Background and Aims

After 9/11 and several global epidemics, syndromic surveillance has become an even more important part of a country’s national emergency preparedness. The challenge is to detect the spread of an infectious disease as early as possible. This project is based on a hypothesis that through analysis of health data from people with Type 1 Diabetes (T1D), it is possible to detect a disease close to the point of infection, i.e., long before the disease onset.

Methods

An in-house Electronic Disease Surveillance Network, EDMON system, has been developed to test our hypothesis. EDMON is a real-time early disease outbreak detection system that uses self-recorded health data from people with T1D. The models have been tested using 11 years of high precision continues data recorded from 3 individuals with T1D.

Results

By combining continuously recorded health data (blood glucose, insulin, carbohydrates, physical activity and geographical location) from people with T1D, we are able to detect abnormal changes in their BG, changes that indicate that they have been infected with a pathogen. The discrepancies are identified by the use of advanced statistical and machine learning models.

Conclusions

Through the EDMON system, we can receive real-time information about the spread of infectious diseases in our neighborhood, and thus, can take the necessary action in order to reduce the risk of becoming infected. EDMON might be a very useful tool for people with T1D as well as the rest of the population in disease prevention.

Background and Aims

Personal health-sensors and devices are quickly entering the marked, answering the needs of people with diabetes’ self-management. This has led to an increasing amount of patient-gathered health data, which we foresee will be important in meetings between healthcare personnel and patients. Building on the previous FI-STAR project, we address this issue in the current FullFlow project.

Methods

Prior to testing an in-house developed system that allows people with diabetes to share their self-gathered data during consultations, we queried healthcare personnel (n=17; 12 GPs, 4 nurses, and 1 nutritionist) about their perceptions of, and suggestions for, the proposed system.

Results

All the healthcare personnel informants reported that they expected the designed system to be useful during consultations. Ten of them gave specific suggestions about how they expected the system to function, including: 1) possibilities for remote consultations; 2) support for keeping track of types of carbohydrates, not only amount; 3) support for keeping track of lipid levels; 4) automatic data transfer from apps, e.g. Strava, and devices, e.g. glucose meters and insulin pens; 5) support for all kinds of mobile phones; 6) integration of this system’s functions with electronic health record systems; 7) highlighting changes since last consultation; 8) transfer of consultation notes and hospital system information into the patients’ app to provide them with tailored recommendations for follow-up at the next consultation.

Conclusions

Healthcare personnel are positive to a system for using patient-gathered data, and they contribute with creative and specific suggestions for how such systems should work.

Background and Aims

Fear of hypoglycemia (FOH) refers to phobic avoidance reactions associated with hypoglycemia. FOH is associated with poor glycemic control and its consequences, necessitating the development of effective methods to address this condition. We present a pilot study, combining biofeedback (BF) with a virtual reality (VR) smartphone application.

Methods

Patients with type 1 diabetes suffering from FOH as evaluated by the "Hypoglycemia Fear Survey-II" questionnaire [composed of two subscales; Behaviors (HFS-B) and Worries (HFS-W)] were randomly assigned to either treatment with BF or BF+VR.

All participants were instructed to use a smartphone application daily for two weeks. They were exposed to virtual stimuli: mild low glucometer readings (85-125) and practiced reducing their physiological arousal using a galvanic skin response (GSR) biofeedback system. In the BF+VR arm successful relaxation was reinforced by exposure to their own virtual smile; this stimulus has the potential of activating brain reward mechanisms. Primary outcome was defined as the change in the HFS-Worries scale.

Results

Five participants were recruited and randomly assigned to receive BF+VR (n=3) or BF alone (n=2). Participants demonstrated a significant improvement in their HFS score (an average reduction of 7.4 points from a baseline of 83.4), mostly attributed to a reduction in the HFS-Worries subscale (5.8 points). This improvement was durable for a 2 month period. There was no clear trend towards superiority of the BF+VR over BF alone.

Conclusions

The results of this feasibility study, using a smartphone-based virtual reality application to treat FOH, call for a larger randomized controlled trial.

Background and Aims

Research on health technologies traditionally report clinical measures. However, with mHealth and online resources for diabetes self-management, individuals are calling for more, diverse evidence. We compare two reviews to determine to what extent mHealth and online intervention studies address patient-reported needs.

Methods

A systematic review (Review 1) searched for reported outcomes of mHealth and online intervention studies (PROSPERO registration: CRD42018115246). A literature review (Review 2) searched for patient-reported needs for diabetes self-management. Both covered articles published between 2015 and 2019. For ease of comparison, the co-authors categorized the results.

Results

Reviews 1 and 2 resulted in n=31 and n=21 articles, respectively. Main categories of reported outcomes were: support from/access to resources, usability/suitability, patient empowerment/engagement, clinical outcomes, and data protection. Main needs categories were: support/access to services, information, coping and patient engagement/empowerment, and technology. Thus, the research outcomes and patient needs were in general very different. For example, under the category support/access to services, reported intervention outcomes included peers, coordinated-care services and relevant information. However, specific patient-reported needs included resources and services to self-management activities, e.g. gyms, feedback on self-management performance and reminders.

Conclusions

A reason for these differences is that research interventions occur within closed and controllable systems, whereas patient-reported needs result from experience in the real-world, with a multitude of resources. Future interventions can address this by include more contextual information, e.g. about participants’ access to resources, as baseline measures. In doing so, we can provide evidence of the relationship that these resources have on the success of the intervention.

Background and Aims

The use of an online study-management system can help to ease the burden of both participation in, and administration of, mHealth interventions. We describe the benefits and challenges of using such a platform to manage an intervention (FullFlow Project) involving both patients and their providers in the testing of an mHealth data-sharing system.

Methods

Our remote study-management platform consists of: a website used to monitor status and message the participants, a local server for automatic data-collection and analysis (Piwik, now Matomo) and an open source survey tool (LimeSurvey). Patient recruitment was initiated through health providers and continued through the platform. Two researchers and one developer administrated the study.

Results

The benefits of this platform included security and efficiency in distributing study-information and messages, as well as supporting participants from a single platform, based on open-source systems. For example, if a participant was not actively engaged in the intervention, we could then send messages specific to their situation. In the platforms’ current implementation, we have experienced three main challenges: 1-Participant follow-up requires manual tracking and initiation of messaging; 2-Data-collection requires manual review of data and interaction logs, from separate sources; and 3-Data-analysis requires specific programming to combine the differently structured output from each data source.

Conclusions

Future improvements to the system can include automation of tasks and additional software that can facilitate the organization of these data for analysis. For example, automatic merging of data-sources and generation of simple reports would make the system more efficient, which is especially important for mHealth interventions.

Background and Aims

An ontology model is presented that represents medical and metabolic profiles of type 2 diabetic patients. This model is constructedas follows: Design of the ontology contemplating the most relevant aspects of the treatment, such as the patient's medical profile (hereditary-family history, patient's medical history, signs, symptoms, diet, physical activity, and complications of diabetes). Construction of the ontology model by acquiring data of diabetic patients, and biomedical data. Development of a rule-based knowledge base to reason about the metabolic profile of diabetic patients. Finally, the evaluation of the ontological model.

Methods

Ontology Requirements. Identification of the scope of the ontology, the definition of possible scenarios, users, the competence of the ontology, and the quality characteristics.

Ontology Design. Consists of: term elicitation, hierarchical relationships definition, data properties relationships definition, object properties relationships definition, and axiomatization.

Ontology Construction. Instantiating the ontology with individuals, relations and axioms. Consistency checking is executed to verify that none of the class definitions and axioms has logical contradictions.

Ontology Evaluation. Evaluation of the ontology by competency questions.

