docs/WEF_Scaling_Smart_Solutions_with_AI_in_Health_Unlocking_Impact_on_High_Potential_Use_Cases.pdf (2023).

Subhan

Mirza

Su’ud

MBM

, et al. AI-enabled wearable medical internet of things in healthcare system: a survey. Appl Sci 2023; 13: 1394.

Tiwari

Waoo

. Transforming healthcare: the synergistic fusion of AI and IoT for intelligent, personalized well-being. In: AI integration with IoT for enhanced patient outcomes 2024 Sep 24. Berlin: Springer Nature Switzerland, 2024, pp.109–149.

10.

Abdulmalek

Nasir

Jabbar

, et al. IoT-based healthcare-monitoring system towards improving quality of life: a review. Healthcare 2022; 10: 1993.

11.

Cho

Kim

H-S

Yoo

, et al. An internet-based health gateway device for interactive communication and automatic data uploading: clinical efficacy for type 2 diabetes in a multi-centre trial. J Telemed Telecare 2017; 23: 595–604.

12.

Bouchi

Izumi

Ishizuka

, et al. Internet of things-based approach for glycemic control in people with type 2 diabetes: a randomized controlled trial. J Diabetes Investig 2024; 15: 1287–1296.

13.

Osborn

Egede

. Validation of an information–motivation–behavioral skills model of diabetes self-care (IMB-DSC). Patient Educ Couns 2010; 79: 49–54.

14.

Hollingsworth

Redden

. Tiny Habits® for gratitude-implications for healthcare education stakeholders. Front Public Health 2022; 10. Epub ahead of print 16 May 2022. doi: https://doi.org/10.3389/fpubh.2022.866992

15.

FOGG B. Tiny Habits toolkit, https://tinyhabits.com/wp-content/uploads/2021/02/The-Official-Tiny-Habits-Toolkit-by-BJ-Fogg-Paperback-Edition.pdf.

16.

Peyrot

. Behavior change in diabetes education. Diabetes Educ 1999; 25: 62–73.

17.

Yao

Wang

Yin

, et al. Factors associated with the utilization of community-based diabetes management care: a cross-sectional study in Shandong Province, China. BMC Health Serv Res 2020; 20: 407.

18.

Jelínek

Fischer

. The role of emotional expression in behavior change coaching by a social robot. In: International conference on persuasive technology 2021 Apr 12. Berlin: Springer International Publishing, 2021, pp.193–199.

19.

Padarthy

Crespo

Rich

, et al. Using high-tech and high-touch methods for effective population health management. Iproceedings 2016; 2: e17.

20.

Ong

Stafford

McLaughlin

, et al. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the global burden of disease study 2021. Lancet 2023; 402: 203–234.

21.

Jeon

Park

H-A

. Development of the IMB model and an evidence-based diabetes self-management mobile application. Healthc Inform Res 2018; 24: 125.

22.

Almansour

Beloff

White

. Type 1 diabetes mellitus Saudi patients’ perspective on the adopting IoT-enabled CGM: validation of critical factors in the IAI-CGM a framework. In: 2023 18th conference on computer science and intelligence systems (FedCSIS), 2023 Sep 17, pp.73–81: IEEE.

23.

American Diabetes Association. 6. Glycemic targets: standards of medical care in diabetes—2021. Diabetes Care 2021; 44: S73–S84.

24.

Wright

Morgan

, et al. Time in range: how to measure it, how to report it, and its practical application in clinical decision-making. Clin Diabetes 2020; 38: 439–448.

25.

Beck

Bergenstal

Riddlesworth

, et al. Validation of time in range as an outcome measure for diabetes clinical trials. Diabetes Care 2019; 42: 400–405.

26.

ElSayed

Aleppo

Aroda

, et al. 6. Glycemic targets: standards of care in diabetes—2023. Diabetes Care 2023; 46: S97–S110.

27.

Leelarathna

Beck

Bergenstal

, et al. Glucose management indicator (GMI): insights and validation using guardian 3 and navigator 2 sensor data. Diabetes Care 2019; 42: e60–e61.

28.

Using the GMI to Estimate Your A1C: How accurate is it?, https://diatribe.org/diabetes-technology/using-gmi-estimate-your-a1c-how-accurate-it.

29.

American Diabetes Association. Standards of medical care in diabetes—2022 abridged for primary care providers. Clin Diabetes 2022; 40: 10–38.

30.

MOH. Legal regulations for telehealth services, https://www.moh.gov.sa/en/Ministry/Rules/Documents/Legal-Regulations-for-Telehealth-Services.pdf.

31.

Jackson

Dai

Skeete

, et al. An examination of gender differences in the National Diabetes Prevention Program’s Lifestyle Change Program. Diabetes Educ 2020; 46: 580–586.

32.

Peters

SAE

Muntner

Woodward

. Sex differences in the prevalence of, and trends in, cardiovascular risk factors, treatment, and control in the United States, 2001 to 2016. Circulation 2019; 139: 1025–1035.

33.

Tramunt

Smati

Grandgeorge

, et al. Sex differences in metabolic regulation and diabetes susceptibility. Diabetologia 2020; 63: 453–461.

34.

Misra

Lager

. Ethnic and gender differences in psychosocial factors, glycemic control, and quality of life among adult type 2 diabetic patients. J Diabetes Complications 2009; 23: 54–64.

35.

Kautzky-Willer

Harreiter

. Sex and gender differences in therapy of type 2 diabetes. Diabetes Res Clin Pract 2017; 131: 230–241.

36.

DeFronzo

Ferrannini

Groop

, et al. Type 2 diabetes mellitus. Nat Rev Dis Primers 2015; 1: 15019.

37.

Kahn

Hull

Utzschneider

. Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature 2006; 444: 840–846.

38.

Aleppo

. Clinical application of time in range and other metrics. Diabetes Spectr 2021; 34: 109–118.

39.

Wang

Shen

, et al. Time in range in relation to all-cause and cardiovascular mortality in patients with type 2 diabetes: a prospective cohort study. Diabetes Care 2021; 44: 549–555.

40.

Raj

Mishra

Jha

, et al. Time in range, as measured by continuous glucose monitor, as a predictor of microvascular complications in type 2 diabetes: a systematic review. BMJ Open Diabetes Res Care 2022; 10: e002573.

41.

Stone

Mayberry

Clouse

, et al. The role of habit formation and automaticity in diabetes self-management: current evidence and future applications. Curr Diabetes Rep 2023; 23: 43–58.

42.

Minari

Manzano

Tácito

LHB

, et al. The impact of a nutritional intervention on glycemic control and cardiovascular risk markers in type 2 diabetes. Nutrients 2024; 16: 1378.

43.

Małkowska

. Positive effects of physical activity on insulin signaling. Curr Issues Mol Biol 2024; 46: 5467–5487.

44.

Bilger

Shah

Tan

, et al. Process- and outcome-based financial incentives to improve self-management and glycemic control in people with type 2 diabetes in Singapore: a randomized controlled trial. Patient - Patient-Centered Outcomes Res 2021; 14: 555–567.

45.

Hulbert

Michael

Charter-Harris

, et al. Effectiveness of incentives for improving diabetes-related health indicators in chronic disease lifestyle modification programs: a systematic review and meta-analysis. Prev Chronic Dis 2022; 19: 220151.

46.

Bigsby

Seitz

Halpern

, et al. Estimating acceptability of financial health incentives. Health Educ Behav 2017; 44: 513–518.