Research article

Quality evaluation of digital voice assistants for diabetes management

  • Background 

    Digital voice assistants (DVAs) are increasingly used to search for health information. However, the quality of information provided by DVAs is not consistent across health conditions. From our knowledge, there have been no studies that evaluated the quality of DVAs in response to diabetes-related queries. The objective of this study was to evaluate the quality of DVAs in relation to queries on diabetes management.

    Materials and methods 

    Seventy-four questions were posed to smartphone (Apple Siri, Google Assistant, Samsung Bixby) and non-smartphone DVAs (Amazon Alexa, Sulli the Diabetes Guru, Google Nest Mini, Microsoft Cortana), and their responses were compared to that of Internet Google Search. Questions were categorized under diagnosis, screening, management, treatment and complications of diabetes, and the impacts of COVID-19 on diabetes. The DVAs were evaluated on their technical ability, user-friendliness, reliability, comprehensiveness and accuracy of their responses. Data was analyzed using the Kruskal-Wallis and Wilcoxon rank-sum tests. Intraclass correlation coefficient was used to report inter-rater reliability.

    Results 

    Google Assistant (n = 69/74, 93.2%), Siri and Nest Mini (n = 64/74, 86.5% each) had the highest proportions of successful and relevant responses, in contrast to Cortana (n = 23/74, 31.1%) and Sulli (n = 10/74, 13.5%), which had the lowest successful and relevant responses. Median total scores of the smartphone DVAs (Bixby 75.3%, Google Assistant 73.3%, Siri 72.0%) were comparable to that of Google Search (70.0%, p = 0.034), while median total scores of non-smartphone DVAs (Nest Mini 56.9%, Alexa 52.9%, Cortana 52.5% and Sulli the Diabetes Guru 48.6%) were significantly lower (p < 0.001). Non-smartphone DVAs had much lower median comprehensiveness (16.7% versus 100.0%, p < 0.001) and reliability scores (30.8% versus 61.5%, p < 0.001) compared to Google Search.

    Conclusions 

    Google Assistant, Siri and Bixby were the best-performing DVAs for answering diabetes-related queries. However, the lack of successful and relevant responses by Bixby may frustrate users, especially if they have COVID-19 related queries. All DVAs scored highly for user-friendliness, but can be improved in terms of accuracy, comprehensiveness and reliability. DVA designers are encouraged to consider features related to accuracy, comprehensiveness, reliability and user-friendliness when developing their products, so as to enhance the quality of DVAs for medical purposes, such as diabetes management.

    Citation: Joy Qi En Chia, Li Lian Wong, Kevin Yi-Lwern Yap. Quality evaluation of digital voice assistants for diabetes management[J]. AIMS Medical Science, 2023, 10(1): 80-106. doi: 10.3934/medsci.2023008

    Related Papers:

    [1] Francesca Valent, Valentina Moretti . A case-control study on cervical cancer screening outcome and HPV vaccination among young women in the Italian area of Udine. AIMS Medical Science, 2024, 11(3): 220-231. doi: 10.3934/medsci.2024017
    [2] Sylvia Kirchengast . Diabetes and Obesity—An Evolutionary Perspective. AIMS Medical Science, 2017, 4(1): 28-51. doi: 10.3934/medsci.2017.1.28
    [3] Ekaterina Kldiashvili, Sophio Bojgua . Liquid Based Cytology Cervical Cancer Screening Program—Georgian Experience. AIMS Medical Science, 2016, 3(3): 272-277. doi: 10.3934/medsci.2016.3.272
    [4] Nádia C. M. Okuyama, Fernando Cezar dos Santos, Kleber Paiva Trugilo, Karen Brajão de Oliveira . Involvement of CXCL12 Pathway in HPV-related Diseases. AIMS Medical Science, 2016, 3(4): 417-440. doi: 10.3934/medsci.2016.4.417
    [5] Anuj A. Shukla, Shreya Podder, Sana R. Chaudry, Bryan S. Benn, Jonathan S. Kurman . Non-small cell lung cancer: epidemiology, screening, diagnosis, and treatment. AIMS Medical Science, 2022, 9(2): 348-361. doi: 10.3934/medsci.2022016
    [6] Ahmed Alharbi, Waleed Mansi Alshammari, Turki A K Alreshidi . Expressional correlation of Toll-like Receptor 9 (TLR9) with angiogenic factors and anti-apoptotic markers in cervical cancer cells. AIMS Medical Science, 2021, 8(1): 11-22. doi: 10.3934/medsci.2021002
    [7] John V. Rider, Shannon Martin, Erin Vieira . Do older adults take action to reduce fall risk after attending a community-based fall risk screening?. AIMS Medical Science, 2024, 11(4): 464-479. doi: 10.3934/medsci.2024032
    [8] Maryam Tanhapour, Asad Vaisi-Raygani, Mozafar Khazaei, Zohreh Rahimi, Tayebeh Pourmotabbed . Cytotoxic T-lymphocyte Associated Antigen-4 (CTLA-4) Polymorphism, Cancer, and Autoimmune Diseases. AIMS Medical Science, 2017, 4(4): 395-412. doi: 10.3934/medsci.2017.4.395
    [9] Sherven Sharma, Pournima Kadam, Ram P Singh, Michael Davoodi, Maie St John, Jay M Lee . CCL21-DC tumor antigen vaccine augments anti-PD-1 therapy in lung cancer. AIMS Medical Science, 2021, 8(4): 269-275. doi: 10.3934/medsci.2021022
    [10] Douglas Wilson, Griffiths Keith, Buttar Harpal, Singh Ram, De Meester Fabien, Wilczynska Agnieszka, Takahashi Toru . Therapy through Social Medicine: Cultivating Connections and Inspiring Solutions for Healthy Living. AIMS Medical Science, 2017, 4(2): 131-150. doi: 10.3934/medsci.2017.2.131
  • Background 

    Digital voice assistants (DVAs) are increasingly used to search for health information. However, the quality of information provided by DVAs is not consistent across health conditions. From our knowledge, there have been no studies that evaluated the quality of DVAs in response to diabetes-related queries. The objective of this study was to evaluate the quality of DVAs in relation to queries on diabetes management.

    Materials and methods 

    Seventy-four questions were posed to smartphone (Apple Siri, Google Assistant, Samsung Bixby) and non-smartphone DVAs (Amazon Alexa, Sulli the Diabetes Guru, Google Nest Mini, Microsoft Cortana), and their responses were compared to that of Internet Google Search. Questions were categorized under diagnosis, screening, management, treatment and complications of diabetes, and the impacts of COVID-19 on diabetes. The DVAs were evaluated on their technical ability, user-friendliness, reliability, comprehensiveness and accuracy of their responses. Data was analyzed using the Kruskal-Wallis and Wilcoxon rank-sum tests. Intraclass correlation coefficient was used to report inter-rater reliability.

