Related Experiment Video
Updated: Aug 26, 2025
Virtual Reality Tools for Assessing Unilateral Spatial Neglect: A Novel Opportunity for Data Collection
Published on: March 10, 2021
Developing attention deficits/hyperactivity disorder-virtual reality diagnostic tool with machine learning for
Tjhin Wiguna1, Raymond Bahana2, Bayu Dirgantoro2
1Faculty of Medicine, Universitas Indonesia - Dr. Cipto Mangunkusumo General Hospital, Jakarta, Indonesia.
This study created a new digital tool for diagnosing ADHD in young people. By combining virtual reality environments with artificial intelligence, researchers developed a system that translates standard clinical symptoms into observable classroom behaviors. This approach offers a more objective way to assess attention and hyperactivity compared to traditional interviews.
Area of Science:
- Pediatric mental health research within ADHD diagnostic medicine
- Computational psychiatry and machine learning integration
Background:
No prior work had resolved the limitations inherent in traditional subjective assessments for pediatric neurodevelopmental conditions. Current diagnostic practices rely heavily on parent-child interviews and observational reports. This reliance often introduces bias and variability into clinical evaluations. That uncertainty drove the need for objective, technology-driven alternatives. Virtual reality offers immersive environments that simulate real-world settings for patients. Such platforms allow for the standardized measurement of behavioral responses. This gap motivated the development of digital tools that align with modern clinical standards. Researchers sought to bridge the divide between standardized diagnostic criteria and digital assessment capabilities.
Purpose Of The Study:
This study aimed to develop a comprehensive diagnostic framework for attention deficits and hyperactivity disorder using virtual reality. The researchers sought to address the limitations of traditional, interview-based diagnostic methods. They intended to create a digital tool that reflects the requirements of the current technological era. The project focused on constructing a research domain construct based on standard diagnostic criteria. The team aimed to integrate this construct with machine learning applications to produce an intelligent diagnostic model. They wanted to enable the system to process multifaceted clinical symptom data effectively. The study also sought to expand algorithmic models to generate more accurate diagnostic predictions. Finally, the researchers aimed to establish a foundation for future digital diagnostic innovations in clinical practice.
Main Methods:
The researchers conducted an exploratory qualitative study across two distinct phases. Phase one utilized the Delphi technique to reach expert consensus on behavioral manifestations. Three rounds of consultation ensured that clinical criteria were accurately translated into observable actions. Phase two involved over ten focus-group discussions to finalize the research domain construct. The team employed conceptual content analysis to process the qualitative information gathered. Microsoft Excel for Mac facilitated the initial data organization and analysis. This approach allowed for both manifest and latent interpretation of expert feedback. The final prototype development required integrating these findings into a functional programming framework.
Main Results:
The study identified 13 specific student behaviors that reached a consensus threshold above 75% among experts. These behaviors formed the basis for the three primary domains and six sub-domains of the diagnostic construct. The resulting model incorporates reward-related processing, emotional lability, and inhibitory control. It also includes sustained attention, specific timing of playing, and arousal as key metrics. The research successfully transformed these clinical concepts into activities suitable for virtual reality programming. The team established a clear link between standard diagnostic criteria and digital behavioral tasks. This structure enables the machine learning application to receive and analyze complex clinical input. The final output provides a predictive model designed to improve diagnostic accuracy for children.
Conclusions:
The authors propose that their constructed framework serves as a foundational reference for future digital diagnostic development. This study demonstrates that expert consensus can successfully translate clinical criteria into measurable virtual behaviors. The researchers suggest that integrating machine learning with virtual reality enhances the precision of behavioral predictions. These findings indicate that a multi-domain approach captures the complexity of pediatric attention disorders. The team maintains that their six-domain model provides a comprehensive structure for programming future diagnostic prototypes. They emphasize that the Delphi technique ensures high levels of expert agreement on behavioral indicators. The study concludes that this methodology supports the creation of intelligent models for clinical use. Future investigations may build upon these specific domains to refine diagnostic accuracy in diverse populations.
Frequently Asked Questions
The researchers propose a six-domain model including reward-related processing, emotional lability, inhibitory control, sustained attention, specific timing, and arousal. This structure allows the machine learning algorithm to process multifaceted clinical data into a predictive diagnostic output.
The team employed the Delphi technique to achieve expert consensus on behavioral indicators. This method involved three rounds of consultation to translate standard diagnostic criteria into concrete, observable actions suitable for a classroom-based virtual environment.
The study required three rounds of Delphi consultation followed by over ten focus-group discussions. These steps were necessary to finalize the research domain construct before the programming phase of the prototype could commence.
The researchers utilized conceptual content analysis with a manifest and latent approach. This qualitative method allowed the team to synthesize expert perceptions into the final six-domain structure used for the machine learning prototype.
The first stage identified 13 specific student behaviors that achieved a consensus rate exceeding 75% among the participating experts. These behaviors serve as the foundation for the virtual reality tasks.
The authors suggest that this framework provides a reference for future studies aiming to integrate machine learning with virtual reality. They claim this approach offers a more accurate diagnostic value for children and adolescents.
