Online Diagnostic Services Are Wrong In Two Thirds Of Cases

2023 Author: Abraham Higgins | [email protected]. Last modified: 2023-11-27 23:16

Online diagnostic services are wrong in two thirds of cases

The accuracy of symptom checkers has not improved over the years. Scientists believe that the diagnostic algorithms that are available on the internet can harm people.

Photo: highexistence.com

Millions of people turn to "Doctor Google" and his "colleagues" to understand the reasons for their poor health. In other words, many people try to diagnose themselves using the Internet. Among the "tools" that the global network offers for these purposes, there are so-called "symptom checkers" - algorithms that make a preliminary diagnosis by asking users questions about the symptoms of the disease.

Australian scientists have tested the accuracy of 36 popular free symptom checkers that are available online and as mobile apps. They published the results of the check in The Medical Journal of Australia. It turned out that such diagnostic algorithms are more likely to deceive than tell the truth.

Scientists have chased symptom checkers through descriptions of 48 different diseases. It turned out that the programs give correct answers on average 36% of the time. That is, in two thirds of cases they misinform people. The most accurate symptom checker tested gave correct answers 58% of the time.

Diagnoses and advice from different programs and sites about the same symptoms can vary greatly. Scientists noted that recommendations to see a doctor for signs of serious illness were given correctly in 60% of cases, but were offered much less often in milder conditions.

“The reality is that these websites and applications need to be handled with great care because they don't show the big picture,” said Michella Hill of Edith Cowen University, co-author of the study. She noted that such services are unreliable at best and dangerous at worst.

Interestingly, a similar study from five years ago yielded almost the same results: 23 symptom checkers were tested and found to be accurate 34% of the time. That is, it can be argued that during this time the reliability of "symptom checkers" has not grown at all.