# Joint multi-task learning improves weakly-supervised biomarker prediction in computational pathology

Source: [Joint multi-task learning improves weakly-supervised biomarker prediction in computational pathology](<https://www.sota2.com/research/paper/68fa72771e209131cea56894>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-04T04:42:56.982Z

Authors: Omar S. M. El Nahhas, Georg W\\"olflein, Marta Ligero, Tim Lenz, Marko van Treeck, Firas Khader, Daniel Truhn, Jakob Nikolas Kather

Published: 2024-03-06T17:51:04.000Z

## Abstract

Deep Learning (DL) can predict biomarkers directly from digitized cancer histology in a weakly-supervised setting. Recently, the prediction of continuous biomarkers through regression-based DL has seen an increasing interest. Nonetheless, clinical decision making often requires a categorical outcome. Consequently, we developed a weakly-supervised joint multi-task Transformer architecture which has been trained and evaluated on four public patient cohorts for the prediction of two key predictive biomarkers, microsatellite instability (MSI) and homologous recombination deficiency (HRD), trained with auxiliary regression tasks related to the tumor microenvironment. Moreover, we perform a comprehensive benchmark of 16 approaches of task balancing for weakly-supervised joint multi-task learning in computational pathology. Using our novel approach, we improve over the state-of-the-art area under the receiver operating characteristic by +7.7% and +4.1%, as well as yielding better clustering of latent embeddings by +8% and +5% for the prediction of MSI and HRD in external cohorts, respectively.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [HRD prediction on TCGA-LUAD](<https://www.sota2.com/research/sota/hrd-prediction-on-tcga-luad.md>): HRD prediction; TCGA-LUAD; AUC: 0.734.

- [HRD prediction on CPTAC-LUAD](<https://www.sota2.com/research/sota/hrd-prediction-on-cptac-luad.md>): HRD prediction; CPTAC-LUAD; AUC: 0.861.

- [MSI prediction on CPTAC-CRC](<https://www.sota2.com/research/sota/msi-prediction-on-cptac-crc.md>): MSI prediction; CPTAC-CRC; AUC: 89.9.

- [Mutation Prediction on TCGA-CRC](<https://www.sota2.com/research/sota/mutation-prediction-on-tcga-crc.md>): Mutation Prediction; TCGA-CRC; AUC: 94.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2403.03891>)

[PDF](<https://arxiv.org/pdf/2403.03891.pdf>)

[Implementation](<https://github.com/Avic3nna/joint-mtl-cpath>)