    Results 

    Google Assistant (n = 69/74, 93.2%), Siri and Nest Mini (n = 64/74, 86.5% each) had the highest proportions of successful and relevant responses, in contrast to Cortana (n = 23/74, 31.1%) and Sulli (n = 10/74, 13.5%), which had the lowest successful and relevant responses. Median total scores of the smartphone DVAs (Bixby 75.3%, Google Assistant 73.3%, Siri 72.0%) were comparable to that of Google Search (70.0%, p = 0.034), while median total scores of non-smartphone DVAs (Nest Mini 56.9%, Alexa 52.9%, Cortana 52.5% and Sulli the Diabetes Guru 48.6%) were significantly lower (p < 0.001). Non-smartphone DVAs had much lower median comprehensiveness (16.7% versus 100.0%, p < 0.001) and reliability scores (30.8% versus 61.5%, p < 0.001) compared to Google Search.

    Conclusions 

    Google Assistant, Siri and Bixby were the best-performing DVAs for answering diabetes-related queries. However, the lack of successful and relevant responses by Bixby may frustrate users, especially if they have COVID-19 related queries. All DVAs scored highly for user-friendliness, but can be improved in terms of accuracy, comprehensiveness and reliability. DVA designers are encouraged to consider features related to accuracy, comprehensiveness, reliability and user-friendliness when developing their products, so as to enhance the quality of DVAs for medical purposes, such as diabetes management.



    Cervical cancer (CC) stands as the fourth most frequently diagnosed cancer and the fourth cause of cancer-related mortality among women globally [1]. Alarmingly, approximately 90% of CC cases (mainly locally advanced and metastatic carcinoma of cervix) and premature deaths occur in low-and middle-income countries (LMICs) compared to high-income countries (HICs) [1],[2]. Moreover, Human Papilloma Virus (HPV) infection is the most common causative etiology for CCs in LMICs [3].

    Primary prevention, notably through HPV vaccination, and secondary prevention measures such as screening have proven effective in reducing CC incidence and mortality [2]. Nevertheless, their implementation has exhibited significant heterogeneity [2]. Indeed, HPV vaccination coverage remains limited in LMICs, with less than 30% of countries providing vaccination services [4]. Similarly, only 20% of women in LMICs have undergone prior screening examinations, in contrast to the over 60% observed in HICs [5].

    While most HICs boast well established CC screening programs, the healthcare systems in LMICs have to face several challenges, hindering their ability to offer effective and sustainable services due to deficiencies in technical, human, and financial resources [5]. However, implementation remains disjointed even in LMICs with active programs, and participation rates are low, influenced by several organizational and individual factors [6][8]. Noteworthy organizational factors include aspects such as program structure, modes of recruitment, accessibility, infrastructures, procedures for the return of results, and connections to further assessments and therapy. All these aspects pose particular challenges in LMICs due to shortages in population registries, resource constraints, and the scarcity of adequately trained healthcare professionals [6][8]. Additionally, individual determinants, such as CC awareness, and socio-cultural and religious barriers exert a substantial influence on screening participation [6][8].

    With respect to the CC burden, the World Health Organization (WHO) launched the “Global Strategy to Accelerate the Elimination of Cervical Cancer as a public health problem” in November 2020, outlining ambitious targets to be achieved by 2030 [9]. These goals include ensuring that 90% of girls are vaccinated by the age of 15 years, 70% of women are screened by 35 and again by 45 years, and 90% of women diagnosed with CC receive high-quality treatments [9]. The attainment of these targets is particularly challenging in LMICs [5].

    Since 2010, a multidisciplinary network (EuMedCN) was launched, which gathered experts in cancer control, national health authorities, and international organizations; this supported the implementation of effective and sustainable cancer screening programs across the Mediterranean region. This network holds potential significance in bolstering the WHO's ambitious strategy [10]. EuMedCN encompasses cancer experts from Mediterranean HICs, specialists and professionals from non-EU Balkan countries, and Northern African and Middle Eastern states bordering the Mediterranean Sea [10].

    Three countries (Albania, Montenegro, Morocco) belonging to the aforementioned network were selected for a trial with the primary objective of evaluating women's participation in CC screening using different recruitment strategies [11]. The secondary objectives include investigating the main barriers and facilitators for either attending or refusing CC screening invitations. Within the EuMedCN countries, these three areas were selected due to their availability to be involved in the trial, their robust CC epidemiological framework, the presence of ongoing pilot screening projects, and a steadfast political commitment to cancer prevention. Thus, Albania, Montenegro, and Morocco emerged as optimal environments to pilot and analyze the organizational complexities in inviting women for CC screenings. This paper presents the trial's results.

    This study, conducted during the pre-COVID-19 period (2017–2019), aimed to investigate CC screening practices in Albania, Montenegro, and Morocco, with a focus on comparing the “usual care” (ongoing invitation) to an “intervention strategy” (alternative approach) in each country. Notably, Montenegro had an established CC screening program, while Albania and Morocco only had pilot/local experiences. The diverse screening organization, settings, study periods, invitation modalities, tests, and study protocols across the countries are illustrated in Table 1. In particular, the specific target populations, varying across the three countries, were identified and mutually established with the respective country leaders. The decision to focus on a narrower target population, rather than adopting the broader one recommended by the International Guidelines, is attributed to both the experimental nature of the trial and the constrained economic, structural, and personnel resources.

    The identification of the more appropriate comparator with the standard invitation procedure was the result of detailed discussions and evaluations with stakeholders and screening managers in the three regions. The identified solutions were those that encountered acceptability, feasibility, and affordability criteria from both the health operators and the women's point of view. In Albania, the traditional FF invitation by community nurses was compared to PCs, which is a direct and proactive approach that has been seen as an effective strategy to encourage female attendance to CC screening. In Morocco, FF was compared with a specifically designed strategy, named “invitation made in Morocco” (IMM), which primarily relied on PC invitations and, if unsuccessful, employed home visits. In Montenegro, where an active screening program was already in place, it was agreed to juxtapose the usual care via phone contacts with a personalized invitation letter, as in many HICs. Sending a personalized invitation letter holds the potential to extend the program outreach and achieve a wider coverage of the target population in a cost-effective manner, requiring fewer time and personnel resources.

