Keywords

1 Introduction

Diabetes is one of the most common chronic diseases in the world. According to the data provided by the World Health Organization (WHO) in 2016, there are 422 million diabetics in the world [1]. The WHO further estimated that 1.6 million deaths in 2016 were directly related to diabetes, and another 2.2 million in 2012 were related to the critical levels of high blood sugar [2].

As high glucose levels can damage the body, untreated diabetes can cause different diseases, such as hypertension, depression, blindness, kidney failure, stroke, and heart attacks. These diseases together with diabetes adversely affect the life quality of patients and complicate their daily life [3]. If the treatment if provided timely and effectively, including precise blood sugar control, diabetes patients can live a long life, especially, when receiving appropriate treatment from the early stages of the disease. However, treatments need to be tailored to the individual patient groups and be initiated immediately after diagnosis.

In Europe, 286.000 children were diagnosed with diabetes in 2017 with 28.200 new diagnosed children registered every year [4]. All over the world, there are 1.1 million children and adolescents under the age of 20 with diabetes [5], and the numbers are continuously increasing by approximately 3%–4% corresponding to the cases of newly diagnosed children with diabetes in Europe each year [6]. Currently, the recovery treatment for diabetes does allow curing the disease completely but helps to slow its progression down [7].

With regard to children, treatment can specifically be challenging, as blood sugar needs to be monitored on a regular basis, and respective actions need to be undertaken. Often the parents need to take over responsibility of performing appropriate monitoring and treatment, particularly, if the children are diagnosed at a young age [8]. Therefore, the need to perform continuous monitoring of blood sugar is highly valued by the parents as a means to prevent hypoglycemia as well as hypoglycemia-related anxiety [9]. In accordance with these facts, technology advances should be employed to develop and improve an approach aiming to support children with diabetes and their families in their daily life.

Therefore, in the present study, we developed the prototype of a smartphone application connected to a smart toy used to support the children with diabetes and their families. In the present paper we describe the development process and share the results of the two iterations of evaluation involving experts and parents of children with diabetes.

2 Related Work

Various studies evaluated the usage of technologies for diabetes mellitus. Investigations were mainly dedicated to continuous monitoring systems, however, tailoring health games for adults and children was also studied.

The advantages of digital approaches in comparison to the traditional paper-based diaries have been outlined by Palermo et al. [10]. They showed that children had a higher acceptance rate toward such digital solutions and perceived them as easy to use. Furthermore, they identified out that the digital version contained less errors compared to the entries of a paper-based control group, and that the compliance rate was significantly higher for male participants.

The research conducted by Franklin et al. in 2008 revealed a novel approach for early notification systems for the diabetics [11]. In this study, a messaging system was proposed to encourage and support young people with diabetes. It was found that young people accepted this system and used it to send positive responses to the central message provider. Therefore, notifications and alerts are deemed as significant components of smartphone applications, as they are widely accepted by users.

Mougiakakou et al. introduced a complex technology-driven method to support diabetics in 2009 [12]. According to this research, the intervention included a combination of physical activity monitoring, food intake counter, blood sugar measuring, and a pressure measurement device. The application they developed combined all features and measurements within a single platform. Moreover, it included a physician platform to monitor the patient health situation. Therefore, the physicians could access the data from their phones or computers. To facilitate the connection between the patients and physicians, they employed a mobile or Wi-Fi network. As a result, they proposed a two-sided complex system for supporting diabetes patients. However, the proposed system implied having an assistant partner to perform monitoring and intervention.

Van der Drift et al. discussed the advantages of using a robot to motivate children for keeping a diary [8]. They demonstrated that the robot allowed increasing the motivation of children to write more details into their diaries with the average number of characters written augmented from 37 to 83, indicating a highly efficient interaction between child and a toy.

Further, Al-Taee et al. in 2017 illustrated how a robot assistant could help children to manage their diabetes [13]. The robot interacted with the child and talked to the child to understand how the situation of the child was. Moreover, the robot provided advises, and thereby educated the child how to cope with diabetes. After recording interactions with the child, it sent the dialog data to a health care partner, so that it was possible to monitor the health situation of the child through a web application. As a result, they reported positive outcomes indicating high acceptance rates of over 80%.

