Colorectal cancer (CRC) is thesecondleading cause of cancer-related deaths worldwide, with 1.93 million new cases diagnosed in 2020 and 48% mortality rate. Most of these deaths could have been prevented by increasing the use of recommended screening tests. CRC has a 5-year survival rate of 90% if diagnosed in Stage I but only 14% if diagnosed in Stage IV.

Current screening methods have limitations, either because they are invasive (colonoscopy), because they have a high false positive rate (fecal tests) or have limited capacity to detect both the tumor and its precursor lesions.

Compliance is the main challenge for CRC screening programs, as only 10% in EU of the eligible population (50-75 years old) is undertaking the necessary screening test. In addition, it has been recently shown that the incidence of CRC in people under 50 who currently do not have access to screening programs is increasing. There is a clear unmet medical need for early, accurate and easy cancer detection.

AMX DATA, analyzing Big Data to identify people at risk

The EU-funded AMX DATA project tries to address this medical need by analysing anonymised Electronic Health Records (EHRs) and identifying eligible individuals with growing risk of developing CRC who are due for screening or who have never been screened for increasing screening rates.

The complete information about a patient’s risk of developing cancer within their history does not exist in one place. It is fragmented from records created by many authors, over many years, in different formats and usually from different institutions where patients have received care over their lifetime.

The current development of language technologies makes possible to address the problem of automatic analysis of EHR with guarantees of success by extracting clinical features that are associated with CRC or advanced adenomas (AAs). Therefore, the Innovation Associate (IA), Miklos Gabriel Tulics, brought its knowledge and experience in the analysis of experimental data. The EHR data used in this study have been anonymised, extracted and analysed  from multiple sources and multiple formats. The goal was to identify risk factors associated with the development of CRC and AA.

Statistical tests were used to compare the characteristics of the clinical features.

AMADIX has a pipeline of innovative and non-invasive tests under development for screening and early diagnostic uses in colon, lung and pancreatic cancer. The company has conducted clinical studies in more than twenty hospitals for different types of cancer. The identified risk factors during the research were compared with the previous findings of AMADIX and with findings from scientific literature, also new hypothesis were formulated for future studies.

AMX DATA, the way to improve CRC diagnostics through AI

The results of this project have been very promising. The next steps will be to develop a new combined algorithm based on risk factors extracted from EHRs and analysed in this project with the previous blood-based test developed by AMADIX.

By incorporating these risk factors in a combined model with molecular biomarkers we will improve the sensitivity without reducing specificity of the test in recognising CRC and AA.

For further information, visit AMADIX website.

Source: European Commission I EISMEA (