    Table 1.  Screening characteristics and study protocol by country.
    Country Setting (Study period) Age Invitation list (source) Screening test Usual care Intervention strategy (comparator)
    Albania Fier District (Jan–Dec 2017) 30–49 years General Practitioner HPV-test Face to face Phone call
    Montenegro Podgorica (Nov 2018–Feb 2019) 30–36 years Montenegrin Health Insurance Fund Informatics System HPV-test Phone call Personalized letter
    Morocco Tangier
    Kenitra
    Marrakech
    (Jan 2018–Feb 2019)
    30–49 years Anti-tetanus vaccination Visual inspection with acetic acid (VIA) Face to face Invitation-made in Morocco

     | Show Table
    DownLoad: CSV

    The criteria to declare women as “not reached” were established based on the number of unsuccessful attempts in each arm (after two home visits in the FF arm and in case of incorrect phone numbers or after three unsuccessful attempts in PC arm). Additionally, this alternative approach, which is inherently proactive, could enhance the interaction with subjects, ultimately resulting in improved communication.

    The exclusion criteria at the moment of the test included the following: pregnant women from the eighth month onward, those who had not engaged in prior sexual activity, individuals with a history of hysterectomy, and those with a previous CC diagnosis and/or treatment.

    A supplementary qualitative analysis involved anonymous questionnaires to assess the participation barriers and facilitators, and employed a 5-point Likert scale. Women who refused screening (at least 30%) completed the barriers' questionnaires, while all screening attendees were provided the facilitators' forms. Additionally, their satisfaction with the invitation strategy and intentions for future screening participation were investigated among all questionnaire respondents.

    The study involved a sample size of 2000 women in Albania, 2000 women in Montenegro (with a balanced representation of 1000 from rural areas and 1000 from urban areas), and 1800 women in Morocco (600 from each province, encompassing both urban and rural areas). Data acquisition involved local entry, and subsequent anonymized transmission to the Coordinating Centre for analysis.

    Endpoints of interest encompassed various steps of the screening process:

    1. Women reached: expressed as a proportion of the eligible population, this parameter served as an indicator of practicability.

    2. Women scheduling a screening appointment: representing the intention of women to participate in CC screening, this metric was calculated as a subset of women reached.

    3. Women tested: reflecting the actual participation, this metric was derived as a subset of the women reached.

    4. Examination coverage: a comprehensive parameter denoting the proportion of women screened out of the eligible population, providing insights into the quality of the screening process.

    The individual pathways were assessed 3.5 months subsequent to the invitation, which was a time interval agreed upon with local providers and was considered sufficient for a woman to decide whether to comply with the screening invitation within the study period.

    Further stratified analyses, specific to the setting and age groups, were exclusively conducted for Albania and Montenegro. These analyses were not applied to Morocco due to the strategy adopted for data collection, which did not include absent women.

    The questionnaire analyses involved a grouping strategy, which combined the responses of “not at all/slightly” and “moderately” versus “very/extremely” for the relevant metrics. Descriptive statistics, two-sample independent t-tests, and χ2 tests were deployed for the quantitative analyses. Statistical computations were executed using STATA v. 15, thus ensuring robust analytical methodologies.

    Ethical considerations adhered to the principles of the Helsinki Declaration and national legislations, thus ensuring detailed information dissemination and obtaining written informed consent from all participants. Three distinct protocols were introduced, each tailored to specific areas, carefully developed with contextual diversities in mind, and inspired by collaborative decision-making with local stakeholders. Each Ethical Committee approved the trial. Qualitative questionnaires were collected in anonymous.

    In Albania (Table 2), the comparison between “usual care” (FF) and the “intervention strategy” (PC) involved 1996 women (urban: 996, 49.9%; rural: 1000, 50.1%). Due to the non-updated phone numbers, a 1:1 randomization was not feasible, and women without a correct phone number were assigned to the usual care arm. This resulted in 711 (35.6%) women being recruited through PC and 1285 (64.4%) being recruited through FF, with an unbalanced distribution between the two arms. This issue was particularly relevant in rural areas (PC: 17.7%; FF: 82.3%), while balancing was maintained in urban contexts (PC: 53.6%; FF: 46.4%). The mean age of the invited women was 41.6 ± 5.7 years.

    Usual care, represented by FF invitation, proved to be more effective across multiple parameters:

    1. Reaching (FF: 96.9% vs. PC: 67.5%; p < 0.01);

    2. Screening scheduling (FF: 98.3% vs. PC: 87.1%; p < 0.01);

    3. Testing (FF: 89.9% vs. PC: 68.3%; p < 0.01);

    4. Examination coverage (FF: 87.1% vs. PC: 55.8%; p < 0.01).

    Differences were noted by setting for reaching, screening scheduling, testing, and examination coverage, with rural areas consistently exhibiting higher rates (p < 0.01). No significant age-related differences were observed (p > 0.10).

    In Montenegro (Table 2), the usual care (PC) was compared to the intervention strategy (invitation letters), which involved 1758 women (urban: 1000, 56.9%; rural: 758, 43.1%). Following randomization, 876 women (49.8%) were contacted by phone and 882 (50.2%) were contacted by letter. The mean age of the invited women was 32.7 ± 2.2 years.

    The invitation letters exhibited a higher reachability (65.2% vs. PC: 52.2%; p < 0.01), while PC demonstrated higher rates for screening scheduling (53.4% vs. letter: 14.4%; p < 0.01), testing (33.9% vs. letter 11.7%; p < 0.01), and examination coverage (17.6% vs. letter 7.6%; p < 0.01). The urban areas consistently outperformed the rural areas across all steps (p < 0.05 or p < 0.01). Age-related differences were noted for scheduling, testing, and examination coverage, which were all higher in younger women (<34 years) (p < 0.01).

    In Morocco (Table 2), the usual care involved an opportunistic FF invitation for women that attended local healthcare centers, while the intervention strategy employed IMM, inviting women by phone, and with FF visits if they were not reachable.

    In Morocco, a random sample of women from the vaccination archives were selected for the intervention strategy (IMM), whereas an equal number of women that accessed health centers for other reasons formed the opportunistic group. Due to the nature of this comparison, randomization was not feasible.

    A total of 1854 women participated, which were distributed across Kenitra (32.4%), Marrakech (35.3%), and Tangier (32.4%). The recruitment was comprised of 947 women (51.1%) through opportunistic FF and 907 (48.9%) with IMM.

    No significant differences were observed for screening scheduling between FF (93.7%) and IMM (92.8%). All women that were reached through IMM were tested, thus ensuring congruence between the intention and the actual participation. The examination coverage was high, with no disparities by the recruitment strategy (FF: 93.7% vs. IMM: 92.8%; p > 0.10).

    The subjects were presented with either a barriers' questionnaire if they declined the screening invitation or a facilitators' questionnaire if they accepted. The demographic characteristics of the respondents are delineated in Table 3.