In turn, Toscos et al. in 2012 evaluated the impact of technologies on the parent-child relationships [14]. They conducted interviews involving children and parents. As a result, they demonstrated the potential of the positive influence of pervasive technologies on the parent-child relationship and highlighted the possibilities to design an application, which could be used to support the users in changing their behavior by focusing on their emotional reaction to the corresponding health data.

3 The System

The aim of the present study was to provide the parents with a tool to assist them to nudge the children to measure their blood glucose level regularly and thereby to maintain a healthier lifestyle [15].

3.1 Development Methodology

Seeking to develop a user-friendly and scientifically solid application, we followed the design science research approach [16].

To obtain insights into the existing methods of developing applications for diabetics, as well as to investigate functional requirements corresponding to child-friendly applications in this area, we conducted a structured literature review based on the approach of Webster and Watson [17]. This helped us to obtain first insights into important features of the planned application as well as to evaluate the possibilities of motivating and teaching children through the technologies, as shown in Sect. 2 of this paper.

Thereafter. we proceeded with the definition of requirements. With this purpose, we analyzed the findings obtained from the literature review to derive essential functionality of the planned application.

After the first iteration, the system design was evaluated involving experts to ensure inclusion of important features in terms of medical treatment from the viewpoint of a doctor. The obtained feedback was incorporated during the next development session. After completing the second development iteration, the prototype was evaluated involving several experts of the first evaluation session as well as parents of children with diabetes.

3.2 The Prototype

The main purpose of the proposed system is to provide the patients with a child-friendly solution, while supporting parents in monitoring their child’s health.

As shown in Fig. 1 the proposed system is composed of the four following components: a blood glucose measuring sensor placed on the skin of the child (1), a plush toy, which functions as a data receiver connected to the sensor (2), a parent application to process and represent currently measured blood glucose values (4), and a cloud database to exchange the data between the toy and the parent application (3).

Fig. 1.
figure 1

The four different components of the proposed system.

Full size image

The Blood Glucose Sensor.

A Freestyle Libre NFC-sensor [18] was employed as an example of a state-of-the-art technology for continuous measurement of the blood glucose level. Furthermore, its high prevalence among diabetes patients made it a suitable example to present to the target group. The sensor is placed on the skin to measure the glucose level of diabetics constantly. For this purpose, a small, sterile needle is placed under the skin and fixed by a plaster. To send the data to the receiver, the sensor uses near-field communication (NFC). Therefore, the data are only transmitted when the receiver is placed at a maximum distance of 4–5 cm in front of the sensor.

The Sensor-Value Receiving Toy.

A teddy bear was chosen as the children’s companion, as shown in Fig. 2. We decided to use a widely popular plush toy due to multiple reasons. First, the children can easily get attached to it and build up an emotional relationship. We hoped that this could not only increase the probability that the children would take it with them everywhere, but also that such form would encourage them to follow the nudges. Second, it is a simple and portable toy, which provides the sufficient space for the hardware, and its material does not severely restrict the wireless connection. Finally, it is made of soft materials and thereby provides certain protection to the hardware inside.

Fig. 2.
figure 2

During the prototyping process, a smartphone was used as hardware within the teddy bear.

Full size image

The application implemented into the teddy bear is able to register the data from the NFC sensor. After receiving the information, the most recent data are sent out and uploaded into the online database.

Furthermore, we attempted to provide the experimental teddy bear with the ability to speak. This would allow parents to enter phrases in the parent’s app which would be send to the bear and there synthesized to an audio output via TextToSpeech (TTS) technology. The idea behind this feature was to provide the supervisors with an option to encourage the child to measure their current blood glucose (through pushing the “Hug me” button, so that the child would attach the bear near to the sensor) or to give instructions if the child would be in a state close to hypoglycemia or hyperglycemia.

The Database.

We used the open-source database MongoDB as a back-end program and Amazon Web Services as a cloud platform for the storage area. To integrate the back-end with the toy and the parent app, a REST API running on a Flask9 microframework was implemented. This architecture enabled the back-end to send and retrieve the data to and from two applications in the JavaScript Object Notation (JSON) format.

Furthermore, as the application is intended to gather personal information of users such as name and age as well as highly sensitive health data, such as measurement values, data security is an important issue to address. In the proposed system, the data are secured via state-of-the-art hashing, access management, and encrypted transmission.

