1. Upload the Image
Drag your Kaplan-Meier chart or efficacy curve (in PNG, JPG or JPEG format) and drop it inside the dashed canvas area, or click "Select File" to browse locally.
2. Axis Calibration
Calibrating the image before digitizing points is essential to convert pixel coordinates into real clinical values.
- Click "🚀 START CALIBRATION" to launch the interactive wizard.
- The cursor will change and the wizard will guide you step-by-step to place 4 reference points:
- X1 (X Origin): Click on the zero or starting point of the X axis (time).
- X2 (Max X): Click on the known maximum value of the X axis.
- Y1 (Y Origin): Click on the zero or starting point of the Y axis (rate).
- Y2 (Max Y): Click on the known maximum value of the Y axis (usually 100% or 1.0).
- Fill in the numeric fields in the sidebar with the real clinical values corresponding to your axis limits (e.g. X2 = 42 months, Y2 = 100%).
3. Curve Creation & Management
In the "2. Manage Curves" panel you can organise the study arms:
- Use the "+ Add" button to create a new arm. You can rename it and change its colour to distinguish arms easily.
- Select the curve you want to draw by clicking it in the list. **Note:** Switching active curves automatically clears the captured color palette to prevent mixing color profiles between arms.
- Choose the Interpolation Type: use "KM Step Function" for Kaplan-Meier (step) curves or "Linear" for continuous curves.
4. Point Plotting & Editing
To add points manually:
- Left-click on the curve directly in the canvas. A point marker will appear.
- You can drag existing points to fine-tune their position with precision.
- To delete a point: right-click on it in the canvas, or click the trash button (🗑️) in the "3. Points" sidebar table.
- If you need to place points very close together for precision, hold Ctrl or Shift while clicking to force point creation without dragging a nearby existing point.
5. Image Filters
If the original chart grid makes tracing difficult, use the sliders to adjust Brightness, Contrast or Binarization. The "Remove Grid" button applies automatic filters to try to clean background lines.
6. Smart Autotracing (Advanced)
To digitize complex curves with a single click:
- Click "👁️🗨️ Eyedropper" and click on several points of the curve in the canvas to capture its color tones (building a "fan" palette).
- Live Color-Matching Preview: When the Eyedropper, Highlighter, or Auto-1Click is active, matching pixels are highlighted on the canvas in the active curve's color. Adjust the **Tolerance** slider to dynamically see what the autotracer sees, ensuring only the target curve is selected.
- Click "🖌️ Highlighter" and draw over the canvas to shade the area where the curve lies.
- Click "🚀 Auto-1Click" and click on a starting point of the curve. The algorithm will scan the shaded area and trace the points automatically.
7. Dashboard & Reports
Once the curves are traced, go to the "3. Dashboard" tab:
- Review the calculated clinical outcomes: estimated Hazard Ratio (HR), Median Survival and Restricted Mean Survival Time (RMST).
- Adjust the integration horizon (AUC / RMST) with the live slider to evaluate benefit at different time points.
- Enable Parametric Extrapolation to project curves beyond the observed study time using Weibull, Exponential, Log-Normal or Log-Logistic models.
- Use the top buttons to export reproducible scripts to R or Python, or click "🖨️ Generate Clinical Report" to get a premium printable PDF.
8. Session Save & Load (New Function)
Your work is auto-saved in the browser. To share or resume a session on another device:
- Click "Save Session" to download a .json file with all digitized data, image background, and calibration coordinates.
- Click "Load Session" to restore a previously saved .json file, including the original image background and all curves.
- Click "Export Dashboard" to generate a standalone interactive HTML file containing the full dashboard — shareable without the app.
9. Export to Excel (New Function)
In the "3. Points" panel, click "📊 Export to Excel" to download a .csv file with all digitized coordinates:
- The file contains a shared Y column and one X column per curve, compatible with Excel, R and Python.
- A study metadata footer is automatically appended to the file.
10. Patients at Risk & IPD Reconstruction (Advanced Biostatistics)
If you enable "Patients at Risk" in the Settings tab, you can manually enter patients at risk at specific time landmarks (e.g. 0m, 6m, 12m, etc.). This unlocks the following premium biostatistical features:
- IPD Reconstruction (Guyot 2012 Algorithm): Reconstructs individual patient micro-data (event times and censoring) by synchronously crossing the digitized curve and the risk table in the browser.
- Real Cox Model: Computes a true Cox Hazard Ratio via Newton-Raphson partial maximum-likelihood optimization, along with its 95% confidence interval.
- Statistical Significance (Log-rank test): Calculates the exact p-value for the hypothesis test between survival curves.
- Greenwood Bands: Shades 95% Kaplan-Meier confidence intervals for each arm over time in the chart.
