10083274AI-VISION: An observational study validating a predictive algorithm integrating multi-modal data for patient prognostication and treatment stratification in Triple Negative Breast Cancer.ClosedCollaborative R&DInnovate UK**Background:** Despite improvements in the treatment of Triple Negative Breast Cancer (TNBC), the cancer returns in half of the women and shockingly 40% are dead within 5 years of their initial cancer diagnosis. There is an urgent need to identify reliable biomarkers of response for chemotherapy and immunotherapy. In this study we will apply our company's computational approach to update our existing algorithm. The new algorithm will predict treatment response and future risk of cancer spreading. **Aims:** To update our existing predictive algorithms specifically for use in women newly diagnosed with TNBC. New added features will include predicting response to chemotherapy with or without immunotherapy treatment, and how an individual's prognosis could change if treatment A is chosen over treatment B. **Study Design:** This is an observational clinical research study planning to collect clinical data and genetic data from patient samples, with their consent over 2 years. This project is divided into 2 stages: a retrospective stage, where the events have already taken place and outcomes are known, and a prospective stage, where patients will be followed. **Retrospective stage:** Historical data will be collected from 200 women previously treated for early breast cancer (stage 2-3) in the NHS to train the algorithm. Concr has UK ethics approval for this part. **Prospective stage:** The algorithm will predict future events, before they have occurred for 50 women with a new diagnosis of TNBC. This will test the algorithm's reliability in real time. To overcome potential biases in the data, the algorithm will initially be trained using retrospective data from an equal number of women who responded (n=100; Arm A) and did not respond to chemotherapy (n=100; Arm B). **Data types:** We plan to collect specific clinical information, like age, ethnicity, cancer stage at diagnosis, cancer treatment, drug response and if the individual is still alive. **Genetic testing**: All patients in this study will have their cancers tested to capture tumour biology. For gene analyses (DNA and RNA) archival formalin fixed paraffin embedded cancer biopsies will be retrieved. This data will be used to train the algorithm and will not be shared with study participants because the significance of many genes is yet unknown. There are no extra biopsies required because feedback from 20 women as part of public engagement highlighted their preference not to have extra research biopsies, but a rather willingness to donate existing clinical samples for research.