---
updatedAt: 2026-08-25T21:20:14.000Z
agentTools:
  projectIndex: https://docs.chartinsight.ai/llms.txt
---

# AI Research

ChartInsight™'s AI Research Assistant lets you have natural language conversations about your medical records, with responses grounded in the actual document content.

## Conversational Interface

### Natural Language Q\&A

Ask questions about your medical records in plain English:

* "What is the patient's treatment history?"
* "Summarize all the imaging studies"
* "What medications were prescribed after the accident?"
* "Are there any references to physical therapy?"

### HIPAA Compliance

* All conversations are HIPAA compliant
* End-to-end encryption for research data
* No conversation data used for model training

***

## Research Topics

### Multiple Topics

* Create **multiple research topics** per record
* Organize different research questions separately
* Each topic maintains its own conversation history

### Research History

* **Grouped by date**: Last 30 days, then by month
* **Sidebar navigation** for quick access to past research
* **Most recent topics** appear at the top

### Topic Management

* **Rename topics** with custom titles (up to 120 characters)
* **Delete topics** when no longer needed
* **New Topic** button to start fresh conversations

***

## Context & Citations

### Page-Aware Responses

* AI responses include **citations to source pages**
* Click citations to navigate directly to the referenced page
* Multiple citations when information spans pages

### Context Panel

* **Resizable sidebar** showing relevant document excerpts
* **Highlighted passages** from the source document
* **Synchronized navigation** between research and PDF viewer

### Source Verification

* Every claim can be traced back to the source document
* Page numbers provided for easy verification
* Context excerpts show surrounding text

***

## Feedback System

### Message Rating

* Rate AI responses for quality
* Feedback helps improve future responses
* Simple thumbs up/down interface

### Continuous Improvement

* User feedback incorporated into model improvements
* Common questions inform prompt optimization
* Error patterns help refine extraction accuracy

***

## Best Practices

### Effective Questions

* Be specific about what you're looking for
* Include relevant time periods or providers
* Ask follow-up questions to dig deeper

### Example Workflows

**Legal Research:**

1. "What injuries are documented in this record?"
2. "When did treatment begin for the back injury?"
3. "What did Dr. Smith say about the prognosis?"

**Medical Review:**

1. "Summarize the patient's cardiac history"
2. "What medications is the patient currently taking?"
3. "Are there any contraindications mentioned?"

**Insurance Analysis:**

1. "What procedures were performed?"
2. "How long was the hospital stay?"
3. "What follow-up care was recommended?"

***

## Technical Details

### Response Generation

* Responses are generated based on the processed record content
* Vector similarity search finds relevant passages
* Large language model synthesizes coherent answers

### Limitations

* Cannot answer questions about content not in the record
* Always verify critical information against source documents