
The cancer-sniffing deep learning artificial intelligence will work side-by-side with human doctors.
Although machines doctors are still a thing of science-fiction, a team of medical researchers from Germany proved that artificial intelligence could outperform human diagnosticians. In a recently published paper, University of Heidelberg scientists revealed that deep-learning AI could predict melanomas, nevi, and benign moles with far better accuracy compared to their human peers.
Artificial Intelligence Indicates Tumor’s Malignancy Based on Picture Database
Professor Holger Haenssle of Heidelberg’s University Department of Dermatology, reveals in a recently published paper that deep learning convolutional neural network could greatly increase the chance of accurate diagnosis.
As explained by Professor Haenssle, the AI resembles a tightly-knit net of artificial neurons that work together based on what they can ‘see.’ Furthermore, the deep learning convolutional neural network or CNN for short learns after each encounter and becomes even better at diagnosing skin cancer.
In a series of field tests, the Heidelberg team pitted the CNN against dermatologists from all over Europe. The results were astounding, to say the least.
As Professor Haenssle wrote in the study, which was recently published in the Annals of Oncology, the dermatologists who participated in this study were able to accurately recognize melanomas in 86.6 percent of cases and nevi in 71.1 percent of cases.
However, even backed by years of clinical experience, the human doctors were unable to outperform the artificial intelligence which managed to diagnose nevi and melanomas in 95 percent of cases.
Conclusion
Professor Haenssle noted that numbers slightly improved on the human side when doctors were granted access to patient records. As a result, in stage two of the trial, human doctors accurately diagnosed melanomas 88.9 percent of the time and benign nevi 75.7 percent of the time.
Unfortunately, the artificial intelligence still trumped them even with the patients’ records to back them up.
The Heidelberg team stressed out that CNN should not be viewed as a replacement for the human doctor but rather as an aid.
Image source: Pixabay