    In the context of the barriers' survey, the participants were instructed to complete the form directly during the recruitment process (either through PC or FF interactions), while those that received the invitations by letter underwent a secondary contact phase via phone. Among individuals that declined the screening invitation, the questionnaire acceptance rates were 35.3% in Albania, 31.9% in Montenegro, and 80.0% in Morocco, which was notably higher. While no disparities in the questionnaire return were discerned in Albania and Montenegro concerning the recruitment strategy and age, a conspicuous discrepancy in the setting was evident in Montenegro (urban: 7.5% vs. rural: 52.2%, p < 0.01).

    The respondents to the barriers' questionnaire exhibited marked distinctions across countries concerning socio-demographic features, including education, marital status, and health insurance (Table 3). In Montenegro, the predominant cited barrier was recent testing (87.0%), with other reasons being infrequently mentioned. In Morocco and Albania, the prevalent refusal factors included fear of testing positive (43.9% and 37.8%, respectively), a limited risk perception (35.6% and 37.8%, respectively), and embarrassment (42.0% and 22.4%, respectively). Across all countries, a negligible percentage of women (<5%) declined the screening due to their husband's advice, health problems, or concerns about fertility. Notably, irrespective of screening refusal, the invitation strategies were highly esteemed (>90%) in all settings.

    Table 2.  Cervical cancer screening pathway by country and type of recruitment.
    Country Usual care n (%) Intervention strategy n (%) p*
    Albania (n = 1996) Face-to-face Phone call
    Target population 1285 711
    Women reached 1245 (96.9) 480 (67.5) <0.01
    Women accepting 1224 (98.3) 418 (87.1) <0.01
    Women tested 1119 (89.9) 328 (68.3) <0.01
    Examination coverage 1119 (87.1) 328 (46.1) <0.01
    Montenegro (n = 1758) Phone call Invitation letter p*
    Target population 876 882
    Women reached 457 (52.2) 575 (65.2) <0.01
    Women accepting 244 (53.4) 83 (14.4) <0.01
    Women tested 155 (33.9) 67 (11.7) <0.01
    Examination coverage 155 (17.7) 67 (7.6) <0.01
    Morocco (n = 1854) Face-to-face Invitation made in Morocco p*
    Target population 947 907
    Women reached 947 (100) 907 (100) >0.10
    Women accepting 887 (93.7) 842 (92.8) >0.10
    Women tested 887 (93.7) 842 (92.8) >0.10
    Examination coverage 887 (93.7) 842 (92.8) >0.10

    *Note: p < 0.01, reported in bold are considered statistically significant.

     | Show Table
    DownLoad: CSV
    Table 3.  Socio-demographic information of women refusing or accepting the invitation for cervical cancer screening.
    Women refusing screening n. (%)
    Women accepting screening n. (%)
    Albania (n = 195) Montenegro (n = 107) Morocco (n = 100) Albania (n = 1495) Montenegro (n = 219) Morocco (n = 839)
    Recruitment
    Usual care 105 (53.8) 105 (85.4) 68 (68.0) 1104 (77.5) 153 (72.5) 412 (49.1)
    Intervention 90 (46.2) 18 (14.6) 32 (32.0) 321 (22.5) 58 (27.5) 416 (49.6)
    Setting
    Urban 153 (78.5) 11 (10.6) 55 (55.0) 545 (38.2) 151 (72.2) 386 (46.0)
    Rural 42 (21.5) 93 (89.4) 45 (45.0) 880 (61.8) 58 (27.8) 453 (54.0)
    Age (years) 40.7 ± 3.2 33.0 ± 2.1 38.0 ± 5.0 40.8 ± 3.2 31.7 ± 2.2 38.4 ± 6.2
    Education
    Elementary school or less 36 (25.4) 11 (10.3) 74 (74.8) 552 (37.5) 15 (6.9) 605 (74.1)
    High school 86 (60.6) 68 (63.6) 20 (20.2) 624 (42.4) 87 (39.7) 158 (19.3)
    College/University 20 (14.1) 28 (26.2) 5 (5.1) 296 (20.1) 117 (53.4) 54 (6.6)
    Marital status
    Married 137 (92.6) 90 (84.1) 80 (82.5) 1391 (94.0) 135 (61.9) 754 (98.1)
    Separated/divorced 3 (2.0) 5 (4.7) 8 (8.2) 31 (2.1) 10 (4.6) 28 (3.6)
    Single 1 (0.7) 11 (10.3) 6 (6.2) 21 (1.4) 71 (32.6) 14 (1.7)
    Widowed 7 (4.7) 1 (0.9) 3 (3.1) 37 (2.5) 2 (0.9) 4 (0.5)
    Health insurance
    None 36 (25.3) 2 (1.9) 32 (33.7) 862 (58.3) 5 (2.4) 185 (23.0)
    Private 20 (14.1) 33 (30.8) 2 (2.1) 291 (19.7) 48 (22.6) 26 (3.2)
    Social 86 (60.6) 72 (67.3) 61 (64.2) 325 (22.0) 159 (75.0) 594 (73.8)

     | Show Table
    DownLoad: CSV

    The screened women were tasked with completing a facilitators' questionnaire, with acceptance rates of 98.5% in Albania, 98.6% in Montenegro, and 45.2% in Morocco. No discernible differences in the questionnaire return were identified in Albania and Montenegro based on setting (p > 0.10) and age (p > 0.10). However, in Montenegro, a pronounced disparity by recruitment method was observed (PC: 98.7% vs. letter: 79.1%, p < 0.01).

    Similar to the barrier's questionnaire, the respondents to the facilitators' survey displayed notable disparities by country concerning socio-demographic features (i.e., education, marital status, and health insurance) (Table 3). Common facilitators included awareness of the importance of cancer prevention (over 94% in all countries), understanding the impact of early treatments to improve the prognosis (Albania: 69%, Montenegro: 95%, Morocco: 98.1%), and trust in the local healthcare systems (Albania: 80%, Montenegro: 77%, Morocco: 85%). Self-perception of CC risk was low in Balkan countries (Albania 21.9%, Montenegro 10.7%), but more substantial in Morocco (56.1%), particularly among non-schooled women (60.0% vs. 51.0%; p < 0.01) and in urban areas (68.7% vs. 43.0%; p < 0.01). The experienced invitation strategies were highly regarded, even among the screened women (>95%). Additionally, the respondents favourably evaluated organizational factors (e.g., waiting time, facilities, cleanliness) and the healthcare professionals' skills (as technical and communication). The majority expressed an inclination to participate in CC screening again if invited (Albania: 97.3%, Montenegro: 94.0%, Morocco: 99.6%).

    This study delved into locally-adapted recruitment strategies in regions characterized by notable CC epidemiology, with a prerequisite of at least a pilot CC screening presence [10]. The selection of tested approaches prioritized feasibility and affordability. Given the recognized impact of the methodologies on uptake [12][14], the testing of alternative strategies and the comprehensive evaluation of barriers and facilitators in LMICs became crucial to ensure the equitable provision of screening services.

