An improved denoising autoencoder (IDAE) was nearly 100% accurate in identifying HCV, indicating that this neural network model could be useful in diagnosing and managing the disease, according to results published in Scientific Reports .
“In order to improve the effectiveness and accuracy of HCV detection, this study proposes an IDAE and applies it to HCV detection,” researchers wrote. “Conventional denoising autoencoder introduces random noise at the input layer of the encoder. However, due to the presence of these features, encoders that directly add random noise may mask certain intrinsic properties of the data, making it challenging to learn deeper features.”
In the current study, investigators aimed to address data information loss that occurs in traditional denoising autoencoding by incorporating the concept of residual neural networks into an enhanced denoising autoencoder. They applied the IDAE to an opensource HCV dataset.
Findings indicated significant results from the IDAE in the extraction of data.
“While existing baseline machine learning methods have less than 90% accuracy and integrated algorithms and traditional autoencoders have only 95% correctness, the improved IDAE achieves 99% accuracy in the downstream HCV classification task,” researchers wrote.
This represents a 9% improvement over a single algorithm and a nearly 4% improvement over integrated algorithms and other autoencoders, according to the study results. The findings indicate that IDAE “can effectively capture key disease features and improve the accuracy of disease prediction” in HCV data.
As a result, researchers note that IDAE has the potential to be widely used in the detection and management of HCV and similar diseases. This is particularly applicable for the development of early warning systems, predicting HCV progression, and personalized treatment strategies.
The study team did note several limitations, as well as directions for future investigations.
“It is still not possible to completely circumvent the possibility that negative test results may be underdiagnosed, which is an important limitation of the current study,” they wrote. “Looking ahead, given the advantages demonstrated by the improved noise reducing autoencoder on the HCV dataset and its potential in medical data research, we believe that its extension to other feature learning tasks within the healthcare domain has positive application prospects.”