AI able to predict chemotherapy dosages developed
February 05, 2019 / Caitlin Stanway-Williams
A group of researchers from Stanford University Hospital have successfully used AI technology to accurately adjust chemotherapy dosages by identifying those who required a lower dose even before treatment had begun, according to a study published in Nature Scientific Reports. Finding the right dose of medication for chemotherapy treatment uses a process of trial and error which can cause unnecessary suffering for patients. Adverse effects from taking the wrong dose of medication results in an estimated 280,000 hospitalizations in the US every year, according to the report. To discover the correct dosage, the team used the "Random Forest Classifier" method, an algorithm which combines the results of decision trees based on slightly different subsets of data. To train the algorithm, the researchers fed it digital information on patients who needed to have their dosages changed from various sources, including analysis results, notes and prescriptions.