    It is noteworthy that the majority of studies that explored CC screening recruitment methods have predominantly focused on HICs [12][14], with a limited number of experiences conducted in LMICs, outlying mixed results [15],[16]. In HICs, the effectiveness and widespread use of invitation letters have been well documented [12],[13],[17]. While this strategy has been demonstrated to be feasible and effective in some LMICs [15], the current absence of randomized trials has introduced a notable gap [18].

    In the current study, the implementation of invitation letters was found to be scarcely feasible, primarily due to the shortage of up-to-date population registries, impeding the identification and the contact of the complete target population. In particular, in Albania, the local providers highlighted that women were not accustomed to receiving healthcare letters through home mail, whereas phone numbers had been more readily available for previous engagements with the local healthcare system [10]. Similarly, in Morocco, previously reliable phone numbers were accessible from the most recent national tetanus vaccination lists. Consequently, in Albania and Morocco, uncertainty regarding the accuracy of the addresses would likely have resulted in many undelivered letters. For these reasons, PC was deemed as a more practicable option. In contrast, in Montenegro, where a population screening program involving PC was already in place, the acceptability and feasibility of implementing a more structured recruitment strategy, such as using letters of invitation, has been considered a viable alternative approach.

    As anticipated, a prevalent barrier encountered in our study was the unavailability and poor update of various contact details, encompassing both phone numbers and addresses. This finding aligns with a recent Armenian study wherein only 60% of eligible women could be successfully reached [15]. In Albania, the incorrectness of over 20% of the phone numbers accounted for the observed imbalance between the two study groups, which is a trend similarly noted in Montenegro, where more than 12% of the contact details were missing. In contrast, Morocco exhibited a less severe occurrence of missing information, attributed to the existence of a comprehensive national tetanus immunization registry with a coverage exceeding 95%.

    Defining the study population is a crucial step that can impact the success of the study. In our trial, the preselection within the eligible population was unavoidable due to the constraint of only reaching women with accurate contact information, even though this may have hampered the equity of the program. In Morocco, the usual care arm employed an opportunistic FF approach, wherein CC screening was exclusively offered to women already visiting healthcare centers. This selection criterion could elucidate the notably high acceptance observed in this setting. However, the process of randomization ensured the reliability of the comparisons, although generalizations should be approached with caution.

    Considering reachability, the most effective approaches were FF in Albania and invitation letters in Montenegro; alternatively, in Morocco, no discernible differences emerged. The limitations of PC included incorrect contact details and the inability to reach a substantial portion of the population. In Montenegro, despite invitation letters yielding better outcomes, the results were sub-optimal, with approximately 35% of letters remaining undelivered. In Morocco, IMM, which primarily involved phone contacts, demonstrated promising results by reserving the more expensive FF contacts exclusively for unreachable women (less than 5%).

    Intentions for CC screening were higher for FF in Albania, PC in Montenegro, and were equivalent in both arms in Morocco. The decline from intention to actual testing remained consistent in the PC groups (Albania: −18.8%; Montenegro: −19.5%), irrespective of the available reminders and the option for rescheduling appointments. This parameter exhibited variations among countries, with more favorable outcomes in Albania (82.3%) and Morocco (93.3%), while being significantly lower in Montenegro (18.8%).

    Regarding the actual participation, the most effective strategies were as follows: FF in Albania (90% vs. PC: 68%; p < 0.01) and PC in Montenegro (34% vs. letter: 12%, p < 0.01); and in Morocco, no significant differences were observed, with both groups surpassing 90%. The success of the FF strategy in Albania was anticipated, given the interpersonal trust established through contacts with healthcare professionals, which was a trend corroborated by a recent study in Brazil [19].

    The observed differences in the indicators between the urban and rural settings aligned with previous literature that outlined such disparities [20],[21]. The recruitment approaches played a substantial role in shaping these variations. In Albania, better outcomes were identified in rural areas, likely linked to the trusting relationships with healthcare professionals in these communities [20],[21]. Conversely, in Montenegro, higher participation rates were found in urban areas, potentially attributable to a greater CC awareness [20],[21]. Notable differences in the women's attendance rates between the rural and urban settings highlighted the need for practical solutions to increase the screening utilization and reduce barriers in rural areas.

    The examination coverage varied among the different recruitment strategies: FF demonstrated a higher coverage in Albania (FF: 87.1% vs. PC: 46.1%), PC showed a higher coverage in Montenegro (PC: 17.6% vs. letter: 7.6%; p < 0.01), and no significant differences were seen in Morocco. The efficacy of FF, in achieving a higher examination coverage, underscores the importance of direct communication with healthcare professionals, particularly for women with limited CC awareness. However, the cost associated with FF remained a challenge [22]. In this context, PC emerged as a promising strategy to enhance both the participation and the affordability, albeit requiring an update of archives to achieve an optimal coverage.

    While both tested modalities (FF vs. IMM) proved effective in Morocco, limiting invitations to women already visiting healthcare centers raised equity concerns. However, IMM could offer an encouraging solution to reach all eligible women. The low experienced coverage in Montenegro, albeit consistent with national data [9],[10], emphasized the necessity for additional awareness initiatives.

    The qualitative questionnaires revealed an overall satisfaction and favorable attitudes toward CC screening in all countries. The identified facilitators and barriers aligned with those found in investigations across other LMICs [23][25]. Given the established role of CC education programs to increase the screening knowledge and participation [26], the implementation of similar initiatives should be considered.

    The strengths and limitations of the study should be acknowledged. A notable strength lies in the testing of recruitment strategies in countries where screening services were in their early stages. All protocols were adapted to the context and collaboratively developed with local stakeholders and healthcare professionals. Despite challenges in the data collection, the study rigorously analyzed the recruitment strategies through multiple step-by-step assessments, providing a comprehensive evaluation for optimizing screening. Lastly, the common membership of the involved countries to a shared screening network can be instrumental in disseminating findings across the Mediterranean area.

    However, certain limitations need to be considered. A common challenge was the lack or limited updating of the information archives, resulting in a significant portion of the study's sample being unreachable, limiting the generalizability and undermining the equity of the proposal. Furthermore, obtaining reliable data in LMICs is challenging, which is exacerbated by the information systems' scarcity and the reluctance of local healthcare professionals to collect and provide data to monitor the healthcare processes. Finally, the study exclusively tested traditional contact modalities (letter, PC, or FF), while the exploration of new technologies could be beneficial, particularly in LMICs [27],[28].

