In 2020, global healthcare systems have accelerated operations to tackle the COVID-19 pandemic and leveraged technologies including AI to meet the medical needs of the tens of thousands affected. AI in healthcare has been pushed into action and demand for its deployment across a host of medical disciplines has grown. Can AI bring healthcare the coveted triple win proposition, can it be good for providers, patients, and payers? Many believe it can. A crucial factor in successful training and deployment of models is the availability of structured datasets, enriched by specialists with a deep understanding of the medical domain.
Read the White Paper to:
- Understand the medical domains where Computer Vision is being leveraged
- Delve deeper into explainable AI and its increased importance in medicine
- See how AI has been put to work in the COVID-19 research efforts
- Follow the process of building a workforce of domain specialists in the medical imaging field
First 300 words
Although a year ago may now seem like the distant past in the tech world, the promise of artificial intelligence and machine learning has not abated. In healthcare, while advances continue to be largely in the research and development, medical practitioners are increasingly looking to AI to play a greater role in healthcare than they did even a year ago. One of the questions uppermost on the minds of clinician attendees at radiologic conferences in 2019 was, how long before AI takes my job? While some of those worries may have faded rapidly, in part due to the coronavirus pandemic which has demonstrated how unprepared most world health systems are for a large-scale emergency, the demand for deployment of AI across a host of medical disciplines has only grown. Can AI bring healthcare the coveted triple win proposition, can it be good for providers, patients, and payers? Many believe it can.
Unlike other market segments—transportation, entertainment, manufacturing, and commerce, for example—accelerating AI deployments in healthcare cannot be accomplished simply by throwing more resources at the problem. AI models require vast amounts of data to learn to navigate in the real world. Whether the data is in the form of text or images, it must often first be annotated into a language the algorithm can understand. This work is often performed by humans—domain experts—who are trained to label relevant data elements in literally millions of images. Accuracy is of utmost importance, for the old computer adage “garbage in, garbage out” applies even more strongly to AI than it does to conventional computer programming. Correcting a faulty AI model is not simply a matter of rewriting a few lines of code. The entire learning process may have to be restarted from scratch.
Where do domain experts come from? The words “expert” and…