Standard therapy often fails people whose brains work differently. 

Like those with autism. 

Meet Elias. He’s 22, has been diagnosed with Autism Spectrum Disorder (ASD) and Social Anxiety Disorder (SAD), and faces executive dysfunction, sensory overload, and communication barriers. Daily tasks overwhelm him. Unpredictable settings make talking about feelings nearly impossible. And traditional therapy hasn’t helped.

In February 2025, MyndStories, along with Zensible, set out to explore the possibilities for better support for neurodivergent individuals.

Together, we launched the “Innovation Challenge for Therapy & Technology” – a call for students to turn their ideas into a short paper. With a focus on research-backed, tech-integrated therapeutic approaches, the challenge encourages participants to refine their concepts and think about how digital tools, AI, or other advancements could make mental health support more accessible and effective for neurodiverse communities.

In March, 2025, we selected 3 winners. We already published the paper that won the 1st prize and the second prize. Now, we’re proud to publish the final and third prize winner here from Shruthi Suresh.

Shruthi explores how a neuro-adaptive intervention, which involves AI-driven emotional recognition, Quantum Neural Network (QNN) sentiment decoding, and Extended Reality (XR)-based interactive simulations can foster a structured yet flexible environment.

This model provides an innovative, personalized, and practical therapeutic approach for neurodivergent individuals, which can enhance their quality of life.

The paper has been reviewed by Soumya Choudhary, Aparna Divakar, and Tanmoy Goswami.  

Introduction

Neurodivergence encompasses a range of cognitive variations, including ASD, ADHD, dyslexia, and Tourette syndrome. Individuals with neurodivergence often struggle with traditional therapeutic models, which may not accommodate their unique processing styles. 

The rise of AI-driven therapy offers unprecedented potential for personalized, adaptive interventions. This paper explores how QNN-enhanced sentiment analysis and XR therapy can revolutionize neurodivergent mental health treatment by improving accessibility, engagement, and therapeutic outcomes. It also examines the practical integration of these technologies into existing mental health ecosystems while addressing ethical, accessibility, and implementation challenges.

The need for adaptive therapy solutions is evident in recent studies indicating that traditional talk therapy is often ineffective for neurodivergent individuals due to sensory processing differences and atypical cognitive patterns (Gómez et al., 2021). These individuals often require structured, predictable, and multi-modal interventions that align with their unique cognitive and emotional processes.

The application of AI-driven sentiment analysis and XR-based simulations holds immense promise in bridging these gaps, offering a more engaging, responsive, and data-informed approach to therapy. By leveraging real-time emotion recognition and immersive virtual environments, these technologies create tailored interventions that dynamically adjust to an individual’s needs. 

However, successful implementation requires rigorous ethical standards, transparency in AI decision-making, and adaptability to individual client needs. Addressing concerns related to data privacy, algorithmic bias, and accessibility in XR platforms is crucial to ensuring equitable and effective mental health care for neurodivergent populations.

Case Study: Elias’s challenges and needs

Elias, a 22-year-old with ASD and SAD, experiences:

• Executive dysfunction – Difficulty with organization, task initiation, and emotional self-regulation.
• Sensory overload – Hypersensitivity to sounds, lights, and social environments, leading to avoidance behaviours.
• Social communication barriers – Struggles with nonverbal cues and emotional expression, impacting his ability to build relationships.
• Anxiety and avoidance – Prefers predictable, structured environments, making traditional in-person therapy overwhelming and ineffective.

Despite multiple therapy attempts, Elias finds it difficult to articulate emotions in real time and struggles with cognitive overload in unstructured therapeutic settings. His difficulty in recognizing and expressing emotions leads to miscommunication and frustration, further intensifying his social anxiety. 

Elias’s experience aligns with broader trends in neurodivergent mental health. Research shows that approximately 40% of autistic adults experience significant anxiety due to sensory hypersensitivity and executive dysfunction (Gómez et al., 2021). These challenges often create a cycle of avoidance and distress, making it difficult for individuals to engage in therapy effectively. 