    The pursuit of a uniform recruitment strategy for all LMICs appears both impractical and unfeasible, given the complex interplay of diverse local factors that necessitate careful consideration. Invitation strategies facilitating direct contact with healthcare professionals (FF and PC) have shown better outcomes. These approaches provide comprehensive and reliable information, building trust and encouraging women to engage in screening, particularly in regions with limited awareness of CC prevention. However, their sustainability is questionable, especially for the FF approach, which considered time, economic, and personnel constraints to scale up the efforts. Therefore, while these strategies may not be viable for the entire target population, it could be reserved for hard-to-reach groups.

    For these reasons, a sustainable approach could foster collaborations with professionals across different healthcare sectors, specifically those specialized in women's health, including maternal and child health. In this direction, family physicians could play a crucial role in informing and sensitizing women about CC prevention. Moreover, a partnership with prominent voluntary associations to sustain awareness on CC and broader women's health issues could actively promote health education and support women grappling with this cancer.

    Attaining higher cancer screening coverage and participation in these regions demands a robust political commitment coupled with a multidisciplinary approach. A fundamental prerequisite to foster equity in these contexts is the establishment of complete and regularly updated population registries. Population-based registries are pivotal to enhancing patient care programs. Consequently, it is imperative for LMICs to prioritize investment in the planning and development of such registries, facilitating the implementation of modern and effective screening initiatives.

    Furthermore, the integration of new technologies, such as mobile health interventions and leveraging social media platforms, warrants careful assessments due to their inherent attributes of cost-effectiveness and expansive reach. These technological innovations present promising avenues for promoting increased participation in cancer screening initiatives.

    In this endeavor, international networks can play a pivotal role by offering substantial support to cancer screening providers. These networks facilitate the exchange of knowledge among their members and contribute to the capacity building efforts, thus enabling the implementation of more contextually appropriate and effective screening services. These synergies are instrumental in navigating the complex and heterogeneous landscape of CC screening in LMICs, addressing the unique challenges posed by diverse regional contexts.

    The authors declare they have used an Artificial Intelligence (AI) tool (ChatGPT) in the creation of this article for improving the English language.


    Acknowledgments



    The authors would like to thank Ms Wing Lam Chung and Ms Amanda Shue Qi Chia for conducting the pilot test; and Ms Alyssa Elyn Pua and Mr Shandon Shi Kai Ng for performing the quality evaluations. The authors would also like to thank Mr Jovan Kris Kamadjaja, Mr Francis Tee Heng Chia, Mr Shandon Shi Kai Ng and Mdm Betty Gan for lending their smart devices for this study (Google Nest Mini, Samsung Galaxy Note 9, Echo Dot and iPhone 6S respectively).

    Conflict of interest



    No competing financial interests exist. All authors have no conflicts of interest with any of the evaluated apps or their companies in this study. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

