Researchers at the University of Waterloo and elsewhere has made a big advancement in medical diagnostics using artificial intelligence (AI).
Led by engineering professor Alexander Wong, the team has developed a new AI-based method to enhance the diagnosis of diseases like COVID-19, pneumonia, and melanoma.
This research is published in the journal Sensors and introduces the Trustworthy Deep Learning Framework for Medical Image Analysis (TRUDLMIA).
TRUDLMIA represents a major leap in developing reliable and high-performing healthcare models.
Dr. Wong explains that TRUDLMIA not only surpasses existing models in diagnosing specific diseases but also addresses the crucial aspects of performance and trust.
The system is currently being refined to tackle future pandemics and long-term effects related to COVID-19.
The integration of medical imaging and deep learning in medical AI has the potential to revolutionize diagnosis, prediction, and prognosis.
However, challenges such as data bias, low trust, and interpretability have hindered progress. TRUDLMIA addresses these issues through a three-stage training process for the AI system.
The first stage involves learning from a vast set of labeled general data. The second stage uses both general and domain-specific data (like medical images) in a self-supervised learning process without labels.
The final stage fine-tunes the AI with task-specific labeled data, focusing on mitigating data imbalances and biases and enhancing trustworthiness.
The team highlights that TRUDLMIA was developed with direct input from medical professionals. The system aims to improve diagnostic accuracy, foster trust, and adapt across various medical specialties.
If you care about medicine, please read studies that vitamin D could help lower the risk of autoimmune diseases, and drug for inflammation may stop spread of cancer.
For more information about medicine, please see recent studies about which drug can harm your liver most, and results showing this drug can give your immune system a double boost against cancer.
The research findings can be found in Sensors.
Copyright © 2024 Knowridge Science Report. All rights reserved.
