Addressing the Challenges of Adopting AI in Healthcare

AI in healthcare is one of the fastest-growing landscapes for machine learning innovation companies. While today’s CAD technology was developed to improve breast cancer detection and diagnosis, research shows it has not been sophisticated enough to meet the needs of radiologists and streamline their workflow. The technology attempts to balance false positives with false negatives without providing transparency as to how the algorithm makes its determinations. Additionally, CAD technology was evaluated on no more than a few hundred cancers in laboratory settings and did not  include post-market monitoring to assess at scale over time. Ultimately, these factors have led to a decrease in use of CAD technology among radiologists.

Emerging technology could develop improved methods to maximize the efficiency of artificial intelligence in breast cancer detection and diagnosis, and overcome some of the fundamental challenges of today’s technology. For example, AI can learn from reviewing hundreds of thousands of normal mammograms — a scale much larger than any physician could review in his or her lifetime — and AI can be assessed by evaluating more than 2,000 cancers across multiple institutions.

Additionally, we believe once emerging AI technology with a total lifecycle plan design that includes field monitoring — and routine feedback to radiologists — is deployed, it could ensure safety and efficacy at scale once the technology is in clinical practice. Continuous evaluation of the technology could measure model drift and include similar monitoring techniques used for new pharmaceuticals, such as the COVID-19 vaccine. Data from post-market monitoring could also be assessed by an impartial body to ensure the safety and accurate performance of the technology.

We believe once emerging AI technology is deployed, performance metrics would be able to give radiologists significant visibility into the functionality of the application, which they could  take into consideration as they review mammograms. As the technology expands, there could also be an opportunity for site-specific autonomous AI — the development of unique AI configurations that ensure the technology produces the same level of results submitted to the FDA — based on each facility’s data and population. Accrediting facilities for autonomous AI could be a key function of the primary organizations focused on radiology.

With advanced learning, improved feedback and transparency, emerging AI technology has the ability to improve the interface of artificial and human intelligence to build a stronger national healthcare system for all people.

The post Addressing the Challenges of Adopting AI in Healthcare appeared first on AiThority.

artificial intelligence
ai
machine learning
healthcare
fda
research