Israeli AI and machine learning developer, Medial EarlySign, has developed an algorithm platform, called ColonFlag, that can accurately identify patients who are at high risk of having colorectal cancer.
The platform analyses patient blood tests and data from electronic health records (EHR) to identify signs of colorectal cancer often months before clinical signs are present.
A recent peer-reviewed study published in Digestive Diseases and Sciences found that ColonFlag was able to identify risk for colorectal tumours up to 360 days earlier than conventional diagnosis practices.
The analysis was based on data collected from blood samples and demographic information from 17,095 Kaiser Permanente Northwest patients in the US, including a random sample of 900 colorectal cancer patients. ColonFlag analysed patients’ complete blood count results and factored in age and gender to create a colorectal cancer risk score for stratification of each patient.
“Colorectal cancer is the second-highest cause of all cancer-related deaths in the USA, responsible for 49,190 deaths in 2016 alone,” said retired Chief Scientist of Kaiser Permanente Center for Health Research, Mark C. Hornbrook, PhD.
“Early screening for colorectal cancer significantly improves the survival rate. The ability to identify people at high risk for colon cancer and refer them for further testing could help to reduce mortality and prove integral to reducing the overall colorectal cancer burden,” continued Hornbrook.
Colorectal cancer cases identified by ColonFlag were compared with those detected by low haemoglobin (Hgb) levels alone in two adjacent time windows – 0 to 180 and 181 to 360 days prior to diagnosis. For the 0 to 180 day window, ColonFlag demonstrated a 34% and 36% improvement in identifying colorectal cancer cases, compared to low Hgb levels for the 50-75 and 40-89 year old age groups, respectively.
In the 181 to 360 day window, ColonFlag detection was 47% higher for the 50-75 year age group, and 84% higher for the 40-89 year old group.
Further analysis revealed that ColonFlag was successful in detecting tumours in the cecum, a pouch at the beginning of the large intestine, and ascending colon.
“Colonoscopy resources are limited, and screening programmes are being challenged by the need for active participation of asymptomatic individuals,” said Chief Medical Officer of Medial EarlySign, Dr Ran Goshen.
“ColonFlag’s ability to passively identify those at high-risk among non-participating individuals 180 to 360 days prior to diagnosis may allow healthcare providers to better allocate colonoscopy resources,” concluded Dr Goshen.