Using Gamification to Improve the Quality of Educating Children with Autism

Smile1’ game structure

To better measure the issue of diagnosing emotional states in children with autism and according to time constraints, it was decided to study four basic emotions of ‘happiness, sadness, anger and fear’ in this game. Initially, the four senses are taught to the child in the form of social stories, and subsequently, the child is asked to determine the facial expressions related to each part of the story in the form of questions.

The game ‘Smile 1’ was created and manufactured using the Construct 2 game engine. Construct 02 is a great engine for building HTML and two-dimensional games, which was developed by Scirra.

From the very beginning of designing this game, very important points have been taken into consideration, and in fact, considering these points, we made this game specifically suitable for children with autism. One of the points is an attractive, yet simple interface design for children with autism. For this reason, based on research on the characteristics of autistic children, to design a fun and entertaining environment for children, there are no harsh or nasty sounds in this game. This is because autistic children are sometimes more sensitive to hearing and sight than normal people, and intense and nasty sounds, light and colours can harass them. Additionally, because of the focus and attention deficits of these children, the gaming environment should be designed in a relaxed and non-intrusive way. This is especially important during the early stages of the game. For this reason, in the design of the game ‘Smile 1’, we observed this during different stages and sections of the game.

In the game ‘Smile 1’, two cartoon characters of a girl and boy were used. The reason for this choice is to provide familiarity of the children with male and female genders throughout the game. For this reason, with the guidance of autistic experts, these cartoon characters were designed in exactly the same shape and format so that the child with autism would not be confused when identifying their characteristics from the very beginning. The ‘girl’ character in all sections has a scarf because, according to experts, children with autism only identify women as wearing scarves or tents (in Iran). On the basis of equal figures, four senses were designed on the faces in a way that is understandable for these children. You can observe these four modes on the first page of the game in Figure 4:

Figure 4: Four emotional states in the game ‘Smile 1’ on the first page of the game.

In the game entry section, we attempted to create familiarity with the faces by showing these characters to the child. It also includes childish music that does not hurt the autistic children’s hearing and is interesting and exciting. In figure 5, you can see the login screen:

Figure 5: Log into the game ‘Smile 1’.

For example, if the child chooses a ‘happy’ feeling, as shown in Figure 6, the following pictures are shown from left to right, which, by showing pictures, the social story of that stage is read to the child.

Figure 6: Learning the ‘Happiness’ feeling in ‘Smile 1’.

Of course, we attempted to create simple animations to attract the child's attention to various parts of the story.

For example, in the first two images, we showed show both happiness by a smile and a laugh to the child, and when he heard the sound of a laugh and the appearance of teeth while laughing, the child became more familiar with the ‘happy’ sense. Because, according to studies, children with autism recognise happiness and laughter most of the time by seeing teeth. In the third and fourth images, using simple stories and their related images by using the concepts of ‘mom and dad’ in stories and words, such as birthday cake, balloon, canvas, and painting, which children with autism are most often familiar with, the ‘happy’ feeling is taught to the children with autism. Because children with autism do not have correct understanding of many of the abstract concepts around them, in all the stories, elements are used in which the child has an objective conception of, and, in fact, they noticed these concepts in their surroundings.

All elements used in the game ‘Smile 1’ stories were chosen by field experts in this field.

After hearing the ‘happy’ stories, questions related to this sense are as it shows in figure 7.

Figure 7: Questions about the sense of ‘happiness’ in the game ‘Smile 1’.

One each slide, after the story was read to the child, she is now asked: ‘Now tell me what you look like when you're happy?’ Although the stories are first-person, according to studies that will be referred to in the following chapters, children with autism cannot understand first-person stories. Thus, this method of asking a question was used to make the child feel closer to the stories and identify their own sensations.

For these stories, the child's answers to questions and options from easy to difficult are used; thus, in the first question, only two options are available for the child, while in the third question, these choices are divided into four modes that contain all senses. Additionally, if the answer to the question is correct, a star will light up for the child to encourage the child.

As shown on the game pages, all the stories and questions can be repeated or paused. The child can also go back to the previous stages of the game and begin the process from a certain point.

Additionally, in this section, to help the participant, if the child does not choose the correct answer after 25 s from the end of the story, the correct option is displayed in a large size so that the child's attention is drawn to that option. At the stage of answering questions, if the child chooses the wrong option, sentences that indicate the wrong choice are read to the child.

However, to better measure the response rate and progress of the child at each stage, each question is scored as follows. 0: In this sense, the child did not have any cooperation when answering the questions. 1: In the sense, the child ultimately answers the question but has given an incorrect answer and did not reach the correct answer. 2: In this sense, the child after one or more incorrect answers finally found the correct answer. 3: In this sense, the child chooses the correct answer at the very beginning.

