NextStep Health AI Assessment

Demo video: 

The NextStep Health AI Assessment tool is an innovative tool focused on whole-child well-being that enables safe data collection, and identification of learner challenges, alongside Socratic prevention and intervention for learner challenges.

The tool leverages an AI Avatar modality, with conversation data that can be analyzed for both risks, opportunities, and focus areas for growth.

NextStep Health's AI Assessment Tool is designed to address the whole-child well-being of priority learners, particularly Pre-K-8 students who face significant barriers to opportunity. The tool focuses on Black and Latino learners and those experiencing poverty, aiming to level the playing field by providing innovative, evidence-based assessment and intervention.

How the Solution Works:

1. AI Avatar Modality:

   • Engaging Avatar: The AI tool employs an engaging avatar that interacts with learners through friendly, conversational dialogues.
   • Adaptive Conversations: Through these dialogues, the tool gathers insights into each learner's emotional state, challenges, and growth opportunities.
     

2. Comprehensive Data Collection:

   • Safe and Confidential: The AI collects data securely and confidentially to ensure a safe environment for students.
   • Holistic Understanding: Data includes emotional well-being, social-emotional learning (SEL) competencies, and cognitive challenges.
   • Risk and Opportunity Identification: Analysis identifies risks (e.g., social isolation, anxiety) and opportunities (e.g., leadership potential, creativity).

3. Socratic Prevention and Intervention:

   • Personalized Guidance: Based on data analysis, the tool provides personalized recommendations using Socratic questioning.
   • Interactive Coaching: The avatar coaches students in developing skills like emotional regulation, empathy, and decision-making.
   • Cultural Relevance: Conversations and interventions are culturally and linguistically adapted to meet the needs of diverse learners.
     

4. Data-Driven Support for Educators:

   • Actionable Insights Dashboard: Educators receive a dashboard with anonymized, aggregated insights.
   • Custom Intervention Plans: Provides suggested intervention strategies tailored to the class or individual needs.
   • Continuous Improvement: Enables educators to track progress and refine their approach based on real-time data.

Impact on Priority Pre-K-8 Learners and Educators:

1. Improved Social-Emotional Development:

   • SEL Skills: Early learners build essential SEL skills like empathy, self-awareness, and conflict resolution.
   • Resilience and Coping: Priority students learn effective coping mechanisms, fostering resilience.
     

2. Better Identification and Support for Learning Challenges:

   • Early Identification: Identifies learning challenges early, allowing for proactive interventions.
   • Targeted Interventions: Personalized strategies address specific challenges, preventing students from falling further behind.
     

3. Culturally Responsive Education:

   • Increased Engagement: Culturally adapted conversations resonate with priority learners, increasing engagement.
   • Reduced Barriers: By addressing cultural and linguistic barriers, more learners receive appropriate support.
     

4. Enhanced Educator Effectiveness:

   • Insightful Assessments: Educators gain deeper insights into students' well-being and challenges.
   • Informed Decision-Making: Data-driven interventions help educators tailor their teaching strategies for maximum impact.
   • Whole-Child Approach: By addressing emotional and cognitive challenges, educators can foster holistic student development.
     

5. Empowering Parents and Communities:

   • Parental Involvement: The tool engages parents through regular progress updates, encouraging proactive support at home.
   • Community Partnerships: Insights can inform community organizations and programs, aligning support services for priority learners.

NextStep Health’s AI Assessment Tool transforms mental health and well-being support for Pre-K-8 learners by leveraging personalized, AI-driven interactions. By identifying challenges early and providing culturally relevant support, the tool empowers priority learners and educators to navigate barriers to opportunity more effectively. This innovative approach creates a safer, more inclusive learning environment, ensuring every child has the chance to thrive.

The NextStep Health team was built to encompass an array of domain expertise, special skills, and advantageous factors that uniquely position us to tackle the mental health crisis among youth effectively:

Sam Warach's Leadership, Innovation, and Operational Experience: As founder and CEO, Sam brings deep experience from building NextStep Health’s platform from zero-to-one alongside gov. agencies like the State of NH and NYC Mayor’s Office, NGOs like the Clinton Foundation, and CBOs like Big Brothers Big Sisters. Sam’s education from Harvard/MIT and UNH, and innovation fellowship at Stanford’s d.school is complemented by his recognition as a Forbes 30 Under 30 in Boston and NSF-funded innovator. His background in education and health technology innovation ensures a strong foundation in leading initiatives that blend technology with educational content to assess and address complex social issues like mental health.

