Most of my patients know that I’m a big fan of technology. From electronic medical records to viewing diagnostic images over the web, I love finding tools that help me take better care of patients. A study in this week’s New England Journal of Medicine is an important cautionary tale that reminds us that new technologies should always be tested rigorously.
The study examined the use of a technology called computer-aided detection to assist radiologists in interpreting screening mammograms. Computer-aided detection involves computer software that analyzes mammogram images and identifies suspicious abnormalities. It was approved by the FDA in 1998 and has gained popularity since then. The study was also covered in yesterday’s Los Angeles Times.
The study found that the technology actually decreased accuracy when compared to a radiologist reading the mammograms without computer assistance. Computer assistance lead to an increase in potential abnormalities being identified that overwhelmingly turned out to be benign. The number of women recalled for additional imaging increased by 32% and the number of biopsies increased by 20%, but the number of actual cancers detected did not increase. So computer assistance only led to unnecessary procedures and didn’t assist in diagnosis.
So for now, we should leave mammogram interpretation to trained humans, and let computers do what they’re best at — connecting you to educational articles from your doctor!
Tangential Miscellany:
Drum roll, please. The following are the winners of the Pedometer Project for March.
The winner with most steps was Sari A. In her honor I’ll be donating platelets at Cedars-Sinai on April 12.
The winner for most improved walker in March was Victoria W. In her honor I will volunteer at the Simms/Mann Clinic on April 18.
The winner for weight loss was Douglas C. In his honor I will volunteer at the Simms/Mann Clinic on May 2.
My congratulations to all the participants! The Project will continue through the end of April.