Does Artificial Intelligence-generated Image Exposure before the Treatment Reduce Dental Anxiety in Children Aged 6–12 Years: An Observational Study

Abstract

Introduction

Children commonly feel dental anxiety, and the efficacy of treatment is contingent upon the dentist's skill and capacity to establish a productive rapport with the patient.

Aim

To evaluate the effect of anxiety before the extraction procedure by showing artificial intelligence (AI)-generated images and similar non-AI-generated images.

Materials and methods

This observational study, conducted over a 2-month period, involved 30 children between 6 and 12 years of age. The children were divided into two groups, one exposed to non-AI-generated images and the other to AI-generated images, for 4 minutes in the waiting area, who required the extraction. Anxiety levels were measured before and after the image exposure using a pulse oximeter and the Raghvendra, Madhuri, and Sujata digital anxiety scale (RMS-DAS scale).

Results

The pulse rate between the non-AI and AI groups was insignificant; however, RMS-DAS scores were significant (p = 0.001). Intergroup comparison between groups for pulse rate was insignificant, whereas a significant difference was noted in RMS-DAS scores (p = 0.013). Non-AI positively correlated with pulse rate after the procedure and negatively correlated with RMS-DAS after showing AI images.

Conclusion

Both AI and non-AI images demonstrated efficacy in relieving dental anxiety before the dental extraction; however, artificial intelligence images exhibited more reduction in dental anxiety.

How to cite this article

Davangere Padmanabh SK, Dagli ST, Bargale S. Does Artificial Intelligence-generated Image Exposure before the Treatment Reduce Dental Anxiety in Children Aged 6–12 Years: An Observational Study. Int J Clin Pediatr Dent 2025;18(8):986–991.

Introduction

In dentistry, the effectiveness of treatment derives from the dentist's competence and ability to create an effective relationship with the patient. Fear of dentistry in children is common. Kuhn's findings indicated that approximately 22% of children evaluated by pediatric dentists exhibit significant management challenges.¹

The American Academy of Pediatric Dentistry has drawn several behavior management approaches to use in children, such as tell-show-do, voice control, distraction, positive reinforcement, physical restraint, nonverbal communication, conscious sedation, and general anesthesia. Many approaches have been suggested for application in children when anxiety disrupts dental treatment.² Images may be similar to a simple version of “modeling”, successfully used to decrease dental anxiety in children.

One strategy could involve cultivating favorable connections with dentistry by showcasing favorable depictions of youngsters undergoing dental treatment. This approach utilizes social learning principles, suggesting that exposure to positive images will facilitate the formation of an association between positive imagery and dentistry, resembling the concept of modeling.³ Recent evidence demonstrates the impact of model observation on managing dental anxiety in pediatric patients.^4,5

Pediatric dentists aim to prevent, eliminate, or minimize anxiety in young patients, as this emotional state can adversely affect the quality of dental care and the child's overall oral health.⁶ Recent evidence indicates that modeling effectively reduces anxiety in pediatric dental patients. The method entails a child observing another child undergoing dental treatment via direct observation, video, or photographs.⁵

Artificial intelligence (AI) presents considerable potential in pediatric dentistry, enhancing diagnosis and treatment planning, facilitating image analysis, and improving communication with patients and parents.⁷ The emergence of various AI image generators such as Copilot, Dall-E3, and Gemini, capable of producing images that are indistinguishable from natural ones, opens up exciting possibilities for reducing dental anxiety in children. AI image generators employ artificial intelligence algorithms to produce images based on textual descriptions, a process that consists of three steps. In the first step, the AI is trained on the large set of images and their associated text. The AI learns the patterns and style in the data, followed by learning to predict new images based on the text prompt. Our study's null hypothesis was that there is no variance in the reduction of dental anxiety measured by pulse oximeter and the Raghvendra, Madhuri, and Sujata digital anxiety scale (RMS-DAS scale) after showing non-AI and AI-generated images. This study intended to evaluate the outcome of anxiety before dental treatment by showing AI-generated images and non-AI-generated images.

Materials and Methods

The observational study was conducted for 2 months to appraise the effect of exposure to AI and non-AI images on dental anxiety among 6- to 12-year-olds attending the outpatient department of pediatric and preventive dentistry. Written informed consent was taken before starting the study. As per the Helsinki Declaration, the Institutional Ethics Committee clearance was obtained CODS/IEC/230/2024, and our study was registered with the clinical trial registry of India CTRI/2024/07/069768 before starting the study.

