Many people present to their doctor with vague symptoms that may represent cancer or pre-cancerous abnormalities. Many of these are referred to hospital for invasive testing, such as with endoscopy or colonoscopy. There may be hidden patterns within the general health profiles of individuals that could identify those at high or low risk. This research project will use machine learning techniques to identify previously hidden health patterns within the large UK Biobank cohort. The aim is to stratify risk profiles completely non-invasively. If successful. these novel profiles could be used by doctors to target invasive tests to those people who need them and help people who do not need invasive tests avoid having them.