Measuring the effectiveness of ‘Smile1’

Before the experiments, we first reviewed the authoritative sources mentioned in previous sections, considering that teaching different emotional states to autistic patients generally reduces their behavioural problems, increases communication skills, improves conversational skills, and so on, we decided to investigate the effectiveness of ‘Smile 1’ on the behavioural problems of these children. To achieve this goal, after the studies, the Children Behavioural Check List (CBCL) was selected for comprehensive and complete analysis.

The CBCL is a common method for identifying behavioural problems in children. This checklist is one of the parts of the ASEBA system and was designed by Thomas M. Achenbach in 1966. This test is designed for three different age ranges of 1-5 years, 6-14 years and over 14 years of age. By performing this test, a total of 17 child-based behavioural syndrome scales can be identified in a child: 1) Anxiety / Depression, 2) Corruption / Depression, 3) Physical complaints, 4) Social problems, 5) Thinking problems, 6) Attention problems, 7) Behavioural law, 8) Aggressive behaviour, 9) Emotional problems, 10) Anxiety problems, 11) Physical problems, 12) ADHD problems, 13) Behavioural problems, Confrontation, 14) Conduct problems, 15) Internalisation, 16) Extrapolation and 17) General problems.

In the present study, a second type of questionnaire was used for children ages 6 to 14 years. This type of questionnaire includes 113 questions. After the analysis, all 17 factors mentioned for each participant were deduced from this test.

Based on the results obtained, according to relevant documentation, the percentage rating of the scales for children with normal growth is 93. Upon analysing the tests completed by the parents of the participants in the study, in all the children, most behavioural factors are rated above 93, which confirms the existence of multiple behavioural problems in these autistic children.

Participants

The experimental group included children tested using the designed game, whereas the control group included children who did not participate in the game intervention sessions and who only have previous training according to the previous routine.

Then, to test the game, a random sample of people with autism should be selected. Hence, in most similar research studies and activities that aim to determine the effectiveness of an intervention (in this study, Smile1 game) on a specific group that include a control group, 10 to 15 people are examined in each ‘Experimental ‘and’ Control’ group.

Additionally, to confirm this, many studies, such as the research by Ashayerih, and Esteki, (2017); Alves et al. (2013); (Miranda, et al. 2011); Silver and Oake (2001); and many others, have used the same number as a sample.

All of these children are students trained in the Autism Charity Foundation, and their level of autism has been determined by specialists from the very beginning. Therefore, in this study, based on the recognition of specialists in this field, we did not perform a separate test for autism disorder diagnosis of the sample group.

Because pre-test-post-test and control experimental groups were included in this study, after reviewing the resources related to the number of samples required in similar research, ten children with autism in the mild, moderate and severe categories of autism who were both male and female in the age group of 6 to 14 years were included in the experimental group, and ten children with autism matched the experimental group in terms of age, gender and severity of autism were included as the control group.

Ten children were selected in each group, including seven boys and three girls, and five of them had high-functional autism, four had moderate autism and one had severe autism of the classical type.

A. Intervention sessions

After selecting the sample of the test case, the game sessions of ‘Smile 1’ began. The number of sessions required to play the intended game for the experimental group was determined by the opinion of the project adviser. Additionally, to confirm the correct selection, in similar studies, such as in research by Alves et al. (2013); Silver and Oake (2001); Bölte et al. (2002) and many others, the number of sessions of intervention was similarly selected.

Thus, it was decided that the number of sessions for teaching children would be a maximum of 16 sessions of 30 min, or, if the child collaborated, ten sessions of 45 min per child. By starting the intervention and observing the extent of the child's cooperation, it was decided to test the game in 16 sessions of 30 min for each child. Accordingly, if the child learned all four senses earlier and answered the questions correctly, the number of those sessions was also reduced. However, if the child did not cooperate during the sessions, the training effort lasted up to 16 sessions.

In this study, due to time limits of the parents of the participants, these sessions were held at eight 1-hour sessions for each participant, and the first meeting for all participants was to provide the participant with familiarity with the game environment and related explanations. As a result, in the following sections, the seven main intervention sessions that were used for the game are examined.

For each session, depending on the child’s cooperation, stories were taught that focused on each sense, and then questions were asked of the child, and only when the game changed from one session to another was the child partially paying attention to the stories and questions related to the previous section. All of the observations were recorded throughout each session, and the child’s general demeanour during each session was also asked of his parents to justify the process during each session. Below are pictures of the children when performing the game:

Figure 8: Participants of the ‘Smile 1’ game.

