Cigota MKB-10 Clinical Classifier
Performance
Telemetry & Telemetry
Rigorously benchmarked evaluation metrics for the Cigota MKB-10 Clinical Classifier, trained on 105,019 real clinical protocols across 20 diagnostic classes.
Evaluation Metrics
Global Performance
Held-out test set evaluation (n = 20,301).
TELEMETRY
Top-1 Accuracy
48.9%
TELEMETRY
Top-3 Accuracy
81.0%
TELEMETRY
Weighted F1
0.467
TELEMETRY
Macro F1
0.348
Granular Analysis
Per-Disease Breakdown
Top 5 diagnoses by volume with corresponding F1 scores and support validation.
C73
Thyroid CancerF1 0.918
Support: 1,062 test rows
E040
Struma diffusaF1 0.604
Support: 6,352 test rows
E039
HypothyreoidismusF1 0.563
Support: 2,214 test rows
E050
ThyreotoxicosisF1 0.360
Support: 2,123 test rows
E030
Congenital hypothyroidF1 0.217
Support: 1,794 test rows
Data Engineering
Dataset Composition
Sovereign training corpus metrics across institutional endpoints.
Total Rows
105,019
Unique Classes
20 MKB-10
Institutions
Cigota · Cigotica · NM
Direct Technical Inquiry
Deploy Sovereign Machine Learning
Interested in deploying clinical AI or scaling data infrastructure? Let's talk.
Company Info
- Legal: Antikitera d.o.o.
- PIB: 114731407
- MB: 67811305
- HQ: Serbia