When diagnoses from a group of physicians are combined, they are more accurate than a single diagnosis — even when the group may not have specialty expertise, according to a study published March 1 in JAMA.
In what is considered the largest study of collective intelligence in medicine to date, researchers used cases from a multinational data set to determine if gathering multiple clinical opinions could improve diagnostic accuracy. The study looks at data from the Human Diagnosis Project, also known as Human Dx, a medical project in which physicians and medical students from 80 countries solve teaching cases.
The researchers found that when independent diagnoses are combined into a weighted list, they can produce more accurate diagnoses every time, even with just two physician opinions. Accuracy improved as more opinions were used. Whereas an individual physician's diagnostic accuracy is 62.5 percent, a group of nine physicians could produce a diagnosis with 85.6 percent accuracy. Researchers found groups of nonspecialists outperformed a single specialist even when the case in question fell under that physician's subspecialty.
Rather than using a traditional group discussion approach, the researchers were able to combine independent assessments with technology. Their findings suggest pooling and scaling diagnoses with datasets could improve diagnostic accuracy, according to the study.
"Collective intelligence facilitated by software is particularly well suited for medicine because it may offer superior performance with little coordination," the authors wrote.
More articles on quality and infection control:
Amazon removes anti-vaccine content after congressman pens letter of concern
FDA warns against robotic surgery for cancer due to safety concerns
Stem cell therapy sickens at least 17 patients in 5 states