MCP Protocol Compatible

Clinical calculations
your AI can trust

Validated medical scores and lab formulas. One API call away.

⚡ Blazing Fast
✓ Peer-reviewed accuracy
🔒 No PHI stored
Rust Rust
Stripe Stripe
MCP MCP
PubMed Validated
Security No PHI
The Problem

LLMs hallucinate
medical calculations.

Ask any AI to calculate a clinical score. You'll get a confident answer—but it's often wrong. In healthcare, wrong answers mean wrong treatments.

⚠️

Hallucinated Results

Models guess at formulas instead of computing them

🎲

Inconsistent Outputs

Same patient data, different answers every time

📅

Outdated Guidelines

Training data frozen in time, missing current standards

The Solution

Deterministic tools
for non-deterministic AI.

Ferritin gives your AI agent validated, peer-reviewed clinical calculators. Every call returns the same answer—the right answer.

📚

Traced to Source

Every formula linked to its original peer-reviewed publication

🎯

Always Consistent

Same inputs produce identical outputs, every single time

🔄

Always Current

Updated when guidelines change, not when models retrain

The Economics

Why build it yourself?

Building In-House

  • 100+ formulas to implement
  • Validation test suites
  • Server infrastructure
  • Ongoing maintenance
  • ~$50k+ / year

Ferritin

  • 1,000 calls free / month
  • Then $0.001 per call
  • No servers to manage
  • Always up to date
  • ~$10 / month typical
Built Different

Rust. MCP. Native Speed.

Not a chatbot. Not an LLM wrapper. A clinical calculation engine built for the AI-native future.

🦀 Memory Safe ⚡ Native Performance 🔌 MCP Native 🔒 Zero PHI Storage

Clinical Tools

Validation

Every calculation.
Traced to source.

We don't guess. Every formula is validated against peer-reviewed medical literature.

1

Literature Review

Primary source from peer-reviewed journals

2

Reference Implementation

Algorithm matches original publication exactly

3

Unit Testing

500+ deterministic test cases

4

Cross-Validation

Compared against MDCalc, UpToDate

5

AI Agent Testing

LLM-generated edge case verification

53 Clinical Scores
53 Lab Formulas
12 Categories
500+ Test Cases

Sample Citations

CHA₂DS₂-VASc
Lip GY et al. Chest 2010;137(2):263-272
PubMed →
Glasgow Coma Scale
Teasdale G, Jennett B. Lancet 1974;2(7872):81-84
PubMed →
CURB-65
Lim WS et al. Thorax 2003;58(5):377-382
PubMed →
HEART Score
Six AJ et al. Neth Heart J 2008;16(6):191-196
PubMed →

Simple Pricing

Quick Start

Get up and running in under 2 minutes

1 Get Your API Key

Sign up above or use curl:

bash 
curl -X POST https://api.ferritin.dev/v1/billing/signup \
  -H "Content-Type: application/json" \
  -d '{"email": "you@company.com"}'
Response 
{
  "customer_id": "cus_abc123xyz",
  "message": "Welcome! 1,000 free calls/month"
}
2 Configure Your AI Client

Add to Claude Desktop, Cursor, or any MCP-compatible client:

~/.config/claude/claude_desktop_config.json 
{
  "mcpServers": {
    "ferritin": {
      "transport": "sse",
      "url": "https://api.ferritin.dev/mcp/sse",
      "headers": {
        "Authorization": "Bearer cus_YOUR_ID"
      }
    }
  }
}
3 Start Using Clinical Tools

Ask your AI assistant naturally:

💬 Example Prompts 
# Risk Scores
"Calculate CHA₂DS₂-VASc for a 75-year-old female with CHF and HTN"

# Lab Formulas
"What's the corrected calcium if Ca is 8.5 and albumin is 2.8?"

# Complex Workflows
"Assess this patient's stroke risk and bleeding risk together"

⚠️ Important Notice

In summary: These tools are for healthcare professionals only. Always verify calculations and use clinical judgment. Not a medical device.