The Parent App.

To enable parents to monitor the current glucose level of their child through a convenient user interface, we specifically developed the parent app. Furthermore, to use the TTS feature of the plush toy, parents could transmit directives to the child, as shown in Fig. 3. This feature was implemented to encourage the parent to support their children, even if they were far away from each other.

Fig. 3.
figure 3

Directives input (left) and the home page (right) of the parent app. The measuring unit was mg/dl; demonstration target ranges of the blood glucose level were based on [19].

Full size image

However, the main purpose of the user app, was to present the measured data and valuable insights to the parents. The home page was implemented in such way to provide an overview of the daily data including the latest measured blood sugar level and an overview of the measurements during the last twelve hours. In addition, the home page included a battery level indicator, as well as the “Hug Me” button, which was intended to trigger the plush toy to ask the child for taking a new measurement.

Next to the daily overview of the child’s blood sugar level, the app was able to retrieve reports over a flexible period of time, which allowed deriving further insights such as average levels or notable patterns, as shown in Fig. 4.

Fig. 4.
figure 4

Report page (left) and emergency number management (right) of the parent app. The measuring unit was mg/dl; demonstration target ranges of the blood glucose level were based on [19].

Full size image

Additionally, the user app allowed the parents to save important telephone numbers, such as the number of the kindergartner. When the glucose level of the child appeared to be critical, the parents would receive a push notification from the app, which would redirect them either to the home page or to the emergency phone numbers. Using the list of emergency phone numbers into which new numbers could be added flexibly from the list of contacts, they could directly contact the person watching over the child and inform them about the critical blood glucose level.

4 Evaluation

To develop a user-friendly and a scientifically sound end product, we conducted two rounds of evaluation.

4.1 First Iteration Evaluation

In the first evaluation round, we conducted structured interviews involving experts from the medical and psychological fields to obtain professional feedback and inputs. We focused on collecting feedback on the current prototype and thereby obtained insights with regard to additional features of the application, as well as required changes in the design.

Methodology.

We conducted structured interviews to collect qualitative data with regard to the first prototype of the system. Therefore, we aimed to recruit experts from the area of diabetic treatment for children, as well as psychologists to provide feedback on the proposed system and help to improve the design.

We were able to recruit eight experts from two different universities for the interview sessions: six child endocrinology doctors (E1–E6), and two child and family psychologists (E7–E8). To conduct each interview, we visited the experts at their working places. We followed the same structure and asked the same questions in every interview. To facilitate retrospective evaluation, all interviews were recorded.

In the beginning, we explained the functionality and interface of the initial parent application prototype and the matching interactive toy. Additionally, we showed a demo video to help the experts to get an impression of how the prototype was supposed to work. After that, we asked about existing problems faced by children and their parents regarding during the management of diabetes. Then, we asked to describe expectations and requirements for such application from the professional viewpoint and evaluated, which psychological aspects were important to address during the application development.

After finishing all fourteen interviews, we evaluated the individual responses to identify and prioritize change requests to be considered in the next development phase.

Results.

According to the experts, one important feature corresponding to the parent app would be the option to register results of a glucometer (in addition to the values tracked automatically through the sensor), as the results obtained via different tools might have discrepancies. E6 stated that the possibility to observe differences in measurements would provide important information about child’s metabolism, as the sensor results corresponded to delayed values, while the glucometer results were much more precise with regard to the current state. Therefore, glucometer results would be helpful to determine the amount of insulin to be taken by the child to regulate the blood sugar level.

E4 also emphasized the need in multiple visualization options including hourly, daily, weekly and monthly graphs. This was deemed important to assist parents and doctors in identifying the glucose trends and to provide them with better insights into the child’s health situation.

Regarding the interactive toy, E2 and the interviewed psychology experts outlined that the children should not know that their parents are able to control the teddy bear. Additionally, it was suggested to limit the direct control of the toy to a single parent device. The use of all other connected devices was suggested to be limited only to monitoring functionality and displaying the current values.

E7 advised to make it as interesting as possible (e.g., by varying the possible instructions the teddy bear can give to the child) as otherwise, the children might quickly lose the motivation to play with it.

Furthermore, E2 suggested to limit the number of information messages sent to parents/supervisors. This idea was also supported by E7 who stated, that the parent app could result in making the parents overprotective and advised to limit the interaction options between parents and the interactive toy (especially in regard to the TTS feature).