Understanding these barriers highlights the necessity for interventions tailored to neurodivergent cognitive and emotional processes.

Furthermore, studies indicate that traditional therapy settings, which often involve spontaneous verbal communication, can be counterproductive for individuals with ASD, who may struggle with real-time information processing and prefer visual or written communication.

To address these challenges, a therapeutic model integrating QNN-enhanced sentiment analysis and XR-immersion therapy offers a promising alternative. By providing structured, sensory-friendly, and AI-assisted environments, this approach can facilitate emotional expression, reduce cognitive overload, and enhance engagement. Such personalized interventions could bridge the gap between therapeutic intent and effective implementation, ensuring neurodivergent individuals receive support tailored to their specific needs.

Proposed intervention

Quantum Neural Network (QNN)-enhanced sentiment analysis

QNNs, an advanced form of AI, analyse Elias’s language patterns and physiological signals to interpret emotional states in real time. This provides:

• Enhanced emotional expression – AI-assisted journaling converts fragmented thoughts into structured emotional insights, allowing Elias to process emotions without the pressure of real-time articulation.
• Adaptive communication support – The system learns Elias’s unique linguistic markers, improving therapeutic communication through predictive text-based interaction models.
• Real-time feedback for therapists – Provides therapists with sentiment analysis and cognitive load monitoring to adjust interventions accordingly, ensuring therapy remains responsive to Elias’s emotional needs.
• Automated emotion recognition – AI identifies emotional distress before Elias verbalizes it, enabling pre-emptive therapeutic support and reducing therapy dropout rates.
• Pattern recognition for emotional states – By analyzing past therapy sessions, AI identifies recurring stressors and suggests interventions tailored to Elias’s needs, promoting long-term emotional regulation strategies.
• Biometric integration – Heart rate variability, galvanic skin response, and pupil dilation tracking further refine AI-driven emotional interpretation, ensuring real-time emotional insights that enhance therapeutic outcomes.

Neuroimaging studies reveal that individuals with ASD exhibit distinct neural patterns in emotional processing (Pelphrey et al., 2020). By leveraging QNN-based sentiment analysis, therapy can be aligned with Elias’s neural processing differences, ensuring higher effectiveness than traditional cognitive-behavioral approaches. 

Additionally, AI-driven analytics provide valuable insights for therapists, allowing them to customize interventions based on real-time emotional data. This integration not only enhances therapist-client communication but also facilitates longitudinal emotional tracking, allowing for data-driven adjustments in therapeutic strategies. 

Furthermore, sentiment analysis can integrate with speech-to-text and natural language processing (NLP) models to help Elias articulate emotions in a structured, digestible format. This can be particularly beneficial in asynchronous therapy, where Elias can engage with therapeutic exercises at his own pace, reducing the pressure of immediate emotional articulation. The ability to identify emotional dysregulation patterns over time enables therapists to proactively modify treatment strategies, ensuring interventions remain relevant and effective.

Multi-sensory extended reality (XR) immersion therapy

A VR-based therapy space enables Elias to engage in social simulations in a controlled, customizable environment:

• Gradual sensory exposure – Adjustable sensory settings allow Elias to desensitize at a self-directed pace, fostering controlled exposure therapy.
• AI-guided social interaction training – Virtual characters simulate real-life interactions, providing constructive feedback on nonverbal cues, tone modulation, and eye contact.
• Gamified therapy tasks – Encourages emotional expression and problem-solving through interactive scenarios, reinforcing positive social behaviours.
• Cognitive load regulation – The system adjusts task complexity based on Elias’s physiological responses, preventing sensory overload while maintaining engagement.
• Therapist-supervised virtual reality sessions – Ensuring therapy remains guided and responsive to Elias’s progress, optimizing interventions in real time.
• Adaptive feedback mechanisms – Adjustments in difficulty and complexity based on Elias’s performance and biometric indicators, allowing for a personalized therapy experience.

