MysticHealer

Inspiration

The inspiration behind our project stemmed from a desire to improve mental health accessibility and reduce the stigma associated with seeking help. Many individuals struggle to recognize or acknowledge symptoms of depression, while others face barriers such as high consultation fees and lack of support systems. By leveraging AI-driven analysis and real-time peer support, we aimed to create a platform that gently guides users toward help without explicitly diagnosing them, ensuring they receive the support they need in a safe and non-intimidating environment.

What it does

The project detects depression using a combination of survey responses and wearable device data. If signs of depression are detected, it provides accessible mental health support by:

Connecting users with affordable therapists – Professionals offering free or low-cost consultations. Scheduling online consultations – Users can book virtual therapy sessions. Facilitating peer support groups – Users can join online group meetings to connect with others facing similar challenges. This AI-powered system ensures early detection, accessibility, and community support, helping individuals get the mental health care they need.

How we built it

The project was developed iteratively, starting with a user-friendly frontend built with ReactJS to ensure an intuitive experience. The backend infrastructure was powered by NodeJS and Flask, with GraphQL APIs for optimized data communication. Flask served as the bridge between the Convolutional Neural Network (CNN) model and the application, ensuring smooth model inference.

We integrated health sensor data from wearable devices to improve depression detection, allowing users to optionally sync heart rate, sleep patterns, and activity levels for a more comprehensive assessment. The CNN model processed this physiological activity data alongside user responses from a mental health questionnaire, ensuring high accuracy in risk assessment.

For real-time interactions, we implemented Socket.IO and WebRTC, enabling secure, low-latency virtual therapy sessions and peer support groups. This created a safe space for users to connect with therapists and others experiencing similar challenges.

Challenges we ran into

We found two parallel datasets—one containing health sensor data and the other with categorical and survey responses—but merging them was challenging due to differences in structure and missing correlations. Neither dataset alone was sufficient to confirm the possibility of depression, as sensor data lacked psychological context, while survey responses did not include physiological indicators. To address this, we built two separate models—one analyzing biometric data using CNNs and another processing survey responses—and integrated them into a step-function pipeline. If the wearable device model detects potential mental health concerns, the system then considers survey responses for a more comprehensive and reliable assessment.

Accomplishments that we're proud of

We successfully built an AI-driven mental health support system that combines wearable sensor data and survey responses to provide a holistic assessment. The system seamlessly integrates Flask, NodeJS, and GraphQL, ensuring smooth communication between the ML model and the application. Additionally, we implemented real-time peer support features using WebRTC and Socket.IO, enabling users to connect securely through live chat and video sessions. Our approach prioritizes privacy and accessibility, offering users a safe and supportive environment to seek help.

What we learned

We learned how to merge diverse data sources for a more effective mental health assessment and the challenges of integrating ML models into a full-stack system. Implementing real-time support features deepened our understanding of secure, low-latency communication, while working on privacy-focused AI highlighted the importance of ethical considerations in mental health technology.

What's next for MysticHealer

Next, we plan to integrate wearable devices directly into the app to enable real-time data collection and analysis. This will allow continuous monitoring of physiological indicators like heart rate and sleep patterns, improving the accuracy of mental health assessments. Additionally, we aim to develop personalized alerts and insights, helping users track their well-being and seek support proactively.