Diagnosing Leukemia with Artificial Intelligence

2019-12-25

Artificial intelligence is a hot topic in the field of medicine, especially in the field of diagnostics. It can detect a common type of blood cancer, acute myeloid leukemia, by analyzing the gene activity of cells found in the blood. The accuracy of this method is quite high. In practice, this method can support traditional diagnostics and expedite the start of treatment. Researchers from DZNE and the University of Bonn published their findings in the journal "iScience" after conducting conceptual research. In their paper, they stated, "Our goal was to investigate the potential of this method based on a specific example. To do this, we evaluated a large amount of data on blood cell gene activity, collecting almost all available information." The researchers focused on the fingerprint of gene activity, known as the transcriptome. In each cell, only specific genes are "activated" based on the cell's own conditions, which is reflected in their gene activity profile. In the current study, researchers analyzed accurate data spanning thousands of genes extracted from blood samples. The transcriptome holds crucial information about the cell's condition and is the key to integrating artificial intelligence. Researchers can train a specialized algorithm by analyzing the transcriptome. Leukemia has a high mortality rate, and without effective treatment, patients can die within weeks. Therefore, the earlier it is detected, the higher the survival rate. The researchers inputted partial data into their algorithm, including whether the sample came from a leukemia patient. The algorithm then searched for patterns of specific diseases in the transcriptome. This process is largely automated. Based on this pattern recognition, the algorithm analyzed and classified further data, even distinguishing between samples with and without acute myeloid leukemia (AML). The success rate of the algorithm exceeded 99%. If implemented, this method can support routine diagnostics and help save costs. In principle, blood samples can be collected by general practitioners and sent to laboratories for analysis, with a cost not exceeding 50 euros. Classical leukemia diagnostics involve multiple methods, some of which can cost several hundred euros per round. Unfortunately, researchers have not yet developed a feasible solution for widespread use. They have only demonstrated that this method is theoretically feasible.

RECENT AI NEWS

RECENT AI TOOLS