Finally, three of the experts also stated, that supporting would be a very helpful characteristic to implement in the final product, as it would not only allow enabling the toy to track the current glucose levels without the need in direct contact between the child and the toy, but also facilitating measurements during sleeping hours, as E3 suggested.

Overall, the experts appreciated the idea behind the project and agreed that it would be a suitable product for the children aged between 4–10 years.

System Changes.

After finishing the expert interview sessions, we analyzed the collected feedback to define and prioritize required changes to be incorporated in the system. We decided to focus on the following changes.

First, we focused on expanding the available data graphs to include hourly, weekly, and monthly bases. The expanded visualization feature was requested by E1, E2, E4 and E6, as these tables were very crucial for treatment purposes considering that doctors used them as the knowledge base to adjust the insulin amount prescribed to the patients.

Second, we enabled to entering the manual measurement data to compare the glucometer results with the sensor results.

Third, we limited the TTS functionality to predefined directives to prevent the possibility of parents getting overprotective or entering aggressive directives. The new predefined directives were selected in cooperation with experts.

Finally, we refined the alert/notification functionality. We added three different levels (yellow, orange, and red zones), as shown in Fig. 5, with different alarm settings.

Fig. 5.
figure 5

The added zones and setting options in the parent app. The measuring unit was mg/dl; demonstration zone ranges were based on [19]. (Color figure online)

Full size image

The yellow zone was intended for the low glucose level deviations (up to 10 mg/dl) from the target range. The plush toy would autonomously provide the child with the directives corresponding to this zone. If the child followed advises and the blood glucose level normalized accordingly, the teddy bear would not send any notification to the parents.

However, if the child did not follow the directives, and the glucose level reached the orange zone (between 10 and 20 mg/dl from the target range), parents would get notified as well as emergency contacts.

Once the blood glucose level reached the red zone (more than 20 mg/dl deviation), the toy would notify the emergency contacts. Moreover, it would initiate the alarm to notify people around the child.

Implementing these zones, we hoped to ease the burden on the parents, so that they would only need to get involved if situations became critical. Moreover, we hoped that this approach could encourage the children to learn how to cope with diabetes with the help of their toy and support the children to build new healthy habits.

4.2 Second Iteration Evaluation

To ensure the usability and detect possible usage barriers we performed the second evaluation round to obtain feedback from the experts, as well as from the intended target group of the application users. Therefore, the aim was to interview the parents and experts to obtain opinions from the both sided about the final app.

Methodology.

We contacted the experts who participated in the first evaluation round again to request them to analyze the progress and provide feedback about the new version. Two of them, E6 and E7, were willing to participate in the second evaluation round. Recruiting parents to be interviewed was more challenging than expected, as most of families were not willing to share their experience and opinions. We could only find the two parents (P1, P2) having a child in the aimed age range from 4 to 10-year-old, who agreed to participate in the review interviews.

During the parent interview, we first introduced them to the project and provided them with information about the system architecture and the application. We suggested them to consider the two main criteria to evaluate the app. These criteria were defined as perceived usefulness (PU) and perceived ease-of-use (PEOU) and were derived according to the technology acceptance model [20]. Then, we demonstrated the screenshots of the final app prototype. Afterwards, we asked for their opinions about how useful and how easy to use the app was.

During the expert interviews, we introduced them to the final app functionality and explained the new features and modifications. Then, we demonstrated the screenshots of the app and asked for feedback regarding PU and PEOU.

Results.

P1 stated that he considered the toy and the app to be very useful and innovative. Moreover, he outlined that he perceived the app as very clear and simple. He was able to understand and use all available features. Furthermore, he was familiar with the functionality of such apps, as he used other glucose monitoring applications for his child.

P2 confirmed the positive feedback from P1 and outlined that she found the app very useful and simple. The app was deemed to be easy to use and would be helpful for families and doctors.

E7 found the application very functional and stated that it was a novel method to use the toy as a data receiver that children could carry with them. Moreover, she found the app very simple and its user interface very clear. However, she noted that the colors of the app were too light, and they tired her eyes. She suggested us to use stronger color tones to prevent this issue. Her overall opinion on the proposed system was positive, and she assumed the parents would be willing to use this app.

