AI Impressions & Follow-up
AutoScribe uses AI to generate impressions, follow-up recommendations, and longitudinal comparisons to accelerate your radiology workflow.
Introduction
Section titled “Introduction”AutoScribe includes three powerful AI-generated enhancements that are seamlessly integrated into every report template: AI-Generated Impressions, AI Follow-up Appointment Recommendations, and AI Comparisons. These features are extremely intuitive and baked into every report at the beginning, providing radiologists and diagnostic readers with instant, high-quality report summaries.
These AI-generated report summaries are generated near instantaneously and serve a singular purpose: to dramatically speed up radiologists and other diagnostic readers by providing consistent, high-quality AI-generated report summaries. While humans performing repetitive tasks like summarizing reports are prone to errors, AI assistance provides a massive speed boost to reader velocity and throughput while reducing error rates. This allows readers to focus on what really matters — diagnostic reading and providing high-quality care.
Each of these enhancements can be generated in two ways:
- Voice commands: Simply say “Generate Impressions”, “Generate Next Appointments”, or “Generate Comparison”
- Button clicks: Click the generate button in each respective section
All AI-generated enhancements are entirely optional. Readers who prefer to dictate these sections manually can do so with no degradation in reading experience — AutoScribe works seamlessly whether you use AI assistance or traditional dictation.
AI-Generated Impressions
Section titled “AI-Generated Impressions”AI-Generated Impressions automatically extract and summarize the key radiological findings from your report content into concise, clinically meaningful statements.
How It Works
Section titled “How It Works”The AI analyzes your entire generated report content and identifies clinically significant findings (mild, moderate, and severe), consolidating related findings when appropriate. It generates key report facts and summaries based on the reader’s generated report content, preserving precise measurements, anatomical terminology, and qualitative assessments while maintaining professional radiological language.
AI-Generated Impressions work out of the box with no configuration required. To generate impressions:
- Voice command: Say “Generate Impressions”
- Button click: Click the generate button in the Impressions section
The impressions are generated instantly and can be edited, deleted, or supplemented with additional manually dictated content as needed.
AI Follow-up Appointment Recommendations
Section titled “AI Follow-up Appointment Recommendations”AI Follow-up Appointment Recommendations analyze your report findings and generate clinically relevant follow-up recommendations based on the reader’s generated content.
How It Works
Section titled “How It Works”The AI identifies moderate or severe findings that specifically mention or imply follow-up action, then generates professional recommendations for appropriate next steps such as imaging intervals, clinical correlation, or specialist referrals. Related findings are consolidated into unified recommendations, prioritizing actions that impact patient management or safety.
AI Follow-up Appointment Recommendations work out of the box with no configuration required. To generate follow-up recommendations:
- Voice command: Say “Generate Next Appointments”
- Button click: Click the generate button in the Follow-up section
The recommendations are generated instantly and can be edited, deleted, or supplemented with additional manually dictated content as needed.
AI Comparisons
Section titled “AI Comparisons”AI Comparisons generate longitudinal comparisons of findings from the current study and previous studies, tracking changes, resolutions, and new findings over time.
How It Works
Section titled “How It Works”The AI performs comparative analysis between the current study and prior studies, identifying:
- Absent findings: Findings from prior studies that are not visualized or have resolved
- New findings: Findings present in the current study but not in any prior study
- Longitudinal changes: Findings present in both prior and current studies, characterized by trajectory (improved, stable, or progressed) with quantitative comparisons when measurements are available
The comparison focuses only on anatomical regions that were actually scanned in the current study, excluding comparisons for regions not covered by the current imaging protocol.
Configuration Requirements
Section titled “Configuration Requirements”Unlike Impressions and Follow-up Recommendations, AI Comparisons require prior study context to function. These requirements are seamlessly handled by PACS integrators during initial setup, system integration, and study creation, ensuring that prior study data is automatically available for comparison:
- Prior studies in AutoScribe: If previous studies exist in your AutoScribe system, the prior study IDs are automatically provided when creating the current study
- External prior studies: If prior studies are from external systems, text representations of previous reports are automatically entered via the SDK when creating studies so that the AI has sufficient context to generate longitudinal comparisons
To generate comparisons:
- Voice command: Say “Generate Comparison”
- Button click: Click the generate button in the Comparison section
If no prior studies are available or configured, the system will indicate that no comparisons can be generated. When prior studies are available, the comparison is generated instantly and can be edited, deleted, or supplemented with additional manually dictated content as needed.