Professor of Cancer Genetics and Surgery, University of Birmingham; Deputy Director @ Birmingham CRUK ECMC
Interesting news from the NHS England GRAIL Galleri test in the UK - accuracy in the article given as 56% One of the main issues I think with Galleri is its model trained on late stage disease, and is a “fixed” panel. By comparison look at the recent results from Guardant Health and Exact Sciences where their CRC early detection assays pushed > 80% accuracy (COI - I am co-I in Guardant Health CORRECT-MRD-1) We need broader WG assays trained on early disease - we are developing a Nanopore assay with Isidro Cortés Ciriano which combines SNV, CNV, SV, Fragmentomics, Motif analysis and epigenetics (5-mC and 5-hMC) to build the best model possible. https://lnkd.in/eF8QN-q4
Steve Smith
2mo
A wise decision by the NHS. There are further technologies to be explored which offer far more accurate and reliable results. Cancer can be detected in blood and you do not need the presence of a tumour to find it.
Asst Professor of Bioinformatics & ML | Co-Founder at MedAiHealth | Entrepreneur | Inclusive AI | Health Inequality | Human Centric AI
3moThis is very interesting. I remember their results published in ESMO 2020. There was a stark difference in the performances between stage I-II and stage III-IV cancers. The number of samples in the test sets across different cancer types were also quite low.