Chandra Fincke's Technical Expertise: With a background in neuroscience and a focus on machine learning and computational human behavior from Yale, Chandra is well-equipped to lead the technology and data aspects of NextStep Health. His experience in co-founding NIH-funded projects related to overdose prevention showcases his capabilities to build and leverage technology for health-related interventions.

Diverse and Accomplished Advisory Board: The advisory board's diversity, including experts from Harvard Medical School, MIT, Harvard Graduate School of Education, and the Christensen Institute, among others, provides a broad spectrum of insights from clinical mental health to education and business strategy. This collective wisdom guides NextStep Health in maintaining a multi-faceted approach to mental health resilience.

The diversity within the NextStep Health Leadership team offers several advantages in developing a more impactful solution, and is complemented by a short list of ad-hoc staff consultants and informal advisors with deep expertise in software architecture and engineering, data science, psychiatry, and compliance:

Multidisciplinary Perspectives: The team's varied educational and professional backgrounds foster a holistic approach to problem-solving, ensuring that solutions are not only technologically advanced but also grounded in psychological principles and practical for real-world application.

Cultural and Linguistic Sensitivity: The diverse expertise and backgrounds of the team members, including international education and global experiences, enhance the platform's ability to be culturally and linguistically inclusive. This is critical in developing resources that resonate with a diverse teen population.

Innovation and Creativity: The combination of expertise in education technology, neuroscience, machine learning, and strategic operations within the team fosters an environment of innovation. This diversity in skill sets and thought processes is likely to result in creative solutions that stand out in the field of mental health interventions.

Broad Network and Resources: The collective network of the team members and advisors, spanning prestigious academic institutions, healthcare, and business sectors, provides access to a wide range of resources, partnerships, and collaborations that can enhance the effectiveness and reach of NextStep Health's initiatives.

Overall, our team's diverse domain expertise and knowledge, special skills, and the strategic advantage offered by our varied backgrounds position them to effectively address the mental health crisis among youth with innovative, evidence-based, and culturally sensitive solutions.

Although NextStep Health's primary platform (https://nextstephealth.com) is in Growth stage, our new AI-Avatar tool is more so in the pre-pilot closed-testing and fine-tuning Prototype stage.

NextStep Health’s AI Assessment Tool is an innovative solution that revolutionizes the way we approach mental health and well-being support for Pre-K-8 learners and their educators. By combining cutting-edge technology with evidence-based methodologies, it provides a significantly improved approach to the persistent challenges of whole-child assessment and intervention.

What Makes Our Solution Innovative:

1. AI Avatar Modality:

   • Engaging and Adaptive Interaction: The AI Avatar interacts with learners in a friendly, non-judgmental manner, making them feel comfortable and understood. This helps uncover emotional states and challenges that traditional assessments might miss.
   • Conversational Depth and Fluidity: The conversations are not scripted but adaptive, guided by advanced Natural Language Processing (NLP) to respond fluidly to each learner's input, creating a highly personalized interaction.
     

2. Whole-Child Focus:

   • Comprehensive Data Collection: Unlike traditional assessments that focus solely on academic performance or basic behavioral metrics, this tool collects data across multiple dimensions: social-emotional skills, cognitive development, emotional state, and socio-environmental factors.
   • Socratic Prevention and Intervention: By using Socratic questioning, the tool helps students develop critical thinking and self-reflection skills, empowering them to identify and solve their own challenges.
     

3. Culturally and Linguistically Responsive:

   • Personalization for Diversity: The AI Assessment Tool adapts conversations to reflect cultural and linguistic diversity, ensuring it resonates with Black, Latino, and low-income learners.
   • Content Inclusivity: Scenarios, stories, and advice are carefully designed to be inclusive and relatable, breaking down barriers to mental health support.
     

4. Data-Driven Insights and Actionable Feedback:

   • Educator Dashboard: Aggregated and anonymized data is presented in an intuitive dashboard, providing actionable insights to educators.
   • Custom Intervention Plans: Based on assessment data, the tool generates tailored intervention strategies, enabling educators to address each student’s unique needs more effectively.

Broader Positive Impacts and Market Change:

1. Catalyzing Systemic Change:

   • Scalable Model for Other Organizations: By demonstrating the impact of AI-enabled whole-child assessment, NextStep Health can inspire other organizations to adopt similar approaches, leading to systemic improvements in mental health support.
   • Policy Influence: The data and insights from this tool can inform educational policy, driving more comprehensive mental health frameworks.
     

