We started as a collaborative initiative between Climate Resilient Communities (CRC) and Kids Help Phone (KHP), bringing together expertise in climate resilience, youth mental health, and data science. Our mission is to understand and address climate anxiety among young people, using innovative, interpretable AI models to support timely interventions and foster emotional resilience in the face of climate change.
The Challenge
Climate change is not just an environmental crisis, but a mental health crisis for many communities. Increasingly, reports show deep concerns, fear, and anxiety about the planet’s future. Yet, there is limited research on how this anxiety manifests in everyday conversations, and few tools exist to detect and respond early. Without timely support, climate anxiety can escalate, impacting mental health, wellbeing, and hope for the future.
Our Solution
We are developing an open, interpretable Natural Language Processing (NLP) and Large Language Model (LLM) framework capable of detecting early signs of climate anxiety in youth discourse. By analyzing open-source datasets and, in later phases, anonymized youth conversations, we aim to uncover linguistic cues and common themes that indicate distress. This will lead to the creation of an interactive tool for educators, counselors, and community leaders to visualize patterns, understand youth concerns, and take informed action.
Research Methodology
Our research follows a structured, multi-phase approach. In Phase I, we focus on open-source datasets to establish foundational insights and model prototypes. We begin with data acquisition, descriptive analysis, and exploratory data analysis to uncover patterns and themes related to climate anxiety among youth. From there, we develop and test NLP-based and LLM-based models capable of detecting linguistic cues of anxiety. Model performance is validated through rigorous testing, ensuring interpretability, fairness, and scalability. Phase II will expand the study to anonymized youth conversation transcripts, refining the models and tailoring intervention recommendations.
Data Sources
We use publicly available data from Reddit and Twitter.
Analytical Approach
Our analysis pipeline begins with data cleaning, preparation, and feature engineering to capture relevant linguistic and contextual features. We apply a combination of topic modeling (e.g., Latent Dirichlet Allocation) and advanced transformer-based language models to detect patterns of climate anxiety. Interpretability is a core focus, allowing us to identify the specific words, phrases, and themes most indicative of anxiety. Visualization tools are integrated to make insights accessible for non-technical stakeholders, enabling better understanding and actionable use.
Ethical Considerations
We prioritize the privacy, dignity, and wellbeing of youth participants. All data is anonymized and de-identified before analysis, with strict adherence to ethical research standards and applicable privacy laws. Sensitive content is handled with care to avoid harm, and findings are presented in ways that respect youth voices. Our work is designed to inform supportive interventions, not to stigmatize or label individuals. The open-source nature of our tools is balanced with safeguards to prevent misuse.
Open Source Commitment
Our approach prioritizes accessibility and collaboration. The data workflows, detection models, and visualization tools developed in this project will be built on open-source datasets and shared openly where possible. This ensures scalability, transparency, and the ability for other researchers, nonprofits, and communities to adapt and expand the tools for their own contexts.
Impact
By enabling early identification of climate anxiety, our work empowers support networks and mental health professionals to respond proactively. The project will also contribute valuable research on climate perspectives, helping shape better interventions and policies. Our ultimate goal is to build resilience, provide a sense of agency, and help transform climate-related fears into constructive engagement.
Our Team
Current Members
Karim El-Sharkawy (Applied AI Scientist & Project Manager)
Karim El-Sharkawy (Project Manager & Applied AI Scientist) Karim turns complex data challenges into impactful, real-world solutions. He thrives at the intersection of machine learning, MLOps, and applied mathematics, designing and deploying scalable AI workflows that deliver measurable results. From leading cross-functional teams to building production-ready NLP systems, He's driven to bridge innovation and impact in every project he takes on.
Zainab Rehman (Product Manager)
Product Manager with a passion for turning ideas into impactful solutions. I thrive in building and scaling SaaS products, driving growth through Agile methodologies and cross-functional collaboration. From leading large-scale implementations to refining user experiences, I’m all about delivering results that matter.
Paras Jamil (Staff Engineer, Backend)
Paras is the Founder of Syntax AI and a Backend Engineer, with over a decade of experience building language-driven AI systems. She has delivered multilingual NLP solutions, agentic workflows, RAG pipelines, NL-to-SQL translation tools, and REST API deployments, drawing on her background in computational linguistics, backend engineering, and large-scale model deployment across embedded and cloud platforms.
Jorge Rivera (WebApp Dev)
Jorge has been 10 years making Generative AI, Data Science & Insights accessible to business leaders, he is a certified AI Developer & Data Scientist and has earned 3 recognitions as Top Performance Individual from Scotiabank. Jorge has had roles as Analyst, Manager of Data & Analytics, Data Strategist, Data Scientist and Senior Manager of Data & Analytics.
Helena Yu (Partnership Lead and Strategic Support)
Helena is a Data Scientist and AI Strategist that co-leads a multidisciplinary team of 30+ at Climate Resilient Communities. Her past pro bono statistical consulting work include database design for a healthcare nonprofit, descriptive and NLP analysis of a global gender study, and data pipelines and predictive analytics for human trafficking interventions.
Past Members
Amanda Easson (Project Manager) Data Science Manager at TD Bank with a PhD in Cognitive Neuroscience, specializing in forecasting, machine learning, and teaching data science at U of T and Waterloo.
Maryam Tavakoli (Data Scientist) AI/ML-focused Data Scientist with a PhD in Physics and 6+ years of experience building AI solutions and providing data-driven insights using ML, NLP, and reinforcement learning.
Get Involved
We welcome collaboration from researchers, mental health practitioners, climate educators, and developers interested in AI for social good. Whether by contributing to our open-source codebase, supporting outreach to youth communities, or helping design interventions, your expertise can help us make a difference. To join our efforts, please contact us through our team page.
We'd love to talk more about how we can collaborate. Reach out to partnership@crcgreen.com to start the conversation.
If you don’t see your particular skill set here but are still interested in the project, please reach out to volunteers@crcgreen.com to get involved!