KIPAN-AI prediction v1 demo

Introduction

An autoplay manuscript-style abstract covering the clinical question, model inputs, MRSS 2.0 output, validation, and example patient workflow.

MRSS for predicting clinical outcome and treatment responseClinical question - Inputs - MRSS 2.0 output - KM validation - Example patient

RCC Model Modules

AI

AI-clinical

Use clinical variables to calculate AI-clinical score and time-specific OS risk.

SUM

MRSS 2.0

Combine WSI score, six-SNP score, and AI-clinical score using the manuscript formula.

Selected

This v1 web build follows the original WSI + XML annotation workflow and produces the manuscript-aligned MRSS 2.0 output.

Connection

Demo Mode

Off - API Mode

API Base URL

Ready.

Multimodal Inputs

H&E WSI + XML Annotation

Select H&E WSI File

Selected H&E WSI: No WSI selected

Select XML Annotation

Selected XML: No XML selected

Use the selected H&E WSI together with a manual XML annotation. Automatic tumor-region outlining is disabled; the result includes a soft attention heatmap when returned by the backend.

WSI AccessUploaded original H&E WSI + XML

Upload ProfileNative WSI

Accepted formatsSVS, TIFF/TIF

Six-SNP

6-SNP score = -0.1650 - 0.0074*SNP1905666 + 0.0633*SNP8366033 - 0.1466*SNP8443830 + 0.0072*SNP8529733 - 0.2123*SNP8535824 + 0.1186*SNP8588520

The displayed six-SNP score is provided by the prediction workflow.

AI-clinical

AI-clinical score is calculated from the entered clinical variables.

WSI Preview

Select and run a WSI to display its preview

Clinical Variables

Age (18-90) Sex Tumor Size (mm) Pathologic Grade T Stage N Stage M Stage Race Histologic Subtype

Matching clinical inputs use the 2026-11-02 table. Unmatched inputs use the AI-clinical prediction model.

Actions

The web app uploads the original WSI and XML first, then calculates MRSS 2.0 with backend WSI scoring.

WSI Score

No WSI score yet. Select the H&E WSI file and XML annotation, then run the v1 workflow.

Prediction

No prediction yet.

Soft Attention Heatmap

No soft attention heatmap yet.

Notes

Displayed component scores, MRSS 2.0 and risk groups are read from the matched 2026-11-02 preset record. WSI inference and heatmaps run on the backend.

This web build uploads the selected original H&E WSI plus XML annotation to the protected backend. Model weights, reference data, and inference code remain server-side.