BendableSound: Pioneering Haptic Interface for Neurodevelopmental Research
BendableSound: Pioneering Haptic Interface for Neurodevelopmental Research
Embedded Files
Understanding and measuring the unique interaction patterns of children with Autism Spectrum Disorder (ASD) presents significant challenges in early detection and personalized support. Traditional assessment methods often lack the nuance to capture subtle differences in sensory interaction.
Ivonne Monarca, Researcher
September 2017-September 2019
Research | Design | Evaluation with children | Recruitment | Data collection | Quantitative analysis| Machine Learning
Key learnings
Key learnings
Interaction Pattern Analysis
Interaction Pattern Analysis
Successfully identified distinct interaction patterns between ASD and NT children
Discovered three primary digital markers of ASD:
Variation in force usage
Space utilization
Temporal interaction characteristics
Demonstrated the power of interdisciplinary collaboration
Showcased innovative use of technology in understanding neurodevelopmental differences
Highlighted the importance of user-centered design in research technology
Research Goals
Research Goals
Design an innovative passive haptic interface for touch interaction analysis
Design an innovative passive haptic interface for touch interaction analysis
Develop machine learning models to distinguish ASD and neurotypical (NT) interaction patterns
Develop machine learning models to distinguish ASD and neurotypical (NT) interaction patterns
Explore novel digital markers for neurodevelopmental assessment
Explore novel digital markers for neurodevelopmental assessment
Methodology
Methodology
Collaborative Design Process
Collaborative Design Process
🗓️ 12 semi-structured interviews
🎨 🖼️ 10 design sessions
⏳ 21.5 hours of passive observation
🧑🤝🧑👭 A diverse group of HCI experts, children, Psychologists, Musicians, etc.
The data obtained during these design sessions were analyzed using rapid contextual design techniques. Subsequently, we translated this data into sketches, storyboards, and new ideas that were incorporated into the initial activities.
Study Design
Study Design
Participants:
39 neurotypical children
22 children with ASD
Age range: 3-6 years
Data Collection:
3 separate sessions
Multiple institutional settings in Northern Mexico
Total of 21,354 touch interactions recorded
Data analysis
Data analysis
Total Interactions: 21,354 touch interactions recorded
Feature Extraction: 11 general features describing geometry and execution
Analysis Approach:
Feature reduction and selection
Exploratory analysis
Statics analysis
Classification Task: Binary distinction between ASD and NT touch interactions
Key Findings
Key Findings
Digital markers
Digital markers
Variations in the use of strength.
Variations in the use of strength.
Children with ASD tend to use less strength than NT children when interacting with an elastic display.
Smaller gestures.
Smaller gestures.
NT children perform broader gestures than children with ASD.
More time when making a gesture.
More time when making a gesture.
Children with ASD touch the surface longer when making a gesture than NT
Machine Learning Performance
Machine Learning Performance
Classification Precision: 97.2%
Recall: 94.6%
Demonstrated high accuracy in distinguishing ASD touch interactions
Impact and Implications
Impact and Implications
Pioneered a technology-driven approach to neurodevelopmental assessment
Provided objective, quantitative insights into ASD interaction patterns
Opened new possibilities for early detection and personalized intervention
Technical Toolkit
Technical Toolkit
Programming Languages: Python, Processing
Research Methodologies:
User-Centered Design
Machine Learning
Quantitative Data Analysis
Personal Contributions
Personal Contributions
Led interdisciplinary design and research process
Developed feature extraction methodology
Implemented machine learning classification models
Conducted comprehensive data analysis
Coordinated with multiple stakeholders (experts, institutions, participants)
Future Research Directions
Future Research Directions
Refine digital marker identification
Expand interface applications
Develop more comprehensive neurodevelopmental assessment tools
Publications and Further Reading
Publications and Further Reading
Monarca, I., Cibrian, F.L., Chavez, E. et al. Using a small dataset to classify strength-interactions with an elastic display: a case study for the screening of autism spectrum disorder. Int. J. Mach. Learn. & Cyber. (2022). https://doi.org/10.1007/s13042-022-01554-2
Monarca, I., Tentori, M. & Cibrian, F.L. Understanding the musical interaction of children with autism spectrum disorder using elastic display. Pers Ubiquit Comput (2023). https://doi.org/10.1007/s00779-022-01703-y
Page updated
Google Sites
Report abuse