Results

The resulting ontology system is composed of 2201 axioms, 195 classes, 22 object properties, 46 data properties, and 310 individuals. The level of DL expressivity is ALCHOIN(D),

Conclusions

An ontology model is presented focusing on Type 2 Diabetes Mellitus, which has several therapeutic options. This ontological model will serve as a broad knowledge base on the relevant aspects related to the therapy and medical treatment of diabetes. .

Background and Aims

While the potential of existing Type 1 diabetes (T1D) simulators in accelerate research in diabetes technology is well-established, current limitations in the capability to mimic some behavioral aspects of patients’ lifestyle impacting on glucose control call for new investigation. Here, in particular, we focus on carbohydrates consumptions and develop a new model for amount and timing of main meals, snacks and hypotreatments.

Methods

The dataset includes 20 European patients studied for 2-month in free-living conditions in a closed-loop trial (Messori et al., Diabetes Technol Ther, 2016). Time and amount of meals are reported for 2759 main meals, 1218 snacks and 958 hypotreatments. Ten different probability density function models are fitted by maximum likelihood to the distributions of carbohydrates amount of each meal category and the time between consecutive meals (Δt). The Kolmogorov-Smirnov test is used to select the most suitable distributions.

Results

Breakfast, lunch and dinner amounts are described by gamma (μ=40.67 g, σ=17.65 g), lognormal (μ=55.70 g, σ=30.36 g) and loglogistic (μ=60.49 g, σ=37.00 g) distributions. Breakfast-lunch and lunch-dinner Δt are modeled by gamma (μ=303.8 min, σ=74.54 min) and generalized-extreme-value (μ=389.6 min, σ=79.08 min) distributions. Snacks amount follows a gamma (μ=24.51 g, σ=17.08 g) distribution, and their Δt from the previous main meal is modeled by a generalized-extreme-value (μ=199.5 min, σ=109.0 min) distribution. Hypotreatments amount has a loglogistic (μ=19.25 g, σ=16.14 g) distribution independently from the hypo-severity.

Conclusions

Once refined using larger datasets, the newly developed models can be incorporated in T1D simulators to allow more realistic in-silico trials.

Background and Aims

Hypertension is quantitatively the most important risk factor for CVD and death in the world. Despite extensive knowledge treatment results from current methods are poor. Blood Pressure Doctor is one of the world's first all-inclusive digital healthcare system for management of patients with hypertension. The aim is to provide a more effective treatment using remote care and telehealth technologies.

Methods

The care concept is based on a patient smartphone application and a clinical decision support system for doctors. Home blood pressure data are transferred via Bluetooth to the app and medical record forming a basis for continuous monitoring, medical treatment and intense behavioral coaching. Health data compiled together with medication, lab test results and individual risk factor interpretation are presented to the doctor. Automatically generated feedback is reinforced through gamification and given to the patient in the app together with treatment recommendations trough asynchronous physician chat communication.

Results

Both clinical results and patient experience are positive, and the service has received very high ratings from the patients. Study results shows significant further improved blood pressure levels in a majority of patients with hypertension previously treated in outpatient care in Sweden. The combination of engaging, educating and empowering the patients, continuous monitoring of health data with regular feedback and intervention when needed is a strong care concept.

Conclusions

Proper hypertension management requires pharmacological and non-pharmacological interventions. Blood Pressure Doctor offers a new scientifically validated and evidence-based innovative digital care concept well suited to deliver medical care and individual lifestyle advice also for diabetes patients.

Background and Aims

Diabetes is one of the most challenging chronic health conditions in 21th century. Diabetes-related problems need proper patient education, and social media plays a role to disseminate information. The aim of our study was to evaluate the quality of medical information presented in social network.

Methods

We studied 20 Russians top popular influencers, based on those who had more than 100,000 followers in Instagram social media and who had hashtags #diabetes, #diabetic_nutrition.

Results

We found that 14 out of 20 of the blogs (20%) could not be considered credible sources of diabetes management information. These blogs have unrelieble and potentially harmful recommendations like "fruits are undoubtedly products that lower blood sugar" or such kind of not evidence-based information: “The main savior for diabetics is cinnamon. The combination of polyphenols and magnesium mimics the effects of insulin, making sugar levels drop significantly. Garlic is another very useful supplement that makes the pancreas secrete double-acting insulin.”

Conclusions

Social media influencers blogs are not credible resources for diabetes management. Popularity and impact of social media in the context of the diabetes epidemic suggests all influencers should be required to meet accepted scientifically or medically justified criteria for the provision of diabetes management advice online.

Background and Aims

Conducting a cost consequences analysis of the DIABEO system with or without telemonitoring service to improve glycaemic control in patients with diabetes poorly controlled with a basal-bolus insulin regimen.

Methods

TELESAGE was a multicenter, double-randomized, open-label trial comparing a control group (arm 1) vs two DIABEO systems: a software alone (arm 2), and software + nurse-assisted telemonitoring and teleconsultations (arm 3). The primary objective of the study was to investigate the effect on HbA1c changes after 12-months. Both data on the diabetes management and utility data were collected throughout the study and valuated according official guidelines in France in a collective perspective for the year 2016.

Results

Cost analysis was conducted on 665 patients (221/231/213 in arms 1,2,3 respectively). Average yearly direct diabetes management costs were €3,109, €2,404, €1,996 in arms 1,2,3 respectively (p<0.001). Including losses of productivity average yearly costs were €4,289, €,3309, €3,126 (p<0.001). These estimates did not include the price of the DIABEO system. Costs varied according to the utilization rate of the system with lower total costs in the sub-group of patients with a daily use of the system (one or twice a day) compared to the patients with discontinuous use. Health-utilities score were not different across arms 1,2,3.

Conclusions

The DIABEO system especially with nurse telemonitoring reduced yearly diabetes management costs due to a shorter hospital cumulated stay. Regular use of DIABEO may result in lower yearly total diabetes management costs. Impact on quality of life require further studies.

Background and Aims

The Paediatric Diabetes team of an inner-city hospital use Microsoft Excel to record Haemoglobin A1c (HbA1c) measurements of 116 patients. HbA1c levels inform the clinician of long-term glycaemic control and allow adaptation of the patient management plan.

IT literacy of staff, training times, and costs limit the use of technological advances. Microsoft Excel - a cheap, simple, and easily accessible program - empowers teams to create dynamic spreadsheets. We aimed to increase the time efficiency of the Paediatric Diabetes team by improving their spreadsheet.

Methods

The original HbA1c spreadsheet was assessed using a questionnaire and team discussion. The spreadsheet was amended based on team requirements and suggestions. Data grouping was used to simplify the spreadsheet. Conditional formatting highlighted missing data and appointments, upcoming appointments, and colour coded HbA1c values within certain ranges. Summaries were evaluated using “COUNTIF” and “AVERAGE” formulae. Four hours were spent by a medical student implementing these changes.

Results

All participants reported the new spreadsheet was easier to use. Post-implementation questionnaire responses noted that the spreadsheet had higher accuracy and improved layout. The time taken for data input by healthcare professionals reduced by 77.7% (from 45 to 10 minutes per month).

Conclusions

Better data presentation enables healthcare professionals to provide further support to those with higher HbA1c therefore improving patient care. This could be adapted to various teams, especially those managing chronic conditions. Clinicians could consider investing time in straightforward programs like these, as they can be effective at a lower cost than new technological advances.

Background and Aims

Information regarding Latent Autoimmune Diabetes in Adults (LADA) in Mexico is lacking. RENACED-DT1 is a longitudinal national registry for patients with type 1 diabetes (T1D) and LADA in Mexico.

Methods

Currently 1,586 patients have been registered; of those, 48 patients were diagnosed with LADA. We present a descriptive analysis of this group.

