A research team at Samsung Medical Center (SMC)’s Department of Internal Medicine has developed a self-diagnosis prediction model to confirm the risk of developing advanced adenoma that causes colon cancer.
This is the first time that a disease prediction model using big data technology in digestive diseases has been developed. It will likely be helpful for early diagnosis and management of adenomatous colon diseases, the hospital said.
Professors Rhee Poong-lyul이풍렬, Son Hee-jung손희정, and Hong Sung-noh홍성노 used an extensive data analysis of about 50,000 patients who underwent colonoscopy at the SMC between 2003 and 2012.
|(From left) Professors Rhee Poong-lyul, Son Hee-jung, and Hong Sung-noh at Samsung Medical Center.|
The research team first worked on quantifying the medical records of 49,450 patients who visited the hospital during the study period. It used descriptive data such as patient's age, gender, and various test scores, as well as image-based diagnostic readings, to extract and convert into numerical values, regardless of the method of presentation.
Based on this analysis, the researchers analyzed the risk factors, including age, gender, smoking habit, the frequency of alcohol use, and aspirin use, and indexed them.
As a result of evaluating the effectiveness of the new model compared to the existing model, the Area under the Curve (AUC) was 71.6 percent. The confidence interval is even higher than the 67.8-percent hit ratio of the previous model.
The prediction model created by the research team is designed to vary in value depending on the risk factors. If the final value calculated based on the values corresponding to the different risk factor results is lower than -4.195, it is classified as a low-risk group.
In this study, high-risk patients were 3.8 times more likely to have adenoma than low-risk patients. In the prediction model, if a patient belongs to a high-risk group, he or she would need early risk management such as colonoscopy.
"It is significant that we made a risk prediction model by quantifying informal medical records and doing big data analysis,” said Professor Rhee. "We expect to be able to prevent and measure the risk of colon adenoma that can lead to colorectal cancer.”