2. Improving Market Landscape:

   • Shifting Focus to Prevention: Traditional mental health interventions often focus on treatment rather than prevention. By emphasizing Socratic prevention and early intervention, NextStep Health's approach shifts the market towards proactive care.
   • Increased Accessibility and Inclusivity: Our culturally responsive model sets a new standard for inclusivity in mental health solutions, encouraging the market to prioritize diverse learners.
     

3. Collaboration and Research Opportunities:

   • Data for Research: The anonymized data provides valuable insights for researchers studying child development and mental health trends.
   • Partnerships with Schools and Organizations: The solution’s ability to engage with learners, educators, and parents fosters partnerships that can lead to broader community-based interventions.

NextStep Health’s AI Assessment Tool is innovative because it combines AI-driven personalized interactions with a whole-child focus, cultural responsiveness, and data-driven insights. By addressing the challenges faced by Pre-K-8 priority learners and their educators in a significantly improved way, it has the potential to catalyze broader positive impacts, change the mental health landscape, and ultimately improve the lives of millions of children.

NextStep Health’s AI Coach leverages a combination of advanced AI technologies to deliver personalized and impactful mental health support for Pre-K-8 learners and their educators. Here's a closer look at the core AI and technological components that power the solution:

Core AI Technologies:

1. Natural Language Processing (NLP):

   • Conversational AI:
     • Our solution uses NLP to power the AI Avatar's ability to engage in natural, conversational dialogues with learners.
     • The conversational engine is built using Transformer models (e.g., GPT architecture), which are pre-trained on large language datasets and fine-tuned to detect and interpret learners’ responses accurately.
   • Sentiment Analysis:
     • NLP-based sentiment analysis identifies the emotional tone of learners’ responses, providing insights into their emotional state.
     • This information helps personalize the interaction and informs the Socratic questioning approach.
       

2. Machine Learning (ML):

   • Behavior Prediction and Analysis:
     • Our ML models analyze conversational data to identify patterns that may indicate learning challenges, emotional distress, or opportunities for growth.
     • These models are trained using labeled datasets and reinforced through continuous learning from user interactions.
   • Personalization Engine:
     • ML algorithms enable dynamic personalization of the AI Coach's guidance and interventions based on each learner’s unique needs, preferences, and past behavior.
     • This allows for adaptive learning paths that evolve over time.
       

3. Knowledge Representation and Cognitive Computing:

   • Metacognitive Skill Building:
     • The AI Coach uses a knowledge representation system to guide learners through building metacognitive skills like planning, self-monitoring, and self-reflection.
   • Socratic Questioning Framework:
     • Cognitive computing powers the Socratic questioning approach, enabling the coach to ask thought-provoking questions that stimulate critical thinking and problem-solving.
       

4. Speech Recognition:

   • Voice-Based Interaction:
     • The platform offers voice-based interaction capabilities using speech recognition technology.
     • This allows learners who prefer speaking to typing to engage with the AI Coach more comfortably.
       

5. Deep Learning (DL):

   • Emotion Detection:
     • Deep learning models are used to analyze voice tone and facial expressions (when learners consent to video-based interaction) for a more accurate understanding of emotions.
     • These DL models are built using Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs).

Frontend Technologies:

1. React and React Native:

   • The user interface is built using React for the web app and React Native for mobile applications, ensuring a consistent and responsive user experience across devices.
     

2. Unity and WebXR:

   • For VR/AR-based interactions, we use Unity for 3D modeling and WebXR to deliver immersive experiences through the browser.
     

3. GraphQL and REST APIs:

   • GraphQL and REST APIs enable seamless communication between the front-end and back-end, providing real-time updates and data synchronization.

Third-Party Models and Tools:

• Pre-Trained NLP Models:
  • We leverage pre-trained models for conversational AI, fine-tuned to suit our specific educational, whole child and child-safe, and mental health context.
• Speech-to-Text and Text-to-Speech:
  • Speech-to-Text and Text-to-Speech APIs enhance voice-based interactions.

NextStep Health’s AI Coach utilizes a comprehensive blend of NLP, ML, DL, and cognitive computing technologies to create a personalized, engaging, and inclusive platform that helps learners build mental health resilience. This fusion of core AI subdomains and frontend technologies ensures an innovative and impactful solution that can significantly transform mental health support for Pre-K-8 learners.