Results

58.3% are women, 65.2% over 30 years of age at diagnosis; 30.4% have had diabetes for <5 years, 37%, between 6 and 10 years, 15.2% between 11-15 years and 17.4% >16 years. Mean HbA1c and BMI at diagnosis were 8.42% and 23.7 kg/m2, respectively. GAD65-Ab, evaluated in 15 cases, were positive in 86.6%. At diagnosis 64% were treated with oral agents, 23% with insulin and 12.8% with both. Currently 89.5% are on insulin treatment, 16.7% with insulin pump, 56.3% with multiple daily injections, 8.3% on basal insulin and 4.2% on a co-formulation; 50% have additional treatment with metformin, 23% with DPP-4 inhibitors, 19% with sulfonylureas and 6.3% with SLGT2-inhibitors. Subjects treated with DPP-4-inhibitors have lower HbA1c (7.1% [5.8-7.8] vs 8.5%, [5.8-7.8], respectively p<0.01) (Figure 1). Three patients have retinopathy and 3 neuropathy. No nephropathy was found. Dyslipidemia occurs in 30.6% and hypertension in 16.7%.

Conclusions

A high prevalence of LADA cases are misdiagnosed. In order to improve diagnosis and understand the presentation of this condition the registration and follow-up of more cases is needed. Recognizing this condition is important to start optimal treatment (DPP-4-inhibitors in C-peptide positive patients, insulin in C-peptide negative patients) and delay chronic complications.

Background and Aims

The DIABEO system is a telemedicine solution for real time monitoring of insulin treatment. Many people downloading a mobile health app in real-life stop using it after a single utilization. Therefore, we conducted a large pragmatic study to investigate the usage and efficacy of DIABEO in conditions close to real-life (the TELESAGE study).

Methods

Pragmatic, multicenter, randomized, open-label, three parallel-arms study in type 1 and type 2 diabetic patients poorly controlled with a basal-bolus insulin regimen (ClinicalTrials.gov NCT02287532). TELESAGE compared: (i) a control group (arm 1: standard care) versus the software alone (arm 2) and versus the software + telemonitoring by trained nurses (arm 3). The primary outcome evaluated HbA1c reduction after 12-month follow-up.

Results

Six-hundred-sixty-five (665) patients were included in the study. Baseline characteristics of patients were comparable among the three study groups. As expected, 25.1% participants in arm 2 and 37.6% in arm 3 used the DIABEO (± telemonitoring) system one or more times per day (DIABEO users). DIABEO users showed a significant and meaningful mean HbA_1c reduction versus standard care: -0.41% for arm 2 (P = 0.001) and -0.51% for arm 3 (P ≤0.001) without significant differences in hypoglycemic incidence. In the overall population (ITT) HbA1 reduction was comparable between arms.

Conclusions

Under pragmatic conditions, the effectiveness of DIABEO in reducing HbA1c levels in poorly controlled patients depended greatly on the frequency of its use. Significant HbA1c reductions can be obtained by using DIABEO at least once a day.

Background and Aims

As diabetes prevalence continues to increase, inpatient andout patient hyperglycemia becomes more common negatively affecting outcomes. During this session, the presenters will discuss the state of inpatient and outpatient insulin management, transition in care, how digital therapeutics have evolved, supporting research, and key factors to consider when moving from paper protocols to automated insulin dosing solutions.

Methods

This will be an oral presentation reviewing the impact of inpatient and outpatient hyperglycemia, hypoglycemia, and glycemic variability. Describing the differences between types of insulin and dosing algorithms. As well as reviewing the evidence on eGlycemic Management Systems (eGMS) and FDA cleared titration apps.

Results

Review how eGMS and titration apps have evolved over time. Compare and contrast features, benefits and downsides of eGMS and titration apps.

Conclusions

Recognize how eGMS and titration apps can drive engagement and improve outcomes. Discuss needed future areas for research and outcome studies.

Background and Aims

The main component of digital therapeutics in type 2 diabetes management is to engage people with diabetes (PWD) in self-care management. The digital interactions related to self-monitoring of blood glucose (SMBG) and insulin titration use sophisticated algorithms of care that produce improvement in outcomes. However, little is known on the behavior change theories used to support the engagement needed to produce ongoing usage and optimize outcomes. The aim of this oral presentation is to examine the behavior change theories used in the development and deployment of insulin titration apps and SMBG related digital therapeutics.

Methods

This will be an oral presentation of literature review about the use of the behavioral change theories used in development and deployment of digital therapeutics related to insulin titration apps and SMBG.

Results

Using evidence-based behavior change theories and philosophy is instrumental for individualization and scalability of the digital therapeutics. Utilization of behavioral theories in digital therapeutics transform the evidence of effective interventions where PWD are empowered, create concordance, activation, and engagement.

Conclusions

The use of behavioral change theories in development and use of digital therapeutics impacts the cognitive, affective, and behavioral experience for the users.

Background and Aims

Due to the increasing prevalence of diabetes worldwide, treatment and management costs of individuals with diabetes have become a major issue. Supporting self-management to improve glycaemic control by simplifying diabetes management is one way to address this aspect. The utilisation of blood glucose meters with colour-range indicator (CRI) has been shown to lower HbA1c. This budget impact analysis aimed at evaluating the financial impact of using a CRI-based glucose meter in five European countries: France, Germany, Italy, Spain and the UK.

Methods

Data from a randomised controlled trial, evaluating the effect of CRI-based glucose meters, were used to estimate the ten-year risk of patients for fatal myocardial infarction as calculated by the UKPDS risk engine. On the basis of assessed risks for myocardial infarction, the potential monetary impact for the five European healthcare systems was modelled.

Results

Based on a mean HbA1c reduction of 0.36%, as demonstrated in a randomised controlled trial using CRI-based glucose meters, the UKPDS risk engine estimated a 2.4% reduction of the ten-year risk of patients for fatal myocardial infarction. When applied to our economic model, substantial potential cost savings for the healthcare systems of five European countries were calculated: €547 472 (France), €9.0 million (Germany), €6.0 million (Italy), €841 799 (Spain), and €421 069 (United Kingdom) per year.

Conclusions

Utilisation of CRI-based glucose meters can aid the improvement of metabolic management in individuals with diabetes. In turn, this may also positively affect the expenditure of the healthcare systems on the complications of diabetes.

Background and Aims

Research suggests that social media could represent valuable channels for diabetes patient education. This study analyzes the engagement with health education contents posted on the Norwegian Diabetes Association’s (NDA) Facebook page.

Methods

All Facebook posts from NDA published between 01/01/2017 and 30/06/2019 were extracted, and classified according to its main contents’ type. This study was declared exempt by the Ethics Committee (REK-Sør-Øst,Ref:2017/764C). The treatment of personal information was approved by the data-protection officer at the University Hospital North Norway (Ref:0720).

Results

A total of 417 Facebook posts were published during the study period. Health education contents (i.e., learn self-management, self-monitoring) were the most shared type of content (p<0,05). A remarkable significant low engagement around exercise promotion’ contents was found, being both the least liked, shared, and commented (p<0,05). The table shows the engagement around the different types of content.

*Students t-test, p<0,05; **Mann-Whitney U test, p<0,05

Conclusions

Social media users engage considerably in health education content posted on Facebook. Since physical activity is very relevant for diabetes, exercise promotion’ content posted on social media should be presented in different ways to be more engaging.

Acknowledgements

We sincerely thank the Norwegian Diabetes Association (Diabetesforbundet) for their engagement in this study. This study is funded by Helse Nord RHF’s research fund (HNF1370-17).

Background and Aims

Molecular genetic testing is highly recommended for patients suspected of monogenic diabetes as it allows tailoring treatments to specific etiological mechanisms. Using WES, we studied the frequency and the spectrum of genetic variants causative of monogenic diabetes in a cohort of Russian children with non-type 1 diabetes mellitus.

