API input variables

Sex: F, M
Age: Numeric variable
Diameter: Numeric variable
Composition: Solid, p-Solid, p-Cystic, Cyst
Echogenicity: MarkedHypo, MildHypo, Isoecho, Hyperecho
MacroCalcification: None, Presence, EntirelyCalcified
RimCalcification: None, Presence
Margin: Smooth, Spiculated_Microlobulated, ill-defined
Punctate: None, Presence
Shape: Parallel, NP

Endpoint

GET https://tirads.cdss.co.kr/Predict.php — all variables are passed as URL query parameters. The request is stateless and idempotent, so a simple GET is the natural fit: a result can be reproduced, bookmarked, or shared just by keeping the URL.

1. Browser / plain GET

https://tirads.cdss.co.kr/Predict.php?&Sex=F&Age=32&Diameter=32&Composition=Solid&Echogenicity=MildHypo&MacroCalcification=EntirelyCalcified&RimCalcification=Presence&Margin=ill-defined&Punctate=Presence&Shape=Parallel

2. curl

curl -G "https://tirads.cdss.co.kr/Predict.php" \ --data-urlencode "Sex=F" \ --data-urlencode "Age=32" \ --data-urlencode "Diameter=32" \ --data-urlencode "Composition=Solid" \ --data-urlencode "Echogenicity=MildHypo" \ --data-urlencode "MacroCalcification=EntirelyCalcified" \ --data-urlencode "RimCalcification=Presence" \ --data-urlencode "Margin=ill-defined" \ --data-urlencode "Punctate=Presence" \ --data-urlencode "Shape=Parallel"

3. Python

import requests params = { "Sex": "F", "Age": 32, "Diameter": 32, "Composition": "Solid", "Echogenicity": "MildHypo", "MacroCalcification": "EntirelyCalcified", "RimCalcification": "Presence", "Margin": "ill-defined", "Punctate": "Presence", "Shape": "Parallel", } r = requests.get("https://tirads.cdss.co.kr/Predict.php", params=params) print(r.json()) # ['81.35/4 (10-40%)']

Response format

A JSON array with one string: ["81.35/4 (10-40%)"]
81.35 — AI-predicted malignancy probability (%)
4 — K-TIRADS category
(10-40%) — expected malignancy-rate range for that K-TIRADS category

Unknown values for a categorical variable return an error — use exactly the levels listed above.