Product Documentation
Last updated
Last updated
Welcome to the Varion AI's Documentation!
Let's dive in and see what makes Varion so special! 👋
Varion AI is an AI research group dedicated to integrating ICP's decentralized AI into healthcare systems. Our mission is to assess the capabilities and reliability of ICP's DeAI and explore its potential to enhance the modern healthcare systems and a vision of how decentralized technologies can address these issues.
Our current MVP system features a responsive user interface built with React and a backend developed in Rust, utilizing ONNX models for AI-driven insights. It performs on-chain inference for real-time data analysis and off-chain training to enhance model accuracy. Varion AI also implements Internet Identity, ensuring privacy and eliminating the need to store passwords.
Powered by the Internet Computer Protocol, Varion AI provides a secure, scalable, and decentralized infrastructure, seamlessly integrating with centralized infrastructure to bridge between Web2 and Web3 data.
Cardiovascular Disease Prediction: Utilizes DeAI to analyze patient data and predict heart-related risks by evaluating metrics such as heart rate, blood pressure, oxygen saturation, respiratory rate, and temperature. This helps identify high-risk patients for timely interventions, reducing cardiovascular morbidity and mortality rates.
Real-Time Monitoring Dashboard: The user-friendly dashboard enables healthcare professionals to view and analyze patient data in real-time, allowing hospitals to update the spreadsheet with new ECG data every minute. This continuous updating ensures that the dashboard is consistently refreshed with the most recent information.
Patient Management Strategies: This feature allows healthcare professionals to easily search, filter, and sort patient data based on various criteria. Users can search for patients by ID, filter patients by risk level, and sort patients by cardiac probability. The feature provides real-time updates and visual indicators to help identify high-risk patients at a glance.
On-Chain Inference: Run machine learning models on a decentralized platform, leveraging the unique capabilities of DeAI on the Internet Computer.
Let's get into the tutorial on how to use Varion.
If you're not in the mood to read today, you can watch our video demo here 😉:
Welcome to Varion! On this initial page, users can click the "Launch Varion" button to begin the login process using Internet Identity, providing secure access to the dashboard without the need for storing passwords.
Now that you are logged in, you can access all the features and insights Varion AI provides. Let's begin streaming your data from Web2 to get into the action.
Let's set up a connection to Google Spreadsheet through SheetDB.
Here's a guide on how to use SheetDB with an existing Google Spreadsheet:
Login to your SheetDB Account
Look for an option to create a new API or connect to an existing spreadsheet.
When prompted, please enter the Google Spreadsheet URL in the designated text field labeled "Google Spreadsheet URL," as illustrated.
Copy the full URL of your Google Spreadsheet from your browser's address bar and paste it into the designated field in SheetDB.
After entering the URL, click the "Create" button located in the bottom right corner of the dialog box.
SheetDB will generate an API site for your spreadsheet.
Navigate to the Settings tab and generate your Bearer token under the Authentication category. Don't forget to click save!
Good job! You now have a SheetDB API link connected to your spreadsheet, along with an authentication Bearer token to prevent unauthorized access.
Let's get back to the Varion website. Input the data you obtained from step 3 to start fetching data from Web2 to Web3 and click the submit button.
Click 'Start Model' to begin fetching your data at 1-minute intervals. Once the data is received, it will be sent to the Varion backend Canister to generate the cardiac arrest risk for each patient listed on the spreadsheet.
This feature allows hospitals to use their ECG devices to update the spreadsheet with new data every minute, ensuring that the chart is continuously refreshed with the latest information.
You can click 'Stop ML Model' to halt the data fetching process that occurs every minute. (Optional)
Navigate to the sidebar on the left and click 'Patient' to view detailed patient data.
This page enables easy search, filter, and sort of patient data by ID, risk level, and cardiac probability, with real-time updates and visual indicators for quick identification of high-risk patients that requires urgent handle.
Congratulations on making it this far! The Varion team extends our deepest gratitude for testing our application.
PS: Still having trouble? Reach out to our friendly Varion Staff on Discord for additional support. You can contact @shirievz for assistance.
Create a and format it according to the template provided on the Dashboard page. (If you're feeling confused, take a look at our dataset template for guidance on configuring your spreadsheet. You can view it here: .)
Visit