Methods

A total of 60 unrelated patients with diabetes and impaired glucose tolerance (pre-diabetes) were prospectively included in the study (Russian ethnicity and resided in Northwest Russia). Genetic variants were screened by WES in a panel of 35 genes causative of maturity onset diabetes of the young (MODY) and transient or permanent neonatal diabetes.

Results

A total of 38 genetic variants were detected in 33 out of 60 patients (55%). Most of patients (27/33, 81.8%) had variants in MODY-related genes: GCK (n=19), HNF1A (n=2), PAX4 (n=1), ABCC8 (n=1), KCNJ11 (n=1), GCK+HNF1A (n=1), GCK+BLK (n=1), and GCK+BLK+WFS1 (n=1). In 6 patients (6/33, 18.2%), variants in MODY-unrelated genes were found: GATA6 (n=1), WFS1 (n=3), EIF2AK3 (n=1), and SLC19A2 (n=1). Fifteen out of 38 variants were novel ones: in GCK, HNF1A, BLK, WFS1, EIF2AK3, and SLC19A2.

Conclusions

Our data show a high frequency and a wide spectrum of genetic variants causative of MODY in Russian children. The prevalence of GCK variants suggests that diagnostics of monogenic diabetes in Russian children may start with testing for MODY2. However, the remaining variants are present at low frequencies in 9 different genes, altogether amounting to approximately half of the cases and thus highlighting the efficiency of using WES in non-GCK-MODY cases.

Background and Aims

Proactive self-care prevents hypo- and hyperglycemia, improving time-in-range and A1c. To be proactive, people with diabetes guess if BG is rising or falling and act accordingly. An inaccurate guess can lead to a harmful over or under correction (hypo- and hyperglycemia). To prevent such harm, we used app-entered BG data from CGMs to accurately predict upcoming low/high BG.

Methods

Data included contextual information, self-care, and BG values from app users with CGMs. Data were used to train a supervised learning model. The model generated predictions of each user’s BG 30 and 60 minutes into the future, as well as whether BG would be low (<70 mg/dL) or high (>180 mg/dL) in the next 30 minutes, one hour, and four hours.

Results

The mean absolute relative difference (MARD) for 30-minute predictions was 4.3%, with 99.7% of predictions falling in Zone A of the Clarke Error Grid, and 99.9% in Zone A or B. The MARD for 60-minute predictions was 13.4%, with 79.4% in Zone A, and 98.4% in Zone A or B. Hypoglycemia predictions showed 93.2% recall and 89.4% precision at 30 minutes, 83.2% recall and 74.1% precision at one hour, and an area under the ROC curve (AUC) of 91.9% at 4 hours. Hyperglycemia predictions showed 98.9% recall and 97.6% precision at 30 minutes, and 95.0% recall and 92.6% precision at one hour.

Conclusions

Pooling BG data from thousands of One Drop app and CGM users confers accurate, short-term BG forecasts, which can facilitate proactive, safer self-care.

Background and Aims

Medication adherence is important for glycaemic control. However, less than half of the people with Type 2 diabetes take medications as prescribed. Medication non-adherence can be a major health issue carrying significant negative consequences for individuals, as well as for healthcare systems. One of the known barriers to medication adherence is forgetfulness. Given the current ubiquity of smartphones, diabetes self-management apps represent a valuable avenue for providing medication reminders to help users adhere to their medication treatments and improve health outcomes.

Methods

We assessed 181 diabetes management apps for the presence of medication reminder features and compared their characteristics to the evidence provided in studies showing the effectiveness of SMS reminders. Inclusion criteria include: Android and iOS diabetes management apps updated at least once since January 2016, in English, targeted at users with Type 2 diabetes, with medication management functions.

Results

We found that only 55.8% (101/181) of these apps had a basic medication reminder function, although the ones that had them conformed to the characteristics of SMS reminders shown to be effective. The most common format for reminders were pop-up messages/banner (68.3%, 69/101), followed by vibration (48.5%, 49/101), and sound (42.6%, 43/101).

Conclusions

This is an important missed opportunity for diabetes management apps to contribute to improving medication adherence and, ultimately, to improve health outcomes. Further research is needed to establish whether medication reminders in apps are effective for increasing medication adherence (as seen with SMS reminders), and to elucidate the specific features that these reminders should have to improve medication adherence.

Background and Aims

Primary care (PC) is a key setting for the management of diabetes, as the PC provider is in a unique position to provide early, comprehensive, continuous and coordinated care, improving the outcomes for these patients. Digital, intelligent, computer-assisted history taking (CAHT) systems retrieve high-quality patient information prior to the consultation and can improve the PC consultation for diabetes.

Methods

We reviewed PC medical textbooks and published literature on CAHT between June-August 2019, focusing on the diabetes consultation. We conceptually mapped the identified specific elements of the diabetes PC consultation that a CAHT system can improve, summarizing our findings into a proposed framework (Table).

Results

At the initial consultation, a CAHT system can improve screening and diagnosis, by implementing a comprehensive assessment (lifestyle/health behaviours, family, social and work conditions, patient's beliefs) while screening for hypertension, obesity, and other risk factors. Each of these can be explored deeper without taking up consultation time and allowing a streamlined experience with an earlier, more accurate diagnosis. At subsequent consultations, a CAHT system aids with monitoring the patient’s management and progression of diabetes, allowing the interaction to focus on suboptimal behaviour, education, and early identification of complications, with more updated data coming from other specialists or tests.

Conclusions

Accurate and comprehensive patient information obtained outside the consultation through a CAHT system improves screening, diagnosis, monitoring and management of complications. It returns the focus of the consultation back to the provider-patient relationship, allowing for a better quality diagnosis, more consistent treatment, and guided and seamless patient engagement for diabetes.

Background and Aims

Smartphone health applications (apps) provide an opportunity to assist or support patients to better self-manage diabetes and other chronic conditions. A proportion of people may face difficulties when using these apps e.g. those with physical, cognitive, visual, motor challenges, etc. In order to remove such barriers, universal design principles and assistive features should be incorporated into app design. Currently, there are no widely accepted universal design principles for health apps. We developed assessment criteria to evaluate the inclusiveness of health app design.

Methods

Published and grey literature on (app) design for older users, best practice in universal design for websites, and assistive features for (health) technology were used to develop an assessment tool to evaluate the inclusiveness of app design. We focused on the aspects of user experience (UX) that would be important to evaluate from the perspective of older users of health apps.

Results

Possible barriers for health app use for older adults relate to impairments in vision, hearing, psychomotor skills (e.g. movement, hand-eye coordination), cognition (e.g. memory, attention), as well as motivation. As a result, the UX criteria are divided into 3 main sections: visual criteria, audio/sound criteria, and ease of use/navigation criteria (see Figure).

Conclusions

Every day, a larger proportion of older adults with diabetes and other chronic conditions are using health apps to manage their conditions. The Inclusive UX criteria presented here can guide app developers and healthcare providers to develop and recommend apps that would be suitable for this population’s use.

Background and Aims

The glycemic control among patients with diabetes in India is very poor and better self-management is the key to improving glycemic levels. This study evaluates the effectiveness of an iPDM (integrated personalized self-management) to improve blood glucose (BG) levels.

Methods

We included 184 participants (Average = 45.41 years; 72.71% male) who had a minimum BG logging frequency of ≥2 BG logs/wk for ≥4 months and were using an iPDM consisting of the Accu-Chek® Active BG monitor with the Wellthy Care™ digital therapeutic. We analyzed average BG (ABG), Fasting BG (FBG), and Post-meal BG (PBG), at baseline (BG logged 1-3 days from the start of the program), month 1 (M1), and month 4 (M4), and change in estimated HbA1C (eA1C) based on ABG.