A study by Parsons et al. (2021) demonstrated that individuals with ASD using XR-based therapy exhibited a 60% improvement in social confidence over six months. The inclusion of multi-sensory inputs helps address sensory processing challenges, which are common among neurodivergent individuals. By allowing for controlled social exposure, XR therapy reduces anxiety and fosters gradual skill-building. This approach transforms therapy from a passive, dialogue-heavy experience into an immersive, interactive, and engaging therapeutic process.

Moreover, XR environments can be programmed to mimic real-life situations Elias finds challenging, such as ordering at a café, participating in a group discussion, or handling unexpected social interactions. These simulated experiences allow for repeated practice in a low-stress setting, helping Elias build confidence before transitioning to real-world interactions. By incorporating biofeedback mechanisms, the system can also adapt in real time, dimming lights, reducing background noise, or slowing conversation speed to accommodate Elias’s sensory preferences.

Predictive adaptive response systems (PARS)

PARS integrates biometric and behavioural analytics to anticipate emotional dysregulation, enabling pre-emptive intervention through:

• Personalized de-escalation techniques – AI detects stress markers and suggests immediate coping strategies, such as guided breathing exercises or grounding techniques.
• Therapist alerts – Notifies therapists of potential emotional distress, allowing for timely intervention and crisis prevention.
• Ethical AI framework – Uses decentralized encrypted data storage to ensure security and compliance with mental health ethics, addressing concerns about data privacy.

PARS leverages machine learning algorithms to analyse Elias’s physiological and behavioural patterns, identifying early signs of distress before they escalate. This real-time monitoring helps prevent emotional overload and ensures that interventions are delivered at the most effective moment. For example, if Elias’s heart rate and galvanic skin response indicate heightened anxiety during an XR therapy session, the system can automatically initiate a calming visual scene, such as a nature landscape or deep-breathing prompts.

Additionally, predictive analytics allow therapists to assess long-term emotional trends, identifying triggers and stressors that may not be immediately apparent. By integrating wearable biosensors, Elias’s emotional fluctuations can be continuously monitored, ensuring therapy is both proactive and adaptive. The decentralized AI-driven system ensures privacy and security, addressing ethical concerns associated with data collection while maintaining transparency in decision-making.

By combining QNN-enhanced sentiment analysis, XR-based therapy, and PARS-driven predictive intervention, this model offers a holistic, neuro-adaptive therapeutic approach. It not only improves emotional articulation and social skill development but also enhances self-regulation, creating a more accessible and effective therapy experience for neurodivergent individuals like Elias.

Conclusion

There’s an urgent need for adaptive, technology-driven interventions in neurodivergent mental health care. The integrative approach outlined in this paper not only personalizes therapy but also improves accessibility and long-term outcomes for neurodivergent individuals. 

With further research and clinical validation, these innovations have the potential to revolutionize mental health care, making therapy more inclusive, effective, and responsive to neurodivergent needs.

References

• Gómez, C., Ruiz, S., & Molina, J. (2021). Sensory processing and anxiety in autistic adults: Challenges in traditional therapy models. Journal of Autism and Developmental Disorders, 51(4), 1123–1137.
• Pelphrey, K. A., Shultz, S., Hudak, C. M., & Vander Wyk, B. C. (2020). Neural mechanisms of social cognition in autism spectrum disorder: Insights from neuroimaging research. Neuroscience and Biobehavioural Reviews, 118, 229–243.
• Parsons, S., Mitchell, P., & Leonard, A. (2021). Virtual reality and social skill development in autism spectrum disorder: An empirical study on therapeutic outcomes. Autism Research, 14(7), 1258–1274.
• Rajalakshmi, T., & Venkatesan, S. (2022). Quantum neural networks in affective computing: Applications in sentiment analysis and mental health interventions. Artificial Intelligence in Medicine, 125, 102190.
• Smith, A. B., Jones, D. R., & Liu, H. (2023). Extended reality therapy for neurodivergent individuals: Bridging cognitive-behavioural therapy and immersive technology. Journal of Mental Health Technology, 9(2), 45–67.