There is adjacent evidence surrounding the efficacy of AI Avatars in analysis and provision of positive outcomes for whole-child resilience, including reduction of lonliness and suicidality: https://www.nature.com/articles/s44184-023-00047-6

Our AI Avatar is built for assessment around existing evidence-based research. Demo here: 

Ensuring Equity and Combating Bias in NextStep Health’s AI Implementation

NextStep Health's mission is to provide mental health support that is equitable and free of bias for all learners, particularly those from marginalized communities. Our approach includes understanding diverse backgrounds and addressing algorithmic bias through intentional goals, actions, and monitoring strategies.

1. Goals for Equity and Understanding Diverse Backgrounds

A. Equity Goals:

• Ensure equitable access to mental health resources and support for all learners.
• Create a culturally responsive AI Coach that reflects the diversity of learners, particularly Black, Latino, and low-income students.

B. Understanding Diverse Backgrounds:

• Community Engagement: Regularly engage with educators, parents, and students through surveys, focus groups, and workshops to understand their needs and cultural contexts.
• Advisory Board: Our diverse advisory board includes clinical mental health experts specializing in underrepresented groups, ensuring that the AI Coach aligns with diverse learners' cultural nuances.

2. Actions Taken to Shape the Solution in Response to Understanding

A. Inclusive Content Development:

• Culturally Responsive Curriculum:

  • Collaborated with educators and cultural experts to create culturally relevant scenarios and interventions.
  • Adapted the Socratic questioning framework to address the unique social-emotional challenges faced by marginalized groups.
    

• Language and Communication Style Adaptation:

  • AI Coach supports multiple languages, including Spanish.
  • Adjusts communication styles based on learners' cultural norms and linguistic preferences.

B. Diverse Training Data:

• Representation in Datasets:
  • Collected training data from schools serving Black, Latino, and low-income communities to ensure diverse representation in conversational data.
  • Augmented training datasets with publicly available mental health conversations from diverse demographics.

3. Combating Algorithmic Bias

A. Bias Mitigation Strategies:

• Fairness Metrics:
  • Implemented fairness metrics (e.g., demographic parity, equal opportunity) to detect and measure bias in model predictions.
• Adversarial Testing:
  • Conducted adversarial testing to identify and minimize bias by simulating interactions from different demographic groups.

B. Model Fine-Tuning and Calibration:

• Differential Calibration:
  • Calibrated models differently for demographic groups to ensure similar predictive performance across groups.
• Human-in-the-Loop Review:
  • Involved human reviewers from diverse backgrounds to assess model predictions for accuracy and bias.

4. Monitoring and Avoiding Harmful Impacts

A. Bias Monitoring Framework:

• Continuous Monitoring:
  • Established a continuous monitoring system to identify potential biases and harmful impacts in real-time.
• User Feedback Loop:
  • Integrated user feedback into the monitoring system to capture and address any instances of bias.

B. Ethical AI Principles and Guidelines:

• Transparency and Explainability:
  • Ensured that model predictions are transparent and explainable through educator dashboards and user-friendly explanations.
• Impact Assessment:
  • Conducted regular impact assessments to evaluate how interventions affect different learner groups.

NextStep Health is committed to ensuring equitable access to mental health resources by creating an AI Coach that understands and responds to the diverse needs of learners. Our multi-layered approach to combating bias and harmful impacts ensures that marginalized groups receive the personalized support they deserve. We strive to continuously improve our solution through inclusive design, intentional bias mitigation, and constant community engagement.

FTE: 2

Part-time/Contractors: 4

Plan to Be Pilot-Ready Within the Year

Current Progress:

• Pilot Programs: Design pilot programs to serve and engage 500+ students and 50+ educators.
• Diverse Learners: Engage with Black, Latino, and low-income learners, ensuring cultural responsiveness.

Year 1 Goals:

1. Content and Curriculum Development:

   • Finalize culturally responsive curriculum for Transitional Kindergarten and Pre-K, alongside K-8 students.
   • Expand Socratic questioning framework to cover more social-emotional learning (SEL) scenarios.
     

2. Technical Enhancement:

   • Improve the AI Coach's NLP models for higher accuracy in multiple languages (English, Spanish).
   • Enhance AI Avatar modality for a more engaging learner experience.
     