Results

The average number of active days and BG logs by the participants were 66.62 days (95% CI: 60.83-72.4 days) and 93 logs (95% CI: 82-104 logs), respectively.

The BG values at baseline, M1 and M4 are summarized in table 1.

Based on the change in ABG (baseline-M4), the corresponding change in eA1C was 0.93% (7.22% to 6.29%).

Conclusions

The significant drop in BG from baseline to M1 and further reduction at M4 indicate that iPDM along with digital therapeutics could be an effective tool for improving self-management thereby enabling better glycemic control in a resource constrained country like India.

Background and Aims

Mobile healthcare based on smartphone apps is becoming a tool to improve self-care management of diabetics. but research is minimal on the achievement of diabetes education in Korea. We investigated the effect of intensive education by using mobile apps, logbook and booklet.

Methods

It was conducted on type 2 diabetics who visited five general hospitals in Seoul and Gyeonggi Province from April 20, 2016 to April 6, 2017. After randomly assigning the subjects to App group(27), Logbook group(32) and General group(29), All groups conducted pattern management education using booklet. The App group provided real-time education through mobile app. for six months. Logbook group provided education at every hospital visit, using diabetic logbook. General group provided only initial education. HbA1c, self-care behaviors, and quality of life was investigated before and after education to obtain the differences within a group and between groups

Results

The study involved 88 subjects (51M, 37F) with an average age of 55.8 years and an average duration of diabetes of 7.6 years.There were no differences in the self-care management behaviors among the three groups in the changes before and after 24 weeks, but the HbA1c was significantly decreased in the App group compared to the Logbook group and the General group(F = 5.997, p = .004).The quality of life of subjects improved in all three groups after 24 weeks, but there was no significant difference between the three groups.

Conclusions

In this study, Diabetes education is effective for blood sugar improvement. Mobile apps are effective in improving blood sugar management.

Background and Aims

People in Colombia with Type 1 diabetes (T1DM) have limited options for treatment. Sensor-Augmented Pump (SAP) therapy is funded by the Colombian health service, with Medtronic having the dominant market share. Non-pump users, on multiple daily injections (MDI), can choose between Freestyle Libre^TM flash glucose monitoring or capillary blood-glucose testing with a restricted choice of devices: Freestyle Lite^TM, OneTouch^® or GlucoQuick.

Clinicians only access T1D subject data during the medical consultation, via the manufacturers’ proprietary web-platforms. So they must master multiple user-interfaces and cannot monitor and advise T1D subjects remotely.

Methods

An ambulatory non-randomized 12-week, 4-visit pilot study aims to evaluate the potential to leverage Tidepool as a device-independent, cloud-based platform that could be tailored to facilitate data-sharing between Colombian T1D subjects and clinicians. The specific goals are to evaluate glycaemic outcomes and user-experience, as well eliciting user requirements. The study involves two clinicians and eight adults with T1DM: four on MDI and four on SAP therapy. All participants are initially trained to use the Tidepool web-based data-visualisation platform and to communicate via the Tidepool mobile app.

Results

Ethical approval has been granted for execution of the pilot study in 2019. The highly usable Tidepool platform looks like a promising candidate for customisation. Requirements are already emerging, including the obvious one of language.

Conclusions

This pilot study is novel in offering an alternative to the Colombian population with T1DM that is compatible with existing treatment. The next step is to implement the gathered requirements and conduct a larger clinical trial.

Background and Aims

A continuous quality improvement initiative in place in Italy since 2005 (AMD Annals) previously showed that patients with type 2 diabetes (DM2) attending diabetes clinics and treated with insulin had poor metabolic control. Aim of this study was to evaluate the percentage of patients with DM2 treated with insulin, their clinical characteristics, and their level of metabolic control.

Methods

Observational, real-world study based on data extracted from electronic medical records relative to all patients with DM2 seen in 222 diabetes centers in 2016.

Results

Out of 402,700 patients, 140,716 were treated with insulin (34.9%). Of these, 37.5% had HbA1c values> 8.0% and 23.8% had values ​​<7.0%. The most frequently used schemes (± other glucose-lowering agents) were the basal-bolus (basal + 3 injections of short-acting insulin; 46.1%) and basal insulin alone (32.8%). Significant proportions of patients, ranging between 29.9% and 46.0% in the various schemes, presented inadequate metabolic control. The median doses of insulin varied between 0.2 and 0.4 IU/kg in most of the schemes adopted, and rose to 0.6 IU/kg in the case of basal-bolus. Doses increased slightly as the average levels of HbA1c increased. The prevalence of complications was markedly higher in subjects on insulin therapy, indicating a generally late use of this therapy.

Conclusions

This analysis confirms the failure to reach adequate therapeutic targets in high percentages of subjects with DM2 treated with insulin, in disagreement with guidelines recommendations. The late use of insulin and the use of moderately low doses reaffirm the problem of therapeutic inertia.

Background and Aims

Food deserts are defined as “[locations] vapid of fresh fruit, vegetables, and other healthful whole foods, usually found in impoverished areas […] due to a lack of grocery stores, farmers markets, and healthy food providers.” Many studies have proven the existence of a correlation between diets poor in fresh produce and fibers to the metabolic syndrome. The purpose of this research is to investigate the relationship between urban food deserts and MSRD in 7 cities in California (San Francisco, Oakland, Stockton, Santa Clara, San Jose, San Bernardino, Merced).

Methods

Data were obtained from the CDC, California Department of Public Health, and the U.S. Census Bureau. A correlation and t-test were performed to investigate the relationship between census tract income, number of grocery stores located in that census tract and the prevalence among residents of Diabetes, Obesity, Hypertension, chronic Kidney disease (DOHK), Hypercholesterolemia.

Results

There is a negative correlation between median income and: presence of grocery stores (p=0.007), prevalence of DOHK (all p≥0.000) while the prevalence of hypercholesterolemia shows a positive correlation (p=0.015).

A t-test between the upper and lower 50^th percentile of income showed greater number of grocery stores in poorer areas (p=0.034) as well as higher prevalence of DOHK (all p≥0.000). This difference remains consistent and statistically significant even when we focus on the poorest (<25%) and richest (>75%) groups (table below).

Conclusions

This study contrasts the orthodox that food deserts generate chronic disease; the lack of food affordability is more likely responsible for higher prevalence of MSRD in poorer urban environments.

Background and Aims

MDC is an FDA-approved digital solution comprising a smartphone app for people with T2D and web portal for healthcare professionals (HCPs). The HCP defines an individualised long-acting basal insulin (BI) titration plan that is used by MDC to give dose recommendations to the user based on their fasting blood glucose (FBG) readings and hypoglycaemic event data.

Methods

The analysis included people with T2D in India who were registered for MDC by their HCP from 01/09/2018 to 31/03/2019 and received insulin glargine 100 U/mL or 300 U/mL. Users logged FBG and BI doses administered directly in MDC. Users reaching titration target were defined as those who logged three consecutive FBG measurements within their prescribed target range. Hypoglycaemic events were defined as below HCP-defined hypoglycaemia cut-off per the dose plan.

Results

Among 684 people who were registered with MDC and recorded ≥2 FBG readings over a 2-week period, 295 reached their FBG targets. From starting titration to reaching goal, mean FBG decreased from 168.5 to 107.7 mg/dL (−36%), and mean BI dose increased from 17.1 to 19.8 U (+16%). Mean time to reach goal was 18.1 days, with an average of 1 hypoglycaemic event per user. Users who up-titrated had higher mean starting FBG and greater FBG decrease compared with those with stable or decreased dose (Table).

Conclusions

Many MDC users were able to successfully titrate their BI and achieve target FBG, showing the potential for simple digital titration solutions to assist diabetes management in appropriate patient profiles in a resource-limited setting.

