Medical Decisions Extraction, Visualization, and Annotation

Decision Extraction & Classification

This tool allows you to quickly identify and categorize key clinical decisions within a single clinical note.

1. Input Text: Enter or paste the clinical note into the text box. You can also select from the provided examples.
2. Run: Click the "Run" button to process the text.
3. View Results: The labeled output will be displayed, highlighting the identified clinical decisions with color-coded categories.

This provides a simplified overview of the patient's history and key decisions at a glance.

Patient Visualization

This tool allows you to visualize the timeline of clinical decisions across multiple clinical notes for a single patient.

1. Upload Clinical Notes: Upload multiple clinical notes in text format. You can also select from the provided examples.
2. Add/Remove Notes: Use the "+" and "-" buttons to add or remove input fields for additional notes.
3. Filter Timeline:
   • Decision Type: Select one or more decision types from the dropdown menu to filter the timeline. Select "All" to display all decision types.
   • Start/End Date: Enter a start and end date (MM/DD/YYYY) to filter the timeline by date range.
   • Filter: Click the "Filter Timeline" button to apply the selected filters.
4. Generate Summary: Click the "Generate Summary" button to generate a summarized view of the clinical decisions organized by date and category. This will also update the timeline plot.
5. View Timeline: The interactive timeline plot displays the clinical decisions over time. Hover over each data point to view details.

This visualization helps to understand the progression of clinical decisions and identify trends in patient care.

Interactive Narrative Annotator

This tool allows you to manually annotate medical text for detailed analysis and model training. Follow these steps:

1. Input Text: Enter or paste the text you want to annotate into the text box. You can also use the provided examples.
2. Generate Predictions (Optional): Click "Generate Predictions" to pre-annotate the text using a machine learning model. This can serve as a starting point, which you can then review and modify. Note: Generating predictions will lock the input textbox to prevent accidental edits.
3. Select Text: In the output box below, highlight the portion of the text you wish to annotate.
4. Annotate: Click one of the category buttons to apply the annotation to the selected text. The selected text will be highlighted with the corresponding color.
5. Remove/Undo:
   • Remove (q): Click "Remove" or press 'q' to remove the annotation from the selected text.
   • Undo (z): Click "Undo" or press 'z' to revert the last annotation action.
6. Download/View Annotations:
   • Annotator ID & Discharge Summary ID: Enter an identifier for the annotator and an optional ID for the discharge summary. These IDs will be included in the downloaded JSON data.
   • Refresh Annotations: Click "Refresh Annotations" to display the current annotations in JSON format.
   • Download Annotated Text: Click "Download Annotated Text" to save your annotations as a JSON file.
7. Clear Annotations: Click "Clear Annotations" to remove all annotations and start over. This will also unlock the input textbox.

Keyboard Shortcuts:

• Use the keys indicated on the buttons to annotate selections.
• q: Remove annotation
• z: Undo last action