Machines are just as capable as humans, if not more so, on almost every level when it comes to diagnosing patients.
Open source machine learning tools and algorithms are able to identify cancer cases with an accuracy rate that is equal to or greater than that of trained humans. Furthermore, these machine learning tools are also able to give diagnoses faster than their human counterparts, according to a study done by Regenstrief Institute and Indiana University School of Informatics and Computing at Indiana University-Purdue University Indianapolis (IUPUI). The machine learning tools diagnosed the cases based on text data from pathology reports.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
The Regenstrief and IUPUI researchers analyzed 7,000 text pathology reports and used algorithms and open source resources to forecast whether an individual report was positive or negative for cancer. Their central finding was that a fully automated review done through machine learning was correct at least as often as expert human reviewers.
“It’s no longer necessary for humans to spend time reviewing text reports… A human’s time is better spent helping other humans by providing them with better clinical care,” said Shaun Grannis, M.D., interim director of the Regenstrief Center of Biomedical Informatics, in the release that summarized the findings of the study.
Grannis also theorized that physicians, healthcare systems, HIEs and other healthcare organizations can rely on technology, such as machine learning tools, to interpret the growing amount of patient health information. In the research report, first author Suranga N. Kasthurirathne, a doctoral student at IUPUI, stated that the study’s way of deploying machine learning to identify cancer could be applied to other health conditions.
Machine learning can be helpful in other ways as well. In an interview with SearchHealthIT, John Brownstein, chief innovation officer at Boston Children’s Hospital, discussed how machine learning is incorporated into a disease tracking resource called HealthMap that he helped create. Though it doesn’t help diagnose a particular disease, HealthMap uses machine learning to pinpoint the geographical location of patients with a disease such as dengue fever.