iiinova — where your digital self keeps evolving

Reposted by Anders Skanderup

bioRxivpreprint @biorxivpreprint.bsky.social · 24/06/2026

Systematic benchmarking of multi-modal approaches for tumor-naive ctDNA detection and quantification www.biorxiv.org/content/10.64898/20…

021

Reposted by Anders Skanderup

kiranchari.bsky.social @kiranchari.bsky.social · 17/04/2026

Our latest AI model for cancer mutation detection, VarNet-T, is now in Nature Communications! This end-to-end method works on tumor-only data, boosting accuracy by 20–33% and improving TMB-high classification by >3x. 🔬 Paper: www.nature.com/articles/s41... cc @skandlab.bsky.social

011

Anders Skanderup @skandlab.bsky.social · 28/12/2025

A great summary of our new method Fragle - enabling low-cost detection and quantification of cancer DNA from blood samples research.a-star.edu.sg/articles/hig... www.nature.com/articles/s41...

research.a-star.edu.sg

Tracking down cancer’s crumbs - A*STAR Research

A machine learning model called Fragle helps detect cancer and spot signs of relapse by quantifying DNA fragments in the blood.

000

Reposted by Anders Skanderup

Tina Han @tingfordha.bsky.social · 22/02/2025

“Comprehensive benchmarking of methods for mutation calling in circulating tumor DNA” with 2000x WES or 150x WGS. From the lab of @skandlab.bsky.social at Genome Institute of Singapore www.biorxiv.org/content/10.1...

131

Reposted by Anders Skanderup

A*STAR Genome Institute of Singapore (A*STAR GIS) @astar-gis.bsky.social · 17/06/2025

🔬 @astar-gis.bsky.social Researchers - @skandlab.bsky.social unveils Fragle, a novel AI-driven method that significantly enhances the speed, accessibility, and affordability of cancer tracking — by analyzing DNA fragment sizes from a blood test. 📄Read more 👉 www.a-star.edu.sg/gis/home/pre...

a-star.edu.sg

NEW AI METHOD MAKES CANCER TRACKING FASTER AND EASIER USING BLOOD TESTS

021

Reposted by Anders Skanderup

Matt Jensen @mattjensen.bsky.social · 03/12/2024

MSK's Schulz Lab has a great Nature paper combining NLP-abstracted clinical records with tumor sequence data at scale. The combination ("MSK-CHORD") improves outcome prediction and suggests some of AI's potential. 🧪 The paper: www.nature.com/articles/s41... Justin Lee's blog post: bit.ly/3UVGffe

An overview of the MSK-CHORD paper's ideas, this figure contains four subfigures.
a. MSK-CHORD creation overview.
b. NLP model performance and errors audited.
c. MSK-CHORD characteristics and survival data.
d. Patient data visualization in cBioPortal.

0104