# CDC Disease Sentinel AI Agent Connect

> CDC Disease Sentinel monitors the CDC's National Notifiable Diseases Surveillance System. It gives your agent the ability to detect real-time spikes, compare current case counts against yearly baselines, and map disease concentration across states or census regions using live, public health data.

## Overview
- **Category:** government-public-data
- **Price:** Free
- **Endpoint:** https://edge.vinkius.com/vk_preview_oaV2lbFOA8mKmOR6AkStoGf12ybjdkC8Ly37nASm/ai-agent-connect
- **Tags:** cdc, diseases, outbreaks, public-health, nndss, surveillance, sentinel, health-data

## Description

You can use this MCP to turn your AI client into a public health monitor. Instead of digging through messy government spreadsheets, you just ask your agent if a specific disease is spiking in a certain state. The MCP pulls directly from the CDC's NNDSS, providing the latest weekly case counts and comparing them against historical averages. You can check if current numbers are normal, elevated, or a full surge based on a configurable multiplier. It handles the heavy lifting of calculating year-to-date comparisons and 52-week baselines, so your agent can provide a clear verdict on whether a situation is unusual. Whether you need to see the momentum of a single disease over the last few weeks or find out which states are seeing the most cases right now, this tool provides the exact strings and data points your agent needs to give you an accurate answer.

## Tools

### list_diseases
This tool provides the exact names of all diseases currently tracked by the CDC's weekly surveillance system.

### disease_baseline
This tool answers if current disease levels are unusual by comparing the latest week to the 52-week maximum and YTD totals to the same period last year.

### disease_trend
Use this to see the shape of a disease curve by retrieving weekly case counts for a specific disease over a set number of weeks.

### state_scan
This tool ranks all reporting areas by case count for a specific disease to show where cases are most concentrated.

### surge_check
This tool detects spikes by comparing the latest weekly count against the average of previous weeks to return a surge, elevated, or normal verdict.

## Prompt Examples

**Prompt:** 
```
Is there a disease surge this week anywhere in the US?
```

**Response:** 
```
The surge check identifies spikes in tracked diseases and reports the multiplier verdicts for each.
```

**Prompt:** 
```
Show me the last 16 weeks of national meningitis cases and say if it's trending up.
```

**Response:** 
```
The agent retrieves the 16-week series for meningitis and analyzes the count momentum.
```

**Prompt:** 
```
Which states report the most norovirus this week?
```

**Response:** 
```
The agent scans all reporting areas and lists the states with the highest norovirus case counts.
```

## Capabilities

### Surge Detection
Your agent identifies when disease counts cross specific multipliers to flag unusual activity.

### Trend Analysis
The MCP provides weekly case series so your agent can interpret the direction of a disease curve.

### Geographic Mapping
Your agent can pinpoint exactly which states or regions are seeing the highest case concentrations.

### Historical Comparison
The tool compares current year-to-date totals against the same period from the previous year.

## Use Cases

### Outbreak Monitoring
An agent can automatically flag when a disease hits a surge threshold in a specific region.

### Comparative Research
Researchers can compare current YTD case counts against last year's data to spot shifts.

### Regional Health Reports
Users can generate lists of states with the highest reported cases for a specific week.

### Trend Verification
Verify if a disease is actually climbing or just fluctuating within normal limits.

## Benefits

- Eliminates manual data retrieval from CDC portals.
- Provides exact area and disease names for error-free follow-up queries.
- Computes surge multipliers and baselines server-side.
- Uses live, stateless data for the most current information.

## How It Works

The MCP connects your AI client directly to the CDC's surveillance system.

1. Connect to the MCP through your compatible AI client.
2. Ask your agent a question about disease counts or surges.
3. The MCP pulls the latest data from the CDC NNDSS.
4. The tool calculates trends, surges, or baselines based on your request.
5. Your agent receives the data and answers your question.

## Frequently Asked Questions

**Do I need a CDC account to use this MCP?**
No. The MCP uses public CDC open data, so no credentials or accounts are required.

**How does the surge detection work?**
The surge check compares the most recent weekly count against the average of the previous weeks to determine if the activity is normal, elevated, or a surge.

**Can I check data for a specific state?**
Yes. You can use the state scan or specific disease tools to look at national, state, or census region data.

**Is the data real-time?**
The MCP pulls from the latest weekly NNDSS case counts provided by the CDC.

**What happens if there are no cases reported?**
The trend tool shows no data for weeks without reported cases rather than assuming a zero.

**Do I need credentials?**
No. The dataset is public CDC open data on Socrata, readable without a token (anonymous access is throttled and capped at 500 rows per query — every tool stays well under that). No signup, no key, no per-user state.

**How does the surge detector decide "surge"?**
It divides the latest week's reported cases by the average of the previous N weeks (default 12, configurable). At or above the threshold multiplier (default 2.0) it reports a surge; at half the threshold it reports elevated; special cases (no cases this week, or a baseline with zero cases) are called out explicitly instead of producing a division artifact.

**Why do the tools insist on "exact" disease and area names?**
The CDC portal's query engine only matches values exactly (fuzzy matching silently returns nothing), so list_diseases returns the exact labels to copy, and state_scan returns the exact area names — follow-up calls just copy one of them instead of guessing. If a name doesn't match you get an error that says so, never an empty silent answer.

**What data is behind it?**
CDC's NNDSS weekly surveillance: reported notifiable-disease case counts per MMWR week, per geography (national, regions, states, New York City), plus the dataset's own 52-week maximum and YTD comparators. Weeks without reported cases are shown as "no data", not as zeros. MMWR years run July-to-June.