← Back to NorthLaw.ai

🇨🇦 Canadian SST Outcome Predictor

Social Security Tribunal of Canada - CPP, OAS & Employment Insurance Appeals

0.55
MCC Score
87.2%
AUC Score
13,727
Training Cases
LightGBM v3 (middle-text, isotonic calibrated)
Algorithm
2013–2024
Training Period
Data source: Sean Rehaag & Simon Wallace, “A2AJ Canadian Legal Data” (2025), online: GitHub https://github.com/a2aj-ca/canadian-legal-data (updated 2026).

📊 Confidence Tiers (based on 3,768 cases)

High (≥75%): 90.2% accuracy - Strong predictive signal
Medium (60-75%): 68.3% accuracy - Moderate confidence
Low (<60%): 56.2% accuracy - Consider additional factors
How to use: Paste the CPP disability appeal summary, OAS claim details, or EI appeal grounds. The model analyzes language patterns from 13,727 SST decisions to predict appeal outcomes.
0 characters Min: 250 | Ideal: 1500-3500 | Max: 5000
💡 Not sure what to paste? The example cases below show the kind of detail this tool works best with — facts, evidence, legal issues, lower decision.
Try examples:

For Legal Professionals

Take this prediction further with the NorthLaw professional toolkit:

📝 Strategic Memo
Cite-rich brief tailored to your matter
🔬 Multi-Tool Analysis
Cross-reference predictions across tools
⚖️ Opposing Counsel
See how the other side could argue
🧠 SST Knowledge Graph
174 members, collection analytics, trends
Learn more about Pro →

📊 Value Calculator for Lawyers

For legal professionals: estimate the value this tool brings to your practice. Drag the sliders and explore three ways to think about ROI.

80 cases
$350/hr

You spend fewer hours on outcome assessment and redirect that time to other billable work.

Hours freed/year
0
Revenue from freed hours
$0
Tool cost/year
$0
Return on investment
0x

Formula: hours freed = cases × 1.5 hrs  |  revenue = hours freed × hourly rate  |  tool cost = cases × $0.10

Assumes 1.5 hours saved per case on SST outcome research. Manual assessment involves reviewing tribunal decisions, disability evidence, and procedural history — this tool delivers a calibrated prediction in under 10 seconds. Tool cost assumes $0.10 per analysis. Average case load of 12 hours used for capacity calculation.