E6 found the app very useful and stated that this kind of intervention was highly demanded in families. Moreover, he expressed that the app was very easy to use and noted that the user interface was too simple. He emphasized that the graphs and tables could be more detailed and include more information. His overall evaluation was positive and he approved the usefulness and functionality of the app.

5 Discussion

While the overall feedback is positive, and the proposed system is deemed to be a suitable approach to support families with children with diabetes, several ethical concerns remain unclear.

So, we do not yet have any insights into the system’s possible impacts on the family, and therefore, it is not clear how the proposed system may affect its dynamics and single members. During the interviews, one of the psychologists, E7, mentioned the danger that parents may become overprotective. Although we attempt to mitigate such impacts by introducing the zone-based system with the automatic directives feature, parents might still feel responsible to regularly check the blood glucose levels of their child to ensure that they are within the acceptable range. This can be even increased in the case when other persons, such as supervisors in day-care centers or primary schools, have access to the system. The social pressure concerned with being constantly informed about the child health state may therefore increase. Additionally, the proposed system might also affect the child. Although the experts suggest that parents should not tell the child that they can control the toy, children still might feel constantly monitored and therefore pressured. Furthermore, if, by chance, they discover this, they may feel betrayed by their parents.

Another concern is the acceptance by the children. Currently, the design of the teddy bear does not provide a possibility to actually communicate with the child, as communication is one-sided from the toy to the child. Children might perceive this as either boring after a particular period of time, or they may perceive the toy as a supervisor, which may affect the emotional connection and possibly lead to the resistance against bringing the toy everywhere. Furthermore, small children might be confused by the fact that their plush toy talks to them but does not respond to their questions.

Finally, as the system synchronizes the measured data with the phone of the parents, the measured values are sent to an online server. Therefore, the highly sensitive medical data are supposed to be stored in the cloud. Although security measurement was performed to protect the information in the best way possible, the residual risk remains. All data accessible on the Internet are potentially at risk of being stolen in the case of a hacker attack. This has been demonstrated by successful attacks against large and established technology companies, such as Apple and Sony. Due to this reason, we need to investigate whether there are alternative ways to process the data, or whether the medical data in general should not be stored online.

6 Conclusion, Limitations, and Future Work

In the present study we demonstrate that the concept of a supportive toy for children with diabetes is deemed to be appropriate for younger children of the age between four and ten years. We have obtained qualitative feedback to improve the application in two rounds and have developed a system which not only helps parents to monitor the blood glucose level of their child, but moreover, is developed considering the self-caring approach by using automated directives by the plush toy. This might be used not only to help in increasing the child’s self-confidence but also in building new healthy habits.

Overall, the general feedback provided by the parents and experts during the conducted interviews was positive. Their interests and reactions were affirmative and they were willing to use the proposed system. The evaluation outcome indicated that people were interested in using this novel approach to monitor their child’s diabetes. Moreover, all interviewees agreed that the app was easy to use. Therefore, based on the obtained feedbacks, we assume that the users can get adapted to the app.

However, there are several limitations to be considered with the regard to the present study. Although we collected feedback from the experts and parents, we currently do not have any insights yet on how the children might react to the plush toy. Therefore, at present, we cannot make any assumption on how well the children might adopt the proposed toy and how carefully they would follow the predefined directives. Moreover, it is not clear how successfully they might establish a connection and thereby considering the toy their companion and bringing it with them wherever they go.

Therefore, in the future, we plan to conduct an intervention study in which the children will actually use the proposed system for several weeks. This would help to clarify the knowledge gap regarding the potential of the toy acceptance, as well to provide insights on how the proposed tool would influence the behavior of parents.

Furthermore, the system can be expanded by adding various auxiliary features. First, sensors with Bluetooth connection should be integrated to increase the ease of use on the children side. Using Bluetooth sensors can facilitate the plush toy to update the current blood glucose value even when it is several meters away from the child and thereby enabling constant measurement during the night when the child is sleeping. Moreover, adding a speech recognition unit into to the toy may increase interactivity and thereby help establishing a long-lasting interest of the children.

Finally, machine learning-based approaches which can analyze the data on the cloud server, might be applied to extract meaningful patterns from the measured data und therefore, support parents, doctors and children to improve the diabetes treatment.