BOSTON, Sept. 11, 2020 /PRNewswire/ — Today the level of accuracy of an algorithm is its number one selling point. Whilst AI has the potential to revolutionize the disease diagnosis process, its current value proposition remains below the expectations of most radiologists.
IDTechEx gathered the performance measurements of over 35 AI algorithms and found that most are on par with human radiologists for disease detection performance. The goal, however, is for AI algorithms to outperform humans. Ultimately, AI must be more reliable and more accurate than even the most highly trained experts in order to gain credibility as a decision support tool.
IDTechEx have charted the sensitivity and specificity of over 35 algorithms developed for disease types. The top right corner of the graph represents perfect performance and the shaded areas represent the range of human performance. Most points either sit within the range of human performance or remain on par with human levels for detecting negative cases as their specificity is within range of that of humans (yellow box). Only a few algorithms demonstrate super-human sensitivity and specificity. AI has the potential to outperform humans at detecting disease from medical images, but most AI companies have not yet been able to achieve this.
A number of improvements could help image recognition AI to reach its full potential as a decision support tool for radiologists.
1) Increasing diversity in training data sets will widen the software’s applicability
A key technical and business advantage lies in the demonstration of success in dealing with a wide range of patient demographics as it widens the software’s applicability. While training DL algorithms, the training data should encompass numerous types of disease, lesions, and other parameters so that the algorithm can recognize a multitude of demographics, tissue types, abnormalities, etc and perform to the level required by radiologists.
2) Including more negative cases in the training data can raise algorithm specificity
During algorithm training, confirmed disease cases often take priority over negative cases to raise the algorithm’s disease detection performance. This is beneficial for identifying patients at risk of disease but limits the AI’s ability to recognize healthy cases. Low specificity is hence a recurring issue, which can lead to overdiagnosis and costly unnecessary procedures. This problem can be addressed by using more curated negative cases during the training process.
3) Using high-resolution images will maximize algorithm performance
The use of poor-quality data during training negatively impacts the development process and performance levels of DL algorithms. Unclear images reduce the accuracy of insights generated by AI, which can damage its chances for widespread implementation.
Methods that enable doctors and radiologists to capture better images or enhance their resolution can boost the value of AI in medical settings. AI-driven methods for assessing or improving image quality are already commercialized. Companies focused on data quality hold a competitive advantage as dealing only with high-resolution images heightens the reliability of AI-generated insights.
For more information on this report, please visit www.IDTechEx.com/IRAI or for the full portfolio of Healthcare and AI research available from IDTechEx please visit www.IDTechEx.com/research.
IDTechEx guides your strategic business decisions through its Research, Consultancy and Event products, helping you profit from emerging technologies. For more information on IDTechEx Research and Consultancy, contact research@IDTechEx.com or visit www.IDTechEx.com.
Digital Marketing Manager
View original content to download multimedia:https://www.prnewswire.com/news-releases/ai-in-medical-diagnostics-idtechex-discusses-key-strategies-for-faster-clinical-uptake-301128313.html