Sponsored by Sanofi.

Background and Aims

The use of smartphones and social media networking is widespread among teenagers. Multitude of smart phone applications (apps) have been developed to aid diabetes self-treatment. Several publications also support the value of using social networks as psychosocial support. Our aim was to assess the use of apps and social network in diabetes management and whether it could be a role for them to improve diabetes outcomes.

Methods

Transversal descriptive study of type 1 Diabetes aged 14 to 18 years old. Using an online survey, use of apps and social networks were explored.

Results

We obtained answers from 38 adolescents (mean 16.1 years old; 50% males; mean diabetes duration 7 ± 4 years). Almost all of them had smartphone, only 4/38 used them for diabetes management. The rest didn’t use diabetes apps, 51% showing no interest and 30% as punctual users in the past. Out of 34 non users, 56% believed apps could be helpful and they would use them if they were good. Most of the kids used social network several times per day, but only 11% recognised using them for diabetes. Among those who didn’t, 36% of them think are useless, and 24% don’t trust information. Despite this, 76% believe that a proper social network could be helpful in diabetes management.

Conclusions

Type 1 adolescents barely use apps and social network for diabetes management.

Most of them don’t trust information or don’t know good apps for diabetes self-control.

Helping them choosing good apps and using proper social networks would increase the use of these technologies in diabetes self-management.

Background and Aims

The prediction of glycemic profiles based on continuous glucose monitoring contributes to enhance diabetes management, helping to prevent hyper and hypoglycemic events or integrated in artificial pancreas. This paper proposes a new set of parameters to assess glucose predictors by integrating accuracy, delay and noise parameters.

Methods

The new metrics complement the frequently used root mean square error (RMSE), with the following: the delay-compensated RMSE (RMSE_DC), the delay-compensated RMSE into the 25-75% of the extremes (RMSE_DC-25/75), the overshooting/undershooting (OS/US); the mean delay (MD_25/75), the peak delay (PD); and the noise energy ratio (ENOISE_norm). We have applied the parameters to 5 synthetic test-profiles (TP) that emulate prediction profiles. TP_1-2 are built by adding a constant delay (15&10 min respectively) to an original CGM profile, and artificial deformations at the local maxima (TP_2). TP_3-4-5 correspond to a 10 min delayed CGM profile, adding differet types of noise (20dBW WGN in TP_3, plus extra high&low frequency noise in TP_4-5 respectively).

Figure. Synthetic test profiles

Results

The RMSE is not able to discriminate among TP_1-2-3, having clearly distinct performance (Figure 1) (RMSE:13.0-12.2-13.9), while the proposed metrics reveal their differences (RMSE_DC:0-7.0-8.9, RMSE_DC-25/75:0-0-9.4). MD_25/75 detects the delay in 4 out of 5 TPs and PD reveals if that delay is homogeneous. OS/US analysis allows a critical insight about the absence of false-alarms/false-negatives. ENOISE_norm reveals the presence of an extra medium-low-frequency noise in TP_4-5 vs. TP_3 (ENOISE_norm:4.27-8.09-9.85).

Conclusions

The proposed metrics allow an in-depth characterization of glucose predictors, helpint to select the optimum prediction technique depending on the specific application.

Background and Aims

Continuous glucose monitoring (CGM) is poorly accepted at some American diabetes camps. We evaluated the effectiveness of centralized multiplex monitoring of CGM data overnight at diabetes camp.

Methods

Sixty-nine, insulin-treated campers ages 7-18 years were enrolled at a single, week-long American diabetes camp. All campers performed regularly scheduled fingerstick testing, which informed treatment decisions. Campers that monitored their glucose with fingerstick testing at home were fitted with blinded Dexcom CGM at camp and served as control (n=16). Campers that used Dexcom real-time CGM (rtCGM) systems with sharing functionality at home continued to use their self-supplied system at camp (n=53); shared data from all enrolled campers wearing Dexcom rtCGM systems were simultaneously displayed on an investigational, centralized display in the medical cabin (Figure). The multiplex display alerted camp staff to glucose values <4.4 or >13.8 mmol/L, prompting fingerstick testing outside of scheduled intervals. Campers with <50% data coverage were excluded from the analysis. Overnight (22:00-06:00) CGM data were compared between campers wearing blinded and remotely monitored, rtCGM systems.

Results

Remotely-monitored campers experienced significantly less time <3.0 mmol/L and <3.9 mmol/L and significantly more time in range (TIR; 3.9-10.0 mmol/L) overnight compared to campers monitored with fingerstick testing alone. Overnight hyperglycemia was also lower among remotely-monitored campers (Table). Remote monitoring prompted 73 nocturnal interventions for events that might have otherwise gone undetected.

Conclusions

Centralized multiplex monitoring of rtCGM data at diabetes camp facilitates timely hypoglycemia treatment overnight, may disencumber camp staff, and should be encouraged at all diabetes camps.

Background and Aims

A teamwork including parents and people involved in the child’s life is often required in the management of diabetes for children with type 1 diabetes. In particular, family has to pay close attention to the child’s blood glucose level and insulin injections and teach the child to manage his/her diabetes.

To assist in this process, we developed a Furby toy-based system to motivate, educate and keep track of the child’s diabetes.

Methods

The Furby toy-based system has been designed using personas and different scenarios. The main issues were represented: 1) educational part, that teaches children when to take insulin or eat with parental supervision; 2) remote control of blood glucose by parents/babysitter; 3) ease the psychological discomfort of having diabetes by using the toy to play with his/her friends.

Results

The original Furby’s toy electronics were replaced so it could be controlled wirelessly by a Raspberry Pi 3 (Figure 1).

Based on blood glucose values received from Nightscout, Furby expresses different emotions (e.g., happiness, sadness) and announces its change.

Blood glucose values monitoring, text-based games (e.g., play hide and seek, learning the numbers) and control of the toy are possible using a smartphone via the chatbot component.

The Furby toy-based system includes a voice-based interaction/conversation using Alexa and the integrated microphone.

Conclusions

The implemented solution enables future expandability and modularity. It represents an example of how technological solution can be used to ease daily diabetes challenges in families.

Background and Aims

Ongoing patient self-management education and support using technology in diabetes are critical for achieving glycemic control and preventing complications.

My Dose Coach^TM, is a mobile app combined with a web portal, that can suggest optimized basal insulin injection doses using fasting blood glucose and hypoglycemia data. The aim of this study was to evaluate its efficacy on patients reaching Fasting Plasma Glucose (FPG) target range 90-130 mg/dl (5-7.2 mmol/L) without severe hypoglycemic episodes (low glucose levels that required assistance) and the mean reduction in A1c.

Methods

This experimental study involved the use of My Dose Coach^TM previously programed with a titration scheme according to fasting glucose in patients with type 2 diabetes that were treated with basal insulin in Clinica Nova in Monterrey, Mexico. We followed the patients for 16 weeks. Descriptive statistics and student t-tests were used when appropriate.

Results

We included 158 patients (46.8% women), mean (SD) age 51 (10.3) years. One hundred forty-one completed the study (89%). We achieved the FBG goal in 70.9% of the patients [66(28) days], with no event of severe hypoglycemia; 40.5% had 1-9 events of mild-moderate hypoglycemic [58(8) mg/dl] that was successfully treated at home. The mean (95CI) reduction of A1C was 1.78% (1.47-2, p<0.01) with last observation carried forward.

Conclusions

My Dose Coach app successfully helped to achieve FBG, reduced A1c with no events of severe hypoglycemia in patients with type 2 diabetes being treated with basal insulin.

