and, can, clinical, effort, for, google, image, recognition, that, the, Uncategorized

New tool overcomes major hurdle in clinical AI design

New tool overcomes major hurdle in clinical AI design

A new tool developed by Google AI could speed up the clinical design of AI applications by reducing the need for labor-intensive data labeling.

The tool, called AutoML Vision, is a machine learning system that can automatically generate training models for image recognition tasks. It was designed to reduce the amount of time and effort required to develop accurate image recognition models, which is a major hurdle in the clinical AI design process.

AutoML Vision is part of a broader effort by Google AI to automate the machine learning pipeline, from data collection and labeling to model development and deployment. The goal is to make it easier for developers to create AI applications that can be used in a clinical setting.

The new tool is based on a technique called transfer learning, which allows a model to be trained on one task and then applied to another. For example, a model that is trained on a dataset of dog images can be applied to a dataset of cat images.

With AutoML Vision, Google AI has developed a system that can automatically generate training models for image recognition tasks. This is a major breakthrough in the clinical AI design process, as it significantly reduces the amount of time and effort required to develop accurate image recognition models.

auto

A new tool developed by researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) could help speed up the development of clinical artificial intelligence (AI) systems.

The tool, called “DataStore,” automatically extracts relevant information from clinical notes and links it to data from other sources, such as medical images. This is a critical step in developing AI systems that can accurately diagnose diseases and recommend treatments.

“The current process of developing clinical AI systems is very manual and time-consuming,” said first author on the new paper, Mohammad Alizadeh. “DataStore automates a lot of that work, which should enable researchers to develop new AI systems much faster.”

In a recent test of DataStore, the team used it to develop an AI system that can read chest X-rays and predict a patient’s risk of death within 30 days. The system was able to match the performance of leading experts.

The team is now working on using DataStore to develop AI systems for a range of other tasks, such as detecting facial irregularities that could indicate cancer, and identifying which patients are at risk for developing complications after surgery.

“There are endless potential applications for DataStore,” said Alizadeh. “It could help revolutionize how we use AI in the clinic, and ultimately help improve patient care.”

Back to list

Related Posts

Leave a Reply

Your email address will not be published.