Sample Selection

The sample size was derived by comparing two means from using Epi Info version 3.01, the difference in the group means as per the previous study by Gangwal et al.⁸ A sample size of 30 participants was obtained with 80% power and 95% confidence interval. The inclusion criteria were children of both genders aged 6–12 years who required extraction. The exclusion criteria included children with visual impairment, mentally retarded children, and parents who refused to give consent to enroll in the study.

Methodology

A total of 30 children were split into two groups of 15 children each. Group A consisted of non-AI-generated images (control group). Group B consisted of AI-generated images (experimental group). Block randomization was employed in our study to create an allocation list for two comparison groups. A random number generated by the computer was employed to decide on random permuted blocks with a four-block size and an equal allocation ratio. Allocation concealment was achieved using pressure-sensitive paper within the envelope to convey participant information to the designated group. The lead investigator created the random allocation sequence, enlisted the participants, and randomized them to research groups. Neither the study investigators, participants, nor data analysts were blinded, as it was an open-label study design (Fig. 1).

Fig. 1:

Study enrollment flowchart (STROBE: strengthening the reporting of observational studies in epidemiology)

In group A, non-copyright images were selected from a Google image search on July 1st, 2024, between 09:30 am and 11:00 am Indian standard time, consisting of eye-soothing natural images such as waterfalls, valleys, forests, snow, and sky (Fig. 2). Images were revealed to the children in the waiting area who required extraction, and they were allotted 4 minutes to view the images; the coinvestigator measured the time duration using a stopwatch.

Fig. 2:

Non-AI-generated images of waterfalls, valleys, forests, snow, and sky

Similarly, in group B, AI-generated images from Microsoft copilot were obtained by commands such as create high resolution 3D realistic image of waterfall which was eye-soothing; create high resolution 3D realistic image of consisting valley which was eye-soothing; create high resolution 3D realistic image of consisting forest which was eye-soothing; create high resolution 3D realistic image of consisting snow which was eye-soothing and create high resolution 3D realistic image of consisting sky which was eye-soothing (Fig. 3) on July 1st, 2024, between 11:30 am to 1:00 pm Indian standard time. AI and non-AI images were sent to ten pediatric dentists for the content validation across the length and breadth of India. AI and non-AI images were selected on the basis of the nature theme. Images were revealed to the children in the waiting area who required extraction, and they were given four minutes to view the images; even in this group, the coinvestigator measured the time duration using a stop clock. In both groups, anxiety was assessed before and after the visualization of the images with the help of the pulse oximeter and the RMS-DAS scale.⁹

Fig. 3:

AI-generated images of waterfalls, valleys, forests, snow, and sky

Statistical Analysis

Statistical analysis was conducted utilizing the Statistical Package for the Social Sciences (SPSS) software (IBM Corp, version 21.0). Descriptive statistics were conducted for the various parameters evaluated in the study. An independent samples t-test was conducted to evaluate significant differences between the two groups prior to and following the procedure. The paired t-test was employed to evaluate significant differences between the two groups. A p-value below 0.05 was deemed statistically significant.

Results

The mean age of children in group A was 9.73 years, while in group B it was 9.53 years. Group B exhibited a higher female representation at 73.3%, in contrast to group A, which had 53.3% females. In group A, the mean pulse rate (bpm) was 99.2 before and 104.33 after; in group B, the mean pulse rate (bpm) was 105.26 before and 104.13 after. The data indicate an increase in pulse rate following the procedure in both groups (Table 1). The mean RMS-DAS scores for group A before and after displaying non-AI images were 3.0 and 2.06, respectively, while for group B, the scores were 3.06 and 1.46, respectively (Table 2). Statistical analysis using a paired t-test revealed significant results exclusively for RMS scores in both groups when comparing pulse rate scores and RMS-DAS scores (Table 3).