Background and Aims

The use of technology in diabetes is helpful for glycemic control. My Dose Coach^TM, is a mobile app, that can suggest optimized basal insulin injection doses using fasting blood glucose and hypoglycemia data. The aim of this study was to evaluate the effect of using My dose Coach^TM in achivieng Self- Measured plasma Glucose (SMPG) goal of 90-130 mg/dl (5-7.2 mmol/L) and improving WHO´s Five Well Being Index (WBI).

Methods

This experimental study involved the use of My Dose Coach^TM previously programed with a titration scheme according to fasting glucose in patients with type 2 diabetes that were treated with basal insulin in Clinica Nova in Monterrey, Mexico. We followed the patients for 16 weeks. According to WBI, a change higher than 10% is considered significant. Descriptive statistics and student t-tests were used when appropriate.

Results

We included 158 patients (46.8% women), mean (SD) age 51 (10.3) years. One hundred forty-one completed the study (89%). Three patients suspended insulin needs. Sixty eight percent achieved the SMPG goal. There was a mean (95CI) increase of 2.23 (-3, -1.4, p<0.01) points in WBI scale (12.3%). There was no change in mean (95CI) units of insulin needed [-1.48 IU (-3.5,0.5, p= 0.15)].

Conclusions

My dose Coach was effective in helping patients to achieve SPMG goals, and in increase of WBI.

Background and Aims

Currently, there are a lot of free and low-cost diabetes self-management apps available. There is very little and heterogeneous scientific evidence on health outcomes. The objective of this study was to examine the effectiveness of a specific diabetes app (SOCIAL DIABETES) to improve glycemic control and other related outcomes.

Methods

149 people with type 1 diabetes were included in the randomized study in a control group receiving usual face-to-face clinical care (baseline, 3 and 6 months) and the intervention group with baseline visit and 6 months with incorporation into the use of the "social diabetes" app. All of them used the Glucomen areo glucometer. Analytical and clinical data including the download of glucometer and psychosocial data were collected at baseline and at 6 months along with a final survey of satisfaction.

Results

149 patients participated, of which 130 completed the study (67 CG and 63 IG). The mean age was 39.11 years with 19 years of evolution, and baseline HbA1c 8.5 + -1.2%. Both groups improved Hba1c similarly at 6 months. The intervention group (baseline: 8.43%, SD 1.25, 6 months: 8.22%, SD 0.99), compared to the control group (baseline: 8.66%, SD 1.17, and 6 months 8.24 %, SD 1.24) . There was also improvement in the number of mild hypoglycemia, without differences between the two groups. In the psychosocial variables studied there are no differences between intra or intragroup.

Conclusions

The use of a smartphone application related to diabetes impacts metabolic control, and can help in the follow-up of our patients.

Background and Aims

Diabetes technologies have improved diabetes treatment and patient’s quality of life. We aimed to determine the barriers that prevent healthcare professionals and patients from using advanced technologies for treating diabetes.

Methods

Prospective study, starting in December 2018, based on online surveys to diabetic patients and healthcare professionals involved in their treatment.

Results

Among 161 patients (66.5% women, 37.3±13.7 years, 92.5% type 1 diabetes), 89.4% used technologies for diabetes treatment (87.1% continuous glucose monitor [CGM], 30.6% insulin pump). The main barriers for patients were: cost (59.6% CGM, 64.0% insulin pump); device alarms (54.7% CGM, 41.0% insulin pump); desire to spend minimal time treating diabetes (45.3% CGM, 42.2% insulin pump), and fear of device malfunction (36.6% CGM, 43.5% insulin pump). From 95 professionals surveyed (74.7% women, 43.3±13.1 years), 76.8% were physicians (49.5% Endocrinology), and 16.9% nurses. 92.6% recommend the use of diabetes technologies (mostly CGM [90.9%] and insulin pumps [68.2%]). The major barrier to using CGM (69.5%) and insulin pump (72.6%) was the cost for users/coverage by the National Health Service (NHS) or insurer, while 49.5% admited not understanding information/features of insulin pumps, and 48.4% reported difficulty handling devices. Patiens and professionals consider that better NHS/insurer coverage, easier access to devices, more therapeutic education, and improved technology, could ameliorate the use of technologies in diabetes treatment. Most professionals consider important to increase consultation time and education on device's use.

Conclusions

Education of patients/their families and health professionals looks essential to override barriers in using diabetes technologies as an ally for evolution of diabetes treatment.

Background and Aims

Digital health services can support patients and facilitate self-management. However, user experience/satisfaction can impact uptake/outcomes. MyDiabetesMyWay(MDMW), an online service established in NHS Scotland in 2008, has demonstrated positive impacts on clinical outcomes. MDMW was recently piloted in NHS Somerset.

Methods

Participants in Somerset completed questionnaires covering six domains (see table) at enrolment and after six months. Quantitative data were analysed using Wilcoxon signed-rank tests and Generalised Equation Modelling. Free-text items were analysed using thematic analysis.

Results

483 participants completed questionnaires at both timepoints. Participants were: age 25-91 (mean 67.02, SD 9.80); 31.5% female; 10.6% type 1 diabetes.

Patients’ ratings of MDMW at follow-up were positive, with the majority (>60%) agreeing that the product ‘helped them to do what they wanted’, was ‘easy to use’, they ‘could get help when they needed it’, and they were ‘satisfied with the product’. Only low numbers (<4%) disagreed with these statements.

Some respondents reported that MDMW aided self-management, citing easy access to information/increased autonomy. Whilst one aspect of health confidence (‘I know enough about my health’) significantly improved at follow-up compared with baseline, several other scores declined. Combined domain scores echoed these patterns. Declining scores might reflect deteriorating health trajectories, and/or issues described in free text responses, including problems with logging in and incomplete/out of date clinical results, perhaps due to this being the inaugural launch in NHS England.

Conclusions

Digital health services can potentially supplement existing services. Usability/access issues and poor data quality can impact patient use/satisfaction. Improvements in these areas would enhance patient uptake and experience.

Background and Aims

Today mobile phones are fully integrated into daily life. As such, they can be used as tools by patients and healthcare providers to aid in the communication, retention, and utilization of health-related information. In diabetes management, there is potential for mobile apps to help manage outcomes by facilitating behavior change. However, many apps are didactic, not personalized, and do not provide actionable instruction. The purpose of this abstract is to describe a diabetes management application (app) design based on behavioral constructs.

Methods

A three phase approach to the app design included 1) literature review to identify relevant behavior change theories 2) app content and features development based on selected constructs and 3) testing of app content and features.

Results

From the literature review, four theories were identified: Health Belief Theory, Social Cognitive Theory, Fogg Behavior Model and the Transtheoretical Model. Constructs were selected to build a framework for app design. Relevant content and features, as well as methods of delivery, were developed based on selected constructs with the objective of promoting positive behavior change and increasing engagement in diabetes management. As designed, the app creates a user experience that establishes a baseline for desired behavior change and facilitates change through the provision of curated content.

Conclusions

The translation of behavioral theories into the design of this mobile app hopes to empower users to adopt positive health behaviors and enable better self-management of diabetes. Future research will observe and evaluate whether this behavioral methodology leads to positive behavior change and improved diabetes management.

Background and Aims

Mobile devices become usual advancements in daily lifestyles. Smart phone is now cheaper, a plenty of people were drawn into this virtual world. In Thailand, forgetting medication is one factor that affect uncontrolled diabetes. This study implemented research survey to developed primary care units (PCU) ’s portable diabetes application that cooperated diabetes patient and PCU service together. To survey and analyze the requirement of mobile application developing for diabetes patients in PCU that was approved for EASYDM application production

Methods

This survey is requirement analysis of mobile application, conducted August 2019 in PCU, Saimai, Bangkok, Thailand. In this study, 2 questionnaires on EASYDM mobile application requirements was designed: application user accessibility survey questionnaire, application user requirement survey questionnaire and questionnaires were given to 584 outpatients and 8 PCU staffs. SPSS 13.0 was used for statistical analyses.