3. Partnership Building:

   • Establish partnerships with 10 new schools, prioritizing those with a high concentration of Black, Latino, and low-income learners.
   • Collaborate with state education departments to ensure curriculum alignment.
     

4. Pilot Design:

   • Design randomized controlled trials (RCTs) to measure the AI Coach's impact on SEL outcomes.
   • Develop an educator training module for effective implementation.

Evidence of Progress:

• Positive Feedback

• Technical Achievements, Open Beta Launch

• Conversational interpretation

• Educator dashboard launch with personalized intervention recommendations

Key Milestones for the Next Year:

• Q1:
  • Refine content for priority learners, complete ongoing technical enhancements.
• Q2:
  • Launch beta partnership program, onboard 10-15 schools.
  • Begin training educators and collecting baseline data.
• Q3-Q4:
  • Conduct RCTs, analyze preliminary impact results.
  • Refine and expand the solution based on feedback.

NextStep Health aims to pilot the AI Coach with priority learners in 10-15 schools within the next year. Our positive pilot results, technical advancements, and partnership-building efforts ensure we are on track to achieve these goals, ultimately supporting equitable access to mental health resilience resources.

Plans for Availability, Accessibility, and Affordability at Scale

1. Availability

• Multiple Learning Platforms:

  • Mobile and Web Apps: Accessible through iOS, Android, and web browsers.
  • VR and In-Person Workshops: For schools with VR headsets and in-person sessions, ensuring varied access.

• Curriculum Integration:

  • Align the AI Coach curriculum with state and national SEL standards for seamless integration into existing school programs.

2. Accessibility

• Culturally Responsive Design:

  • Multilingual Support: Support for English and Spanish languages, with plans to expand to other languages.
  • Representation in Content: Culturally relevant avatars and scenarios that resonate with Black, Latino, and low-income learners.
    

• Inclusive Usability Features:

  • Text-to-speech and speech-to-text functionalities for learners with disabilities.
  • Socioeconomic-friendly, low-bandwidth mode for mobile apps.

3. Affordability

• Free or Subsidized Access for Priority Learners:

  • Provide free access to public Transitional Kindergarten, Pre-K, and K-8 students, particularly those from Black, Latino, and low-income families.
  • Partner with state education departments and philanthropic organizations to secure funding for subsidized access.
    

• Tiered Pricing Model:

  • Schools and Districts: Affordable subscription-based pricing for schools and districts, ensuring equitable distribution.
  • Non-Profit and Community Organizations: Reduced or free pricing for organizations working with priority learners.

4. Scalability Strategy

• Strategic Partnerships:

  • Partner with NGOs, non-profits, and government education and health departments to expand reach.
  • Leverage existing relationships with organizations like the Clinton Foundation, and build new ones with orgs like CZI and the Bill & Melinda Gates Foundation to scale impact.
    

• Data-Driven Improvements:

  • Continuously refine the solution using insights from pilot programs and user feedback, ensuring it's responsive to the evolving needs of priority learners.

NextStep Health’s AI Coach is designed to be universally available, accessible, and affordable to priority learners. Our strategic partnerships, tiered pricing model, and focus on cultural responsiveness ensure that Black, Latino, and low-income learners receive equitable access to this innovative mental health solution.

NextStep Health is applying to the Learner//Meets//Future Challenge to accelerate our mission of improving mental health resilience in priority Pre-K-8 learners. While we've made significant progress, we face several barriers that the support from Solve and the Bill & Melinda Gates Foundation can help us overcome:

1. Financial Barriers:

• Barrier: Limited funding for scaling up our solution to underserved communities.
• Solution: Financial support from the challenge will enable us to expand free or subsidized access for Black, Latino, and low-income students.

2. Technical Barriers:

• Barrier: Need for further AI model improvement and integration with school information systems.
• Solution: Access to technical experts and resources will refine our AI Coach’s performance and ensure seamless data integration.

3. Market Barriers:

• Barrier: Reaching a wider network of educators and schools.
• Solution: Partnerships facilitated by the challenge will open doors to new schools, districts, and educational organizations.

4. Cultural Barriers:

• Barrier: Ensuring culturally responsive content and approach.
• Solution: Collaboration with advisors at the Bill & Melinda Gates Foundation will help us better tailor our solution to diverse learners.

Solve's network and resources, combined with the Foundation's expertise, will empower us to break down these barriers and transform the mental health landscape for priority learners.