CogniCare is an intelligent platform offering real‑time assessments and personalized suggestions, detecting psychological or psychiatric issues with over 90% accuracy by integrating ML models.
You built generate_mental_health_dataset.py to produce a 500‑row, multi‑label CSV capturing eight conditions (Depression, Anxiety, OCD, PTSD, Bipolar, Insomnia, ADHD, Autism) alongside demographic and symptom‑level features .
By fixing the random seed, you ensured reproducibility, and neatly bundled the dataset as mental_health_dataset.csv for downstream modeling .
In train_model.py, you imputed missing categorical values with the most frequent strategy and numerical values with the mean, ensuring clean inputs for your learner .
Categorical features were label‑encoded, numerical columns standard‑scaled, and these preprocessors saved (label_encoders.pkl, feature_scaler.pkl) for consistent inference .
A MultiOutputClassifier wrapping a Random Forest was trained on the prepared features, and the final model artifact was persisted as mental_health_model.pkl, accompanied by a features.pkl list for aligned predictions .
Your mental_health_app.py script delivers a real‑time Streamlit interface with a gradient backdrop and embedded CSS for an engaging UX .
It ingests user inputs (age, symptom levels, demographics), applies the saved scaler and encoders, handles missing values gracefully, and produces confidence‑scored predictions displayed as success alerts .
Additionally, it classifies overall risk (Low/Moderate/High) and generates a downloadable PDF report via ReportLab, complete with charts and color‑coded risk indicators .
In index.html (and related pages: login, signup, chatbot, appointment booking, health tracker, contributor, blog), you implemented animated gradients, floating icons, and smooth transitions for a modern feel .
Dashboard cards feature 3D hover effects, CSS keyframe animations, and responsive grids for assessment and blog sections, while the online consultation panel uses complimentary gradients and clear CTAs .
Over 60 commits, you refactored code, tightened error handling, enhanced model loading, and polished UI/UX based on feedback, demonstrating agile prototyping from data pipeline to user interface .
Your version history reflects continuous improvements, culminating in a robust, end‑to‑end mental health detection prototype ready for further testing and deployment .
Synthetic Dataset Creation Generated a 500‑row multi‑label mental health dataset covering 8 diagnoses (Depression, Anxiety, OCD, etc.) with demographic and symptom features, saved as mental_health_dataset.csv. GitHub Preprocessing & Model Training Imputed missing values, label‑encoded categorical inputs, scaled numerical features, and trained a multi‑output Random Forest classifier. Artifacts (mental_health_model.pkl, label_encoders.pkl, feature_scaler.pkl, features.pkl) are all versioned. GitHub Streamlit Web App Integration Developed mental_health_app.py—a real‑time Streamlit interface that ingests user inputs, runs predictions, displays confidence bar charts, classifies risk levels, and generates downloadable PDF reports. GitHub Front‑End Templates & Assets Included polished HTML/CSS pages (login, signup, chatbot, appointment booking, health tracker, contributor page, index) with animated gradients, dashboard cards, assessment/blog sections, and online consultation layouts. GitHub Iterative Development 60 commits demonstrate steady improvements—from data pipeline setup through model tuning to UI/UX refinements—culminating in a fully functional, end‑to‑end mental health detection platform. Dues: 1. API key 2. Googlemap API key
We don't get any funds yet.