    [1] Kinsella B, Mutchler A (2018) Voice assistant consumer adoption report. Voicebot.AI . Available from: https://voicebot.ai/wp-content/uploads/2018/11/voice-assistant-consumer-adoption-report-2018-voicebot.pdf.
    [2] National Public Media, Edison Research.The Smart Audio Report. National Public Media (2020) . Available from: https://www.nationalpublicmedia.com/insights/reports/smart-audio-report/.
    [3] Laricchia F (2022) Number of voice assistants in use worldwide from 2019 to 2024 (in billions). Statista . Available from: https://www.statista.com/statistics/973815/worldwide-digital-voice-assistant-in-use/.
    [4] Laricchia F (2022) Factors surrounding preference of voice assistants over websites and applications, worldwide, as of 2017. Statista . Available from: https://www.statista.com/statistics/801980/worldwide-preference-voice-assistant-websites-app/.
    [5] Kinsella B, Mutchler A (2019) Voice assistant consumer adoption in healthcare. Voicebot.AI . Available from: https://voicebot.ai/wp-content/uploads/2019/10/voice_assistant_consumer_adoption_in_healthcare_report_voicebot.pdf.
    [6] Ferrand J, Hockensmith R, Houghton RF, et al. (2020) Evaluating smart assistant responses for accuracy and misinformation regarding human papillomavirus vaccination: Content analysis study. J Med Internet Res 22: e19018. https://doi.org/10.2196/19018
    [7] Alagha EC, Helbing RR (2019) Evaluating the quality of voice assistants' responses to consumer health questions about vaccines: An exploratory comparison of Alexa, Google Assistant and Siri. BMJ Health Care Inform 26: e100075. https://doi.org/10.1136/bmjhci-2019-100075
    [8] Goh ASY, Wong LL, Yap KYL (2021) Evaluation of COVID-19 information provided by digital voice assistants. Int J Digit Health 1: 3. https://doi.org/10.29337/ijdh.25
    [9] Kocaballi AB, Quiroz JC, Rezazadegan D, et al. (2020) Responses of conversational agents to health and lifestyle prompts: investigation of appropriateness and presentation structures. J Med Internet Res 22: e15823. https://doi.org/10.2196/15823
    [10] Miner AS, Milstein A, Schueller S, et al. (2016) Smartphone-based conversational agents and responses to questions about mental health, interpersonal violence, and physical health. JAMA Intern Med 176: 619-625. https://doi.org/10.1001/jamainternmed.2016.0400
    [11] Bickmore TW, Trinh H, Olafsson S, et al. (2018) Patient and consumer safety risks when using conversational assistants for medical information: An observational study of Siri, Alexa, and Google Assistant. J Med Internet Res 20: e11510. https://doi.org/10.2196/11510
    [12] Bérubé C, Schachner T, Keller R, et al. (2021) Voice-based conversational agents for the prevention and management of chronic and mental health conditions: Systematic literature review. J Med Internet Res 23: e25933. https://doi.org/10.2196/25933
    [13] World Health Organization.Global report on diabetes. World Health Organization (2016) . Available from: https://www.who.int/publications/i/item/9789241565257.
    [14] World Health Organization.New WHO Global Compact to speed up action to tackle diabetes. World Health Organization (2021) . Available from: https://www.who.int/news/item/14-04-2021-new-who-global-compact-to-speed-up-action-to-tackle-diabetes.
    [15] Ow Yong LM, Koe LWP (2021) War on diabetes in Singapore: A policy analysis. Health Res Policy Syst 19: 15. https://doi.org/10.1186/s12961-021-00678-1
    [16] Amazon. Alexa developer document—Design your skill. Amazon, (n.d.). Available from: https://developer.amazon.com/en-US/docs/alexa/design/design-your-skill.html.
    [17] Google DevelopersIntegrate with Google Assistant. Google, (n.d.). Available from: https://developers.google.com/assistant.
    [18] Martineau P (2019) Alexa, What's my blood-sugar level?. Wired . Available from: https://www.wired.com/story/alexa-whats-my-blood-sugar-level/.
    [19] One DropAlexa skills: One Drop. Amazon, (n.d.). Available from: https://www.amazon.com/One-Drop/dp/B072QDCSQH.
    [20] DataMysticAlexa skills: My Sugar by Jade Diabetes. Amazon, (n.d.). Available from: https://www.amazon.com/My-Sugar-by-Jade-Diabetes/dp/B0874717X2/ref=sr_1_3?dchild=1&keywords=diabetes&qid=1627894652&s=digital-skills&sr=1-3.
    [21] Epad IncGoogle Assistant: Diabetes Tips. Google, (n.d.). Available from: https://assistant.google.com/services/a/uid/00000063a0945bf1?hl=en-US.
    [22] Epad InccGoogle Assistant: Diabetes Checkup. Google, (n.d.). Available from: https://assistant.google.com/services/a/uid/000000e2388921d2?hl=en-US.
    [23] DietLabsGoogle Assistant: Well With Diabetes. Google, (n.d.). Available from: https://assistant.google.com/services/a/uid/0000000a181531b7?hl=en-US.
    [24] RocheAlexa skills: Sulli the Diabetes Guru. Amazon, (n.d.). Available from: https://www.amazon.com/Roche-Sulli-the-Diabetes-Guru/dp/B08BLTFY75.
    [25] eHealth Support NetworksAlexa skills: My Diabetes Lifestyle. Amazon, (n.d.). Available from: https://www.amazon.com/eHealth-Support-Networks-Diabetes-Lifestyle/dp/B08V8DGL69/ref=sr_1_1?dchild=1&keywords=diabetes&qid=1625367230&s=digital-skills&sr=1-1.
    [26] Heifner M Sulli The Diabetes Guru: Your diabetes voice assistant (2020). Available from: https://beyondtype2.org/sulli-the-diabetes-guru/.
    [27] Schachner T, Keller R, Wangenheim FV (2020) Artificial intelligence-based conversational agents for chronic conditions: Systematic literature review. J Med Internet Res 22: e20701. https://doi.org/10.2196/20701
    [28] Sezgin E, Militello LK, Huang Y, et al. (2020) A scoping review of patient-facing, behavioral health interventions with voice assistant technology targeting self-management and healthy lifestyle behaviors. Transl Behav Med 10: 606-628. https://doi.org/10.1093/tbm/ibz141
    [29] Cheng A, Raghavaraju V, Kanugo J, et al. (2018) Development and evaluation of a healthy coping voice interface application using the Google home for elderly patients with type 2 diabetes. 2018—15th IEEE Annual Consumer Communications and Networking Conference (CCNC) . https://doi.org/10.1109/CCNC.2018.8319283
    [30] Akturk HK, Snell-Bergeon JK, Shah VN (2021) Continuous glucose monitor with Siri integration improves glycemic control in legally blind patients with diabetes. Diabetes Technol Ther 23: 81-83. https://doi.org/10.1089/dia.2020.0320
    [31] Maharjan B, Li J, Kong J, et al. (2019) Alexa, what should I eat?: A personalized virtual nutrition coach for native American diabetes patients using Amazon's smart speaker technology. 2019—IEEE International Conference on E-Health Networking, Application and Services, (HealthCom) . https://doi.org/10.1109/HealthCom46333.2019.9009613
    [32] Statista Research Department.Worldwide desktop market share of leading search engines from January 2010 to July 2022. Statista (2022) . Available from: https://www.statista.com/statistics/216573/worldwide-market-share-of-search-engines/.
    [33] Yap KYL, Raaj S, Chan A (2010) OncoRx-IQ: A tool for quality assessment of online anticancer drug interactions. Int J Qual Health Care 22: 93-106. https://doi.org/10.1093/intqhc/mzq004
    [34] Health On The Net (HON) Foundation.HONcode certification. Health On The Net (2020) . Available from: https://myhon.ch/en/certification.html.
    [35] Charnock D (1998) The DISCERN Handbook: Quality criteria for consumer health information on treatment choices. Abingdon, Oxon: Radcliffe Medical Press 55 pp. Available from: https://a-f-r.org/wp-content/uploads/sites/3/2016/01/1998-Radcliffe-Medical-Press-Quality-criteria-for-consumer-health-information-on-treatment-choices.pdf.
    [36] Charnock D, Shepperd S, Needham G, et al. (1999) DISCERN: An instrument for judging the quality of written consumer health information on treatment choices. J Epidemiol Community Health 53: 105-111. https://doi.org/10.1136/jech.53.2.105
    [37] Robillard JM, Jun JH, Lai JA, et al. (2018) The QUEST for quality online health information: validation of a short quantitative tool. BMC Med Inform Decis Mak 18: 87. https://doi.org/10.1186/s12911-018-0668-9
