[COLONSCORE: THE USE OF MACHINE LEARNING OF BIG DATA TO DETECT COLORECTAL CANCER]

Cohn Adulamy M, Shalev V

Harefuah 2018 Oct;157(10):634-637

PMID: 30343539

Abstract

INTRODUCTION: The use of big data is in its first years of entering the medical world. Big data research enables analysis of very large volumes of data, identifying patterns and findings which traditional statistical methods cannot handle. The diagnosis of colorectal cancer is often missed due to people’s non-adherence to the occult fecal blood screening tool. We describe the case of a patient that led to the development of a novel big data algorithm to diagnose colorectal cancer. A 70 year old, previously healthy man, was diagnosed with metastatic colorectal cancer and succumbed to his illness. He skipped his fecal occult blood screening tests. His blood counts over the previous years showed a steady decrease in hemoglobin, still within the normal range, 3 years before his diagnosis. This trend was confirmed in a large epidemiological study, which has led to the development of a novel algorithm for the diagnosis of colorectal cancer based on repeated blood tests. The use of algorithms created based on the analysis of big data is a new field in medicine. In this case, the “colonscore” algorithm is being applied in a large health program in Israel, and in its first year of operation it has identified cases with higher specificity and sensitivity than the fecal occult blood test.