Nira
Nira is a smart mental health platform offering AI-driven support, mood tracking, mindfulness, and wellness tips integrated with wearables for real-time, personalized, holistic care.
Created on 24th April 2025
•
Nira
Nira is a smart mental health platform offering AI-driven support, mood tracking, mindfulness, and wellness tips integrated with wearables for real-time, personalized, holistic care.
The problem Nira solves
-
Lack of Accessible Mental Health Support
Most people don't have immediate access to therapists or emotional support when they need it most.
Nira’s Solution:
Provides a 24/7 AI-powered chat companion that offers real-time emotional support via text, voice, and facial emotion recognition. -
Poor Mental Health Tracking & Awareness
Users struggle to recognize emotional patterns, triggers, or trends over time.
Nira’s Solution:
Daily mood tracking, emotion analysis, and visual insights help users become more self-aware and identify mood triggers or fluctuations. -
Stress, Anxiety & Overwhelm Without Coping Tools
Lack of structured tools for calming the mind or managing stress in the moment.
Nira’s Solution:
Guided meditation, breathing exercises, and emotional wellness games (e.g. Bubble Wrap, Smash Plates) provide immediate stress relief and emotional regulation. -
Disconnection From Positive Memories During Tough Times
When people feel low, they often forget their past achievements or happy moments.
Nira’s Solution:
The Hope Box feature acts as a digital safe space for uplifting content—reminders, notes, and messages—to reinforce resilience. -
Generic Mental Health Advice
Most apps offer generic tips that may not apply to every individual’s emotional journey or hormonal cycles.
Nira’s Solution:
Delivers personalized suggestions based on mood history, menstrual phases, biometric inputs, and emotional states using AI-powered analysis. -
Mental Health Stigma & Privacy Concerns
Many users are hesitant to seek help due to stigma or fear of privacy breaches.
Nira’s Solution:
Provides a safe, private platform with encrypted journaling and anonymous interaction, allowing users to explore emotional health without fear or judgment. -
Fragmnted Wellness Ecosystem
Users often have to use multiple apps for journaling, meditation, mood tracking, and fitness.
A holistic wellness hub combining journaling, AI support, wearables data, mindfulness practices, and emotional.
Challenges we ran into
- Emotion Recognition Accuracy (Face + Text)
Challenge:
Implementing accurate facial emotion detection and real-time text-based sentiment analysis was tricky, especially with diverse facial features, lighting conditions, and subtle emotional expressions.
Solution:
We improved facial recognition by switching to a CNN-based facial emotion model fine-tuned on the FER+ dataset, which provided better emotion classification.
For text sentiment, we moved from basic rule-based NLP to BERT-based sentiment analysis, which significantly improved contextual understanding.
Added a confidence threshold filter, so if the model is unsure, it avoids suggesting an emotional state to prevent misdiagnosis.
- Voice-to-Text Transcription Errors in Journaling
Challenge:
Users faced frequent misinterpretations with voice entries, especially in noisy environments or for users with accents.
Solution:
Integrated Google Speech-to-Text API, which provided more reliable transcription with accent support.
Added a "preview and edit" mode before saving voice logs, giving users a chance to make corrections.
Implemented basic noise cancellation preprocessing on input audio using WebRTC tools.
- Wearable Integration Inconsistencies
Challenge:
Integrating data from various wearable devices (like Fitbit, Apple Health) posed compatibility issues and inconsistent data syncing.
Solution:
Built modular API connectors for each device type to standardize incoming data.
Introduced a data normalization layer that converted all heart rate/stress data into a common format.
For real-time syncing, we used webhooks and polling fallbacks to ensure up-to-date biometric tracking.
Tracks Applied (1)
Groq track
Groq