Table 1:

Descriptive statistics of pulse rate scores in different groups at different time intervals

Groups	Parameter and time interval	Minimum	Maximum	Mean	Std. deviation	p-value
Group A	Pulse rate (bpm) before	76.00	138.00	99.20	14.75	0.156
	Pulse rate (bpm) after	81.00	135.00	104.33	14.06
Group B	Pulse rate (bpm) before	78.00	126.00	105.26	12.12	0.794
	Pulse rate (bpm) after	75.00	143.00	104.13	20.99

Table 2:

RMS-DAS scores before and after showing non-AI and AI images in different groups at different time intervals

Groups	Parameter and time interval	Minimum	Maximum	Mean	Std. deviation	p-value
Group A	RMS-DAS before showing non-AI images	2.00	4.00	3.00	0.53452	0.001
	RMS-DAS after showing non-AI images	1.00	3.00	2.06	0.59362
Group B	RMS-DAS before showing AI images	2.00	4.00	3.06	0.59362	0.000
	RMS-DAS after showing AI images	1.00	3.00	1.46	0.63994

Table 3:

Before and after comparison of the Pulse rate scores and RMS-DAS scores in different groups

Before and after comparison		Mean difference	t-value	p-value
Group A	Pulse rate (bpm) before vs after	−5.13333	−1.501	0.156
	RMS-DAS before showing non-AI images vs after	0.93333	6.089	0.001
Group B	Pulse rate (bpm) before vs after	1.1333	0.266	0.794
	RMS-DAS before showing AI images vs after	1.60000	8.411	0.000

An independent samples t-test was performed to compare pulse rate scores and RMS-DAS scores across different groups at various time intervals. Statistically significant results were observed between the groups following the procedure for RMS-DAS scores (Table 4). Statistical analysis revealed a significant correlation between age groups and pulse rate scores, as well as between age groups and RMS-DAS scores, following the procedure (Table 5). No statistically significant results were obtained when examining the correlation between pulse rate scores and gender, as well as RMS-DAS scores and gender across various groups at different time intervals (Table 6).

Table 4:

Intergroup comparison of the pulse rate scores and RMS-DAS scores between different groups at different time intervals

Comparison groups	Parameter and time interval	Mean difference	t-value	p-value
Group A vs Group B	Pulse rate (bpm) before	−6.067	−1.231	0.229
	Pulse rate (bpm) after	0.3000	0.031	0.976
	RMS-DAS before	−0.06667	−0.323	0.749
	RMS-DAS after	0.60000	2.662	0.013

Table 5:

Correlation between pulse rate scores and age groups, and RMS-DAS scores and age groups in different groups at different time intervals

Age		Pulse rate (bpm) before	Pulse rate (bpm) after	RMS-DAS before	RMS-DAS after
Group A	Pearson Correlation	0.217	0.524	0.313	0.300
	p-value	0.436	0.057	0.257	0.277
Group B	Pearson Correlation	0.352	0.009	−0.545	−0.227
	p-value	0.198	0.006	0.416	0.036

Table 6:

Correlation between pulse rate scores and gender, and RMS-DAS scores and gender in different groups at different time intervals

Gender		Pulse rate (bpm) before	Pulse rate (bpm) after	RMS-DAS before	RMS-DAS after
Group A	Pearson correlation	0.491	0.170	0.259	0.109
	p-value	0.063	0.544	0.352	0.700
Group B	Pearson correlation	0.245	−0.249	−0.456	−0.276
	p-value	0.378	0.371	0.088	0.319

Discussion

“Dental anxiety denotes a state of apprehension that something dreadful is going to happen in relation to dental treatment, and it is coupled with a sense of losing control.” Since dental anxiety is thought to be a universal issue, it has piqued the interest of numerous researchers such as Aartman et al.,¹⁰ Ramos-Jorge et al.,¹¹ Olumide et al.,¹² Gustafsson et al.¹³

Diverse techniques for evaluating dental anxiety are being mentioned in literature, such as direct and indirect methods. Direct methods included blood pressure, pulse rate, and muscular tension. The indirect method necessitates proficiency in administering scoring assessments. There are several tools and scales available to measure children's dental anxiety, such as Venham's picture test, RMS scales, Facial image scale, Zung self-rating anxiety scale, Jeet wheel scale, Modified Venham's Picture Test, Animated emoji scale, Chotta Bheem and Chutki scale, Modified child dental anxiety scale, and Modified child dental anxiety scale faces. A pediatric dentist can better manage patient behavior and make appropriate behavior guidance or management decisions by measuring dental anxiety and considering the various anxiety levels of their patients.