    [38] Shoemaker SJ, Wolf MS, Brach C (2014) Development of the Patient Education Materials Assessment Tool (PEMAT): A new measure of understandability and actionability for print and audiovisual patient information. Patient Educ Couns 96: 395-403. https://doi.org/10.1016/j.pec.2014.05.027
    [39] Shoemaker SJ, Wolf MS, Brach C The Patient Education Materials Assessment Tool (PEMAT) and User's Guide (2013). Available from: https://www.ahrq.gov/health-literacy/patient-education/pemat.html
    [40] National Library of Medicine.Evaluating Internet Health Information Tutorial. MedlinePlus (2020) . Available from: https://medlineplus.gov/webeval/intro1.html.
    [41] Tan RY, Pua AE, Wong LL, et al. (2021) Assessing the quality of COVID-19 vaccine videos on video-sharing platforms. Explor Res Clin Soc Pharm 2: 100035. https://doi.org/10.1016/j.rcsop.2021.100035
    [42] Ministry of Health Singapore.Your Diabetes Questions Answered. HealthHub (2020) . Available from: https://www.healthhub.sg/live-healthy/1392/your-diabetes-questions-answered.
    [43] American Diabetes Association.Frequently Asked Questions: COVID-19 and Diabetes—How COVID-19 impacts people with diabetes. American Diabetes Association . Available from: https://www.diabetes.org/coronavirus-covid-19/how-coronavirus-impacts-people-with-diabetes.
    [44] GoogleFAQ about Google Trends data—Trends Help. Available from: https://support.google.com/trends/answer/4365533?hl=en&ref_topic=6248052.
    [45] AnswerThePublicSearch listening tool for market, customer & content research. Available from: https://answerthepublic.com/.
    [46] American Diabetes AssociationAmerican Diabetes Association—Connected for Life. Available from: https://diabetes.org/.
    [47] Centers for Disease Control and Prevention.Diabetes. Centers for Disease Control and Prevention (2021) . Available from: https://www.cdc.gov/diabetes/index.html.
    [48] National Institute of Diabetes and Digestive and Kidney Diseases.Diabetes. National Institute of Diabetes and Digestive and Kidney diseases (NIDDK) . Available from: https://www.niddk.nih.gov/health-information/diabetes.
    [49] Cleveland Clinic.Diabetes: An overview. Cleveland Clinic (2021) . Available from: https://my.clevelandclinic.org/health/diseases/7104-diabetes-mellitus-an-overview.
    [50] US Food and Drug Administration.Food and Drug Administration homepage. US FDA . Available from: https://www.fda.gov/.
    [51] Diabetes UK.Diabetes UK—Know diabetes. Fight diabetes. Diabetes UK . Available from: https://www.diabetes.org.uk/.
    [52] UK National Health Service.Diabetes—NHS. National Health Service . Available from: https://www.nhs.uk/conditions/diabetes/.
    [53] Victoria State GovernmentBetter Health Channel—Diabetes. Available from: https://www.betterhealth.vic.gov.au/health/conditionsandtreatments/diabetes.
    [54] Diabetes AustraliaDiabetes Australia homepage. Available from: https://www.diabetesaustralia.com.au/.
    [55] Ministry of Health SingaporeHealthHub Health Services. Available from: https://www.healthhub.sg/.
    [56] National University Health System.National University Hospital: Home. National University Hospital . Available from: https://www.nuh.com.sg/Pages/Home.aspx.
    [57] Royal Australian College of General Practitioners, Diabetes Australia.Management of type 2 diabetes: A handbook for general practice. Royal Australian College of General Practitioners (2020) . Available from: https://www.diabetesaustralia.com.au/wp-content/uploads/Available-here.pdf.
    [58] Ministry of Health Singapore.Diabetes Mellitus—MOH Clinical Practice Guidelines 1/2014. Ministry of Health (2014) . Available from: https://www.moh.gov.sg/docs/librariesprovider4/guidelines/cpg_diabetes-mellitus-booklet---jul-2014.pdf.
    [59] Diabetes Care.Introduction: Standards of Medical Care in Diabetes—2021. Diabetes Care (2021) . Available from: https://care.diabetesjournals.org/content/44/Supplement_1.
    [60] National Institute for Health and Care Excellence.Type 1 diabetes in adults: Diagnosis and management—NICE guideline. National Institute for Health and Care Excellence (2021) . Available from: https://www.nice.org.uk/guidance/ng17.
    [61] Cosentino F, Grant PJ, Aboyans V, et al. (2020) 2019 ESC Guidelines on diabetes, pre-diabetes, and cardiovascular diseases developed in collaboration with the EASD: The Task Force for diabetes, pre-diabetes, and cardiovascular diseases of the European Society of Cardiology (ESC) and the European Association for the Study of Diabetes (EASD). Eur Heart J 41: 255-323. https://doi.org/10.1093/eurheartj/ehz486
    [62] American Diabetes Association.6. Glycemic targets: Standards of medical care in diabetes—2021. Diabetes Care (2021) 44: S73-S84. https://doi.org/10.2337/dc21-S006
    [63] American Diabetes Association.Myths about Diabetes. American Diabetes Association . Available from: https://www.diabetes.org/diabetes-risk/prediabetes/myths-about-diabetes.
    [64] Centers for Disease Control and Prevention.Flu & people with diabetes. Centers for Disease Control and Prevention (2022) . Available from: https://www.cdc.gov/flu/highrisk/diabetes.htm.
    [65] Rahmatizadeh S, Valizadeh-Haghi S (2018) Evaluating the trustworthiness of consumer-oriented health websites on diabetes. Libr Philos Pract, 1786.
    [66] Keselman A, Arnott Smith C, Murcko AC, et al. (2019) Evaluating the quality of health information in a changing digital ecosystem. J Med Internet Res 21: e11129. https://doi.org/10.2196/11129
    [67] National Library of MedicineMedlinePlus evaluating internet health information: A tutorial National Library of Medicine (2018). Available from: https://medlineplus.gov/webeval/EvaluatingInternetHealthInformationTutorial.pdf.
    [68] Smith DA (2020) Situating Wikipedia as a health information resource in various contexts: A scoping review. PLoS One 15: e0228786. https://doi.org/10.1371/journal.pone.0228786
    [69] Google Developers.Policies for actions on Google. Google Developers . Available from: https://developers.google.com/assistant/console/policies/general-policies.
    [70] Amazon.com Inc.Alexa developer documentation: Policy requirements. Amazon . Available from: https://developer.amazon.com/en-US/docs/alexa/custom-skills/policy-testing-for-an-alexa-skill.html.
    [71] Savitha S, Hirsch IB (2014) Personalized diabetes management: Moving from algorithmic to individualized therapy. Diabetes Spectrum 27: 87-91. https://doi.org/10.2337/diaspect.27.2.87
    [72] Patel N 6 timely SEO strategies and resources for voice search. Available from: https://neilpatel.com/blog/seo-for-voice-search/.
    [73] Shafiee G, Mohajeri-Tehrani M, Pajouhi M, et al. (2012) The importance of hypoglycemia in diabetic patients. J Diabetes Metab Disord 11: 17. https://doi.org/10.1186/2251-6581-11-17
    [74] Diabetes.co.uk.Hypoglycemia (low blood glucose levels). Diabetes.co.uk (2022) . Available from: https://www.diabetes.co.uk/Diabetes-and-Hypoglycaemia.html#:~:text=Diabetes%20UK%20recommend%20that%20you,a%20non%2Ddiet%20soft%20drink.
    [75] Centers for Disease Control and Prevention.How to treat low blood sugar (Hypoglycemia). Centers for Disease Control and Prevention (2021) . Available from: https://www.cdc.gov/diabetes/basics/low-blood-sugar-treatment.html.
    [76] Amazon.com Inc.What is natural language understanding?. Amazon . Available from: https://developer.amazon.com/en-US/alexa/alexa-skills-kit/nlu.
    [77] Kinsella B (2019) Voice assistant demographic data—Young consumers more likely to own smart speakers while over 60 bias toward Alexa and Siri. Voicebot.ai . Available from: https://voicebot.ai/2019/06/21/voice-assistant-demographic-data-young-consumers-more-likely-to-own-smart-speakers-while-over-60-bias-toward-alexa-and-siri/.
    [78] Cherney K, Wood K (2022) Age of onset for type 2 diabetes: Know your risk. Healthline . Available from: https://www.healthline.com/health/type-2-diabetes-age-of-onset#age-at-diagnosis.
  • medsci-10-01-008-s001.pdf
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(12839) PDF downloads(294) Cited by(1)

Figures and Tables

Figures(4)  /  Tables(5)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog