THANK YOU FOR SUBSCRIBING
The adoption of AI in health care has been relatively slow, in contrast to the hype surrounding the technology. With some innovative approaches to privacy and the path to regulatory approval, policymakers can help generate useful adoption of new technologies on time.
FREMONT, CA: Advances in machine learning, closely related to data science and statistical prediction, have driven rapid improvements in artificial intelligence (AI) technologies over the past decade. Diagnostics, treatment, administration, and operations are all aspects of the healthcare system that require prediction. Machine learning's abilities and the healthcare industry's needs have led to widespread speculation in the medical community about the possibility that AI will have a big impact on healthcare in the near future.
Check Out This : Gov Business Review
Stay ahead of the industry with exclusive feature stories on the top companies, expert insights and the latest news delivered straight to your inbox. Subscribe today.
There is a tremendous amount of progress and promise being made in the field of clinical AI algorithms, ranging from image-based diagnosis in radiology and dermatology to surgery and from patient monitoring to the interpretation of genomes and the development of new drugs.
Over 17% of the gross domestic product goes to health care, with 11% of workers employed in health care. Compared to other OECD countries, U.S. healthcare spending is higher per capita. Retail and financial services can benefit from AI technologies in healthcare.
Although AI has the potential to transform healthcare, it has not yet been widely adopted.
Understanding why hospitals may hesitate to adopt AI is useful for understanding complementary innovations that might lead to more adoption. The four main barriers to adoption are data access limitations, algorithmic limitations, regulatory barriers, and misaligned incentives.
Algorithmic limitations: Advances in neural networks increased AI possibilities at the cost of interpretability. Some AI algorithms are called black boxes because they are difficult to understand without substantial effort when neural networks are used. The risk is high that if no one looks out for problems with neural network-generated algorithms, they will generate solutions with flaws only discovered after they have been deployed.
Innovations that facilitate AI algorithm interpretation, such as technologies or processes, are fundamental to trustworthy AI. Several large-scale initiatives are promoting the development of trustworthy AI. Interpretable AI could eliminate the black box problem and allow healthcare workers to understand how AI makes decisions.
These innovations will facilitate AI adoption since healthcare professionals will better understand the likelihood that an AI reaches its recommendation based on biased or incomplete data.
Restrictions on data access: AI algorithms also perform better with high-quality data. Data access is another barrier to adoption. Accessing and collecting medical data can be challenging. Data collection interrupts medical professionals' workflows, and the collected data is often incomplete.
Pooling such data across hospitals is also challenging. EHR systems are still incompatible between hospitals and healthcare facilities certified by the government. A patient's medical history is documented locally rather than across his healthcare providers. Artificial intelligence can be difficult to build without large data sets. The technology may, therefore, be slower to catch on with health care providers.
Regulatory barriers: The healthcare industry is more cautious than other industries. The utilization of AI in health care will require complementary innovations in regulation, ultimately allowing patient rights and quality of care to be protected.
Regulations fall into three categories. Healthcare data can be hard to collect and pool because of privacy regulations. A new medical technology's regulatory approval takes time and undergoes extensive scrutiny. Healthcare providers may also hesitate to adopt new technology for fear of tort law implications due to liability concerns.
A complementary regulatory innovation would include changes to all three sections: how AI medical devices and software are approved and where medical providers and AI developers are liable.
More in News