Results of the ‘Smile 1’ game intervention sessions

During each session, the children’s responses to questions were identified and recorded. According to the description of the game after the last section, four feelings of ‘happiness, sadness, anger and fear’ were considered in this game. Each session tried to train all four senses and evaluate questions for each child. For each sense, three questions were asked in which, as previously stated, one of four (0, 1, 2, or 3) scores was assigned to each question.

In this way, the results recorded during the intervention sessions for one of the children are provided in Table 1 as an example.

Table 1: Scores earned by ‘Amir Ali’ for each question in the ‘Smile 1’ game.

Amir Ali, an 8-year-old boy with moderate autism, started his training 1.5 years ago. A lack of eye contact, difficulty with speaking and meaningful conversation, repetitive and stereotypical movements, and low focus and attention are his most important problems. The most troublesome problem of Amir Ali, as described above, has been detecting ‘angry’ feelings, although there has been difficulty with other senses.

Participant performances and feedback about the game and gamification elements

Additionally, during the sessions, the participant's performance and feedback related to the gamification elements that were considered in the game were recorded, as shown in Table 2.

Table 2: Participants' response table regarding the gamification elements.

In this table, the factors included: 1) the child's attention to the game page, 2) the child's attention to game music, 3) the child's attention to the positive feedback of the game, 4) the effectiveness of negative feedback, 5) child's attention to social stories, 6) child's understanding of the stories, 7) identifying the names of emotions (happiness, sadness, anger and fear), 8) the child's interest in the game's rating system (stars), 9) the effectiveness of guidance to the correct answer after the child’s choice, 10) the accompanying tutor in the response and 11) the child's perception of the first personal stories. These were ranked 0 or 1 for each child.

If the child did not show any interest in the factor or did not have any positive performance related to it, it was ranked as 0; it was ranked as 1 if the child was tangibly responsive to the factor involved and if it influenced him. In the next section, this table will be further examined.

Comparison of the CBCL test during pre-test and post-test stages

As discussed in the previous section, this research was performed using pre-tests and post-tests. As the first step, before starting the game intervention sessions, the parents of the children in both groups were asked to complete the CBCL test for their child.

In figure 9, the CBCL analysis of a participant in the pre-test phase is shown.

In the figure 9 below, two diagrams are shown; one is related to CBCL-based scales, and the second one refers to the overall score and child behavioural profiling based on DSM as prepared by ASEBA. As shown, the red line in both graphs shows the normal state line, which is equal to 93 for all scales. The blue line in this graph is related to the condition of the child tested at different scales, which is in most cases higher than 93 for this child with moderate autism.

Figure 9: CBCL analysis of ‘Amir Ali’ - Pre-test phase.

After the end of the intervention sessions, the parents of the children in both two groups were again asked to complete the Child Behavioural Checklist questionnaire to compare the results of these tests during the pre-test and post-test stages to determine the effectiveness of intervention on reducing the behavioural issues of the participants.

As an example of the analysed questionnaires in the post-test, figure 10, shows the behavioural diagrams related to the child ‘Amir Ali’.

Figure 10: CBCL analysis of ‘Amir Ali’ - Post-test stage

Upon comparing the results of the pre-test and post-test for Amir Ali, there were very few changes.

Supplemental intervention sessions to measure training durability

After the end of intervention sessions, with a two-month interval, the game was performed on three selected children during four additional 1-h sessions to evaluate its effectiveness and durability. For this purpose, ‘Khojaste’, ‘Amir Ali’, and ‘Danial’ were selected. The reason for selecting these three children was their cooperation during the intervention sessions and the appropriate level of their autism and their development during the main game sessions. In the tables below, you can see the results in these sessions separately.

According to Table 3, ‘Khojasteh’ and ‘Danial’, who had high-functional autism and showed good performance during the main intervention sessions also completed the stages well in these sessions. Although in the first two sessions, due to the weakness of short-term memory in autistic patients, in some cases, incorrect answers were received from these two. For Amir Ali, because he has a moderate level of autism and had a good but slow progression during the main intervention sessions, during these four sessions during a 2-month interval, he initially responded to some questions. It was difficult though owing to his familiarity with the game environment, and we observed much more cooperation and faster progress.

Table 3: List of scores for the three selected participants in additional intervention sessions over a 2-month interval.

However, instead of understanding the concepts, because patients with autism mostly (except those with high-functional autism who are similar to ordinary children) only remember things that they have heard, the effectiveness of short-term interventions may last like they should. In fact, according to an autism therapist, Mrs. ‘Motie’, teaching one subject to an autistic child requires at least 4 to 5 months. It is important to with repeat the topic again and again to complete the learning process so that the autistic person does not forget it.

