The Index of Suspicion project uses machine learning and computational methods to explore national prescribing data to try to identify patterns in prescribing prior to cancer diagnosis. Such patterns may act as an early signal of cancer diagnosis and could contribute to diagnostic tools to diagnose cancer earlier.
Early diagnosis of cancer could increase survival, improve quality of life, and decrease healthcare costs. One possible source of early indicators of cancer is pre-diagnostic prescribing patterns. These may contain drugs that reflect early signs of an undiagnosed cancer. Previous studies in Denmark have found increases in incident prescribing of several drug groups prior to cancer diagnosis at particular sites.
In 2015, Public Health England’s National Cancer Registration and Analysis Service (NCRAS) entered into a partnership to form an anonymised dataset of all prescriptions dispensed in community pharmacies in England linked to cancer registry data. A pseudonymisation process was used to protect the identities of patients. Health Data Insight is working with NCRAS to analyse this dataset using machine learning techniques and computational algorithms. Machine learning has the potential to identify complex, non-linear relationships between high numbers of variables. Through combinations of these and more classical statistical approaches, we hope to build upon the work of the Danish studies to build a detailed understanding of pre-diagnostic prescribing and find potential signs of cancer.