The term artificial intelligence (AI) denotes a broad spectrum of technical advancements that continue to impact our day-to-day activities. The development of AI makes it possible to analyze vast volumes of data, providing reliable information that improves decision-making.⁸ In our study, we have used the RMS-DAS scale developed by Shetty et al.,⁹ since it can be used in the age group between 6 and 12 years, because it allows children to self-report anxiety in children who have cognitive ability. Tiwari et al.¹⁴ has recommended that RMS, Facial Image Scale, and Venham's Picture Test are easy to administer, equally effective in children, and are less time-consuming, and can be employed by researchers and clinicians in calculating dental anxiety in children.

A study conducted by Fux-Noy et al.¹⁵ concluded that dental anxiety inside a Snoezelen setting decreased the anxiety levels prior to subsequent dental treatment. In our study, the initial recording of the anxiety level was done in the waiting area, which was silent and had a fully covered partition. In the present study, the anxiety level of children in waiting area was found to be highly similar to the study done by Gangwal,⁸ Ramos-Jorge et al.¹¹ and Arangannal and LN,¹⁶ this could be due to a close connection between family dental fear and the development of dental anxiety in children.¹⁷ In the current study, it was shown that both non-AI and AI images exhibited a significant reduction in dental anxiety, as measured by RMS-DAS scores. This finding aligns with previous research conducted by Fox and Newton,² Gangwal et al.,⁸ Ramos-Jorge et al.¹¹; however, a study conducted by Kamel¹⁸ presents contrasting results, did not found any reduction in the anxiety level.

In our study, we have selected Copilot over the other image creators for image generation, as it was very popular in the month of July 2024, as per the analytics. To standardize and match the image obtained from the AI as well as Google image search, was sent to ten pediatric dentists of a private dental institution and who practice exclusive pediatric dentistry. The efficacy of AI-generated images in easing anxiety appeared to surpass that of non-AI images, mostly attributed to the inherent characteristic of AI-generated images to diminish vibrant colors and disparities in contrast, this could be the reason for the difference between the two groups. The pulse rate in the AI group reduced because of the soothing nature of the AI-generated images. The present study provides evidence to disprove the null hypothesis, representing that the AI-generated images were more effective in reducing anxiety than the non-AI-generated images as measured by the RMS-DAS scale. In the present study, the age group was from 6 to 12 years because of their ability to think in a concrete way. In our study, AI and non-AI images were printed on photographic paper and not from any electronic devices in order to avoid electronic screen time among the subjects who belong to Generation Alpha.

The Gangwal et al.⁸ study indicates that positive dental images have a distinct effect compared to neutral images in alleviating anxiety, as assessed by the VPT. Consequently, it was concluded that displaying visuals pertinent to the dental environment effectively alleviated dental anxiety in children. Kamel et al.¹⁸ found that exposure to positive or neutral dental images did not affect the anticipatory anxiety levels in children. Nonetheless, displaying either category of images has proven to be an effective approach for enhancing behavior in children. According to Ramos-Jorge et al.,¹¹ positive dental images showed no significant difference from neutral images in alleviating anxiety, as assessed by VPT. It may be extrapolated that displaying images in photographs, particularly in a dental context, effectively alleviates anxiety in children. Fox and Newton² asserted that exposure to pleasant imagery of dentistry and dentists leads to temporary decreases in anticipatory anxiety in children.

The limitation of the present study is that data were collected at a single time point in the trial, specifically after the intervention, when children had examined the images. The present study did not attempt to recognize the procedure through which anxiety reduction occurred. The anxiety was not recorded following the extraction procedure.

Conclusion

From the results of our study, it can be concluded that non-AI and AI-generated images showed effective results when the RMS-DAS scale scores were assessed individually in the groups. AI images showed a positive effect on extraction procedures by reducing children's dental anxiety. Both artificial intelligence (AI) and non-AI images demonstrated efficacy in relieving dental anxiety; however, artificial intelligence images exhibited more reduction in dental anxiety.

Statement of Author Contributions

SK and SD conceived the ideas; SK and SD collected the data. SD and SB analyzed the data; and SD and SB led the writing.

Orcid

Shital Kiran Davangere Padmanabh https://orcid.org/0000-0003-2896-8446

Shrushti Tusharkumar Dagli https://orcid.org/0009-0006-4592-281X

Seema Bargale https://orcid.org/0000-0000-0000-0002