Consequently, because the present study was conducted over a period of 2 months due to time constraints and the difficulty coordinating time with the children's parents, it could not be expected that the learning process would be prolonged or significant, although the main intervention sessions had very positive and constructive effects considering the children’s backgrounds.

Each participant's progress divided by senses for all sessions

In the previous section, the results and scores of the Smile 1 game sessions were shown as an example for one of the participants in the experimental group. The diagram below shows the progress of this participant during each intervention session for the four ‘happy, sad, angry and afraid’ senses.

‘Amir Ali’ showed slow but noticeable progress during the intervention sessions. From the first session, he found it difficult to identify any of the four senses in the study. Of course, one of the reasons he had inappropriate performance during the first intervention session was his lack of cooperation due to the unfamiliar environment and the lack of anonymity of the coach, which led to a sharp decrease in his focus and increasing behavioural problems during the first session. However, as shown in the diagram, Amir Ali advanced well by the final intervention (seventh), and ultimately, he had the ability to recognise all four senses taught in the game.

As discussed in the previous section, ‘Amir Ali’ is one of three individuals who participated in four additional intervention sessions after a 2-month break to assess the durability of training. In the figure above, these four sessions (8, 9, 10 and 11) are shown. As shown in the figure 11, Amir Ali, due to his serious problems with focus and attention and his weak memory, which exists in most people with autism, his score decreases in the eighth session, although this drop was much better than during his first main intervention session, and certainly, the seven previous sessions created a good background for him. Of course, from the very beginning, ‘Amir Ali’ had trouble recognising the ‘anger’ feeling, which was also observed in four additional sessions. However, according to the figure 11, his progress in these four final sessions was very fast, and we ultimately achieved the desired results.

Figure 11: Separate senses before all meetings for ‘Amir Ali’.

Overall progress of each participant in all intervention sessions

Below show the diagrams for the general development of one child as an example for all the sessions; however, the factors of game and gamification and the child's attention to these factors are taken into account, and its relevance to the child's performance was examined.

Although at the end of the seven main intervention sessions Amir Ali had trained all four senses and answered the questions about them, based on observations at each session and as shown in the above table, although he was paying attention to the related stories, he had no understanding of these stories and did not know the names of some feelings because of the inability to speak.

This is also why Amir Ali, in an intervention session after a two-month hiatus, initially, we observed a slight weakness in his performance regarding remembering names of the senses and related stories and a weakness in short-term memory. However, this weakness was quickly recouped at subsequent sessions. However, Amir Ali was one of the few participants in the study that had negative feedback of the game and more or less considered it when responding to questions, and he was interested in the game's scoring system.

Comparing the overall progress of all participants during the main intervention sessions

When the status and progress of each child during the game intervention sessions was identified, we were able to make comparisons between the performances of all participants in the experimental group during the seven main intervention sessions based on the average responses to the four desired senses, as shown in the figure 13.

Figure 12: Overall progress of ‘Amir Ali’ during all intervention sessions.

Figure 13: Session Development Averages of all senses per each child.

As shown in figure 13, ‘Mohammad Ali’ and ‘Amir Reza’, who answered all the questions correctly, have a linear number of 3. However, ‘Shervin’, ‘Danial’, ‘Parsa’ and ‘Khojasteh’, who showed good performance during the intervention sessions, have numbers between 2.5 and 3 in the figure 13 above, which indicated a good performance. Their progress was good during the intervention sessions. The green line represents ‘Amir Ali’, who, despite moderate performance during the first intervention session, was very good at subsequent sessions. Additionally, there are two lines with very high volatility that represent ‘Amir Hossein’ and ‘Setare’ (crescent and orange lines), which, despite recessions during some of the intervention sessions, made good progress during the final sessions. In the lower part of the figure 13, the brown line represents ‘Yeganeh’, who has severe autism and did not cooperate properly during the intervention sessions.

Comparison of participants' progress during the sessions according to their level of autism

As shown in figure 14, a comparison was made between the participants' progress during the main intervention sessions by distinguishing their level of autism.

Figure 14: Comparison of progress during the main intervention sessions according to the autism levels of the participants.

Based on the figure 14 above, participants with high-functional autism showed less but steady progress than the others. Overall, however, participants with moderate autism progressed more rapidly, although this was more unpredictable. According to previous studies, the main cause of these fluctuations is the disturbance of focus and attention as well as the behavioural and communication problems of these participants, which caused their lack of cooperation during some intervention sessions. In this study, there was no progression or hope for a participant with severe autism (Bhatt, et al. 2014); (Smith, et al. 2010). although this result cannot be generalised to all patients with severe autism. Those who suffer from severe autism have very low levels of education that make learning difficult, time-consuming, very slow and sometimes impossible.