Machine learning and the challenges for its clinical use

Artificial intelligence, particularly in its machine learning variant, is being tested in the medical sector to analyze data patterns and generate projections based on them, without the need to receive direct orders.

Given the complexity of this context, a team of researchers from Harvard and MIT presented some observations for the clinical use of this technology.

Machine Learning for Clinical Use

The relationship between science and technology is close and long-standing. During the last time, the rapprochement between these two areas has been even greater, due to the influence of, among other factors, the COVID-19 pandemic.

The health emergency demanded the acceleration of any investigation that would help to advance with its overcoming. To study the behavior of the virus, its effects and evolution, statistical data were collected in a large volume, submitted in an unprecedented way to AI systems for analysis. Although the first impact of the virus took the world by surprise, the following waves were greeted with a little more information thanks to these systems.

Despite the novel impact of this analysis methodology, the technology behind it is still in its infancy. The research «Deploying clinical machine learning? Consider the following…», developed by eight researchers from Harvard and MIT, points out in their presentation that “While research is important to advance the state of the art, executing it is equally important in bringing these technologies to a position that ultimately impacts patient care and lives up to the broad expectations surrounding AI in medical care ».

The study was carried out based on a survey applied to several participants linked to the development of this specific variant of AI, which in its environment is already called Clinical Machine Learning or CML (Clinical Machine Learning).

The main challenges identified in the research are related to several stages. When developing a clinical diagnosis, the corroboration of diagnoses is considered as a contribution to medical work, reinforcing its weak points. When validating a diagnosis, the use of AI is suggested to avoid the presence of biases, comparing those samples with those from other hospitals. At the level of statistical development, a CML system can generate a large database, highlighting not only critical data, but also positive figures, such as the number of lives saved during some period. At the deployment level, the main challenge of the CML is its clinical validation, since few of the investigations carried out have been able to achieve compliance with the standards required for a technology to be integrated into some stage of the medical care process.

With these observations, the research team contributed its vision as a guide for the development of future research, in order to project its future implementation in a more concrete way.

Source link

Lenny Li

I started to play with tech since middle school. Smart phones, laptops and gadgets are all about my life. Besides, I am also a big fan of Star War. May the force be with you!

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button