Results

Findings: Sample were equal frequent in gender, age 41-55 years (46.4%) and middle school (46.4%). All mobile device type were android and less than 30% have application download experience.
Patients group recommendation were Thai language (54.5%) and less complex (43.7%). Both patients and staffs group range glycemic control as the first goal of the application. Patients group need appraisal function the most (99.3%) when staff group were reminder buzz, health index and laboratory information (100%).

Conclusions

This study shown the requirements and draws a brief picture of application. The survey suggest that languages and complexity are the most concern patient group. Prefer application were tri-parties and communicable.

Background and Aims

Proper training and follow-up for patients new to continuous glucose monitor (CGM) use is required to maintain adherence and improve outcomes. However, CGM training is hampered by a lack of evidence-based standards and poor reimbursement. The aim of this pilot feasibility study was to test a theory-driven, web-based intervention designed to provide extended training and follow-up support to patients newly implementing CGM, and to describe CGM adherence, glycemic control and psychosocial measures pre- and post-intervention.

Methods

Participants were new to CGM, 15-24 years old, with type 1 diabetes for >6 months. Each was randomized to enhanced standard care alone or in addition to the “Intervention Designed to Educate and improve Adherence through Learning to use CGM”, (IDEAL CGM) web-based training intervention using a 3:1 randomization scheme. Outcome measures were collected pre- and post- 3-month study period.

Results

In total, eight participants were randomized. Within the IDEAL CGM group 4/6 participants received exposure to the web-based training. This study demonstrated proof of concept and indicated components of the intervention may be effective despite insufficient evidence to determine the feasibility of the IDEAL CGM intervention and its described impact on diabetes-related outcomes.

Conclusions

This research underscores the importance of continued efforts to 1) establish evidence-based standards for training patients using CGM, 2) determine the dosage and/or level of educational exposure associated with improved outcomes and 3) determine key design elements of the web-based intervention to increase engagement, dissuade attrition, and ensure attainment of the skills necessary to achieve consistent use and improvements in glycemic control.

Background and Aims

Current methods for assessing glycemic control use averaged measures (such as HbA1C) and provide little information on the glycemic pathology of patients. Visual tools and mathematical formulas can allow improved characterization of glycemic behavior for achieving better glycemic control.

Methods

We developed new algorithms and visualizations that quantify and describe glycemic behavior, using device-agnostic data from previously published high frequency SMBG data. We plotted the glucose datasets as temporal glucose histograms, and calculated the glycemic burden (GB, weighted glucose over time) and glycemic severity index (GSI, a composite score of four glycemic variability measures).

Results

Patients with different problems in their glycemic control had histograms with different shapes, GB and GSI. Using these tools, patients who had the same HbA1c level were shown to have significantly different glycemic pathologies. The temporal evolution of glycemic control could be analyzed using intervals as short as 2 weeks, including diurnal variations.

Conclusions

We propose a paradigm change from current diagnosis and treatment methods. We propose to classify patients and their severity not by their HbA1C or single glucose measurements, but rather according to their actual glycemic pathology as determined from the glucose histogram, GB and GSI. By analyzing these parameters, a treatment can be devised that is geared towards the normalization of the glycemic behavior, as exemplified by the glucose histogram. This approach provides insights into the glycemic derangements, enabling the clinician to design a personalized treatment in addition, or as a replacement, to the HbA1C and standard algorithms.

Background and Aims

The Young Leaders in Diabetes (YLD) is a semester-long training, held since 2009. Among the leadership skills developed are: fundraising, basic and advanced management of diabetes, diabetes education, health policies and advocacy, influence, effective communication and interpersonal relationships. YLD are followed to assess their growth on the empowerment ladder, which impacts their community and the entire society. Our aim with this study was to understand how much the 6-month training impacted the HbA1c and diabetes knowledge in a subgroup of YLD.

Methods

Thirteen individuals (5 men, 22±4.5y) performed HbA1c exams and took a diabetes knowledge test of 35 questions: 32 multiple-choice and 3 open-ended questions based on IDF guidelines - twice, before starting and after concluding the training. Wilcoxon Signed Rank test was used to compare knowledge test grades and HbA1c values pre and post training; and also pre-training and current HbA1c of nine YLD.

Results

HbA1c results decreased significantly (pre: 8.0±0.5%, post: 6.9±0.3%; p=0.049); and knowledge test grades increased significantly (pre: 7.9±0.2, 8.6±0.2; p=0.012). HbA1c values continued to be significantly lower even years after concluding the training for all 9 individuals followed for 1 to 9 years post-training (current HbA1c: 7.1±0.2% p=0.035). Interestingly, none of the YLD have started pump or sensor therapy during the training, thus, these technologies were not interfering factor.

Conclusions

We attribute the HbA1c improvement and its maintenance to knowledge and skills developed by YLD during the training. In a future study, we plan to investigate the impact of starting pump and sensor on maintaining the HbA1c after YLD.

Background and Aims

The Young Leaders in Diabetes (YLD) are individuals between 15 and 30 years old who complete a semester-long training. Participants develop the following skills: fundraising, basic and advanced management of diabetes, diabetes education, advocacy, leadership, communication and interpersonal relationships. YLD who stand out are invited to become tutors, responsible for supporting new candidates throughout the training to ensure that they develop leadership skills. In addition, tutors assist in leading activities, conducting group dynamics, organizing and planning training sessions. This study aims to analyze the impact of the YLD training on diabetes knowledge and empowerment.

Methods

Thirteen individuals with average age of 20.4±3.0y answered the Michigan Diabetes Knowledge Test (DKT) and the Michigan Diabetes Empowerment Scale Short Version (DES-SF). Five of them were YLD tutors-candidates (Tutors), while the other eight were the YLD-candidates. Scores between the groups were compared using the Mann-Whitney U Test.

Results

The DES-SF scores were different between the groups (p=0.043, Tutors=4.75 and Candidates=4.13), indicating that individuals who finish the YLD training and stand out are overall more empowered than the ones with no previous YLD training. Tutors scored 20 (19-21), the YLD candidates scored 19 (13-23) on the DKT (no significant difference for DKT).

Conclusions

DES-SF and DKT were useful to reveal that individuals who completed the YLD training and stood out were more empowered and displayed more knowledge - which we attribute to the training - than candidates who had not started the training.

Background and Aims

As machine learning shifts the clinical paradigm in the health care system, more clinicians and patients have been benefited from using big data such as personal health records (PHR). This paper presents lean body mass(LBM) prediction models with features extracted from wearable device data.

Methods

Recruited 300 subjects aged 20-65 were required to wear ActiHeart (Camntech Ltd., UK), and underwent a basic checkup and a treadmill test. After preprocessing the metabolic equivalents and heart rates obtained from the device, we implemented K-mean clustering (6 centroids) in this data and performed regression analysis with these centroids to extract informative features. Prediction models were designed with covariates such as age, sex, height, weight, slope, and intercept.

Results

The results of 5-fold cross-validation of Linear regression and Random Forest models showed satisfactory performance in predicting LBM (with the coefficient of determination of 0.91 and 0.90, respectively). Root Mean Squared Error of LM was 2.98 and of RF is 3.15. Mean Absolute Error of LM was 2.21 and of RF was 2.35.

Conclusions

This paper shows the LM and RF models predicting LBM with additional features from ActiHeart data. Previous studies have found a linear relationship between LBM and peak VO₂ and between HR and VO₂. These linear relationships imply that the features (slope and intercept) used in the prediction models for LBM are reasonable. Continuous monitoring of body composition can motivate users to control weight and provide important insights for clinicians to interpret the association between diabetes and body composition.