An international research team led by Taipei Medical University employed artificial intelligence (AI) in the identifying of high cancer risk groups through blood data obtained from general health examinations. The study was published on Scientific Reports, a journal published by Nature Research, on March 16, 2020.
This research, said Shabbir Syed Abdul, an associate professor at the Graduate Institute of Biomedical Informatics of Taipei Medical University, had been mainly conducted through AI, employing machine learning algorithms in the screening of cell population data (CPD) for hematologic malignancies.
The research team collected a total of 882 hematology-oncology cases from Konkuk University Medical Center, Seoul, South Korea, among which 457 cases were of hematologic malignancies and 425 cases were of hematologic non-malignancies. Then seven models, including SGD, SVM, ANN, linear model, and logistic regression, were employed in AI learning. AI was further used to screen the data obtained from the blood samples collected from the hematology-oncology case subjects; the highest diagnosis rate of 93.5% was achieved by ANN.
Researchers from South Korea, Slovenia, and Saudi Arabia have jointly participated in the study. Associate Professor Shabbir Syed Abdul explained that as blood cancer is harder to diagnose than other cancers and usually requires the combination of blood smear and bone marrow smear examinations, many cancer patients are often diagnosed when cancer had already progressed to the middle or advanced stages, resulting in missed optimal treatment timings.
The new screening method can facilitate early risk detection through blood tests in routine health examinations of patients and timely responses, demonstrating promising research results.
The results of the study will enable medical facilities to detect the risk levels of the future development of hematologic malignancies, such as lymphoma and blood cancer, through blood test data and AI analysis without the requirement of additional examinations, reducing costs of human resources, medical expenses, time, etc. Moreover, as early diagnoses and treatments can be achieved, the effects on the lowering of cancer mortality rates are highly anticipated.