Obstacles that Hinder the Full Implementation of AI In Healthcare
Wed, April 21, 2021

Obstacles that Hinder the Full Implementation of AI In Healthcare

The healthcare care industry may flourish over the years with the help of Artificial Intelligence by improving R&D and by creating machines that could save thousand of lives. / Photo credits by Panchenko Vladimir via Shutterstock

 

 

Artificial intelligence can transform the health industry, as it promises swifter and more accurate clinical decision-making, wrote Roger Kuan of general management magazine Harvard Business Review. AI can also amplify R&D (research and development) capabilities. But it is a double-edged sword. 

According to Alicia Phaneuf of American financial and business news platform Business Insider, machine learning, a subset of AI that aids in identifying different patterns, can provide data-driven clinical decision support (CDS) to physicians and hospital staff. It also uses algorithms and data to generate automated insights for health professionals. 

AI will continue to grow in the health industry as AI spending in healthcare is forecasted to grow at an annualized 48% between 2017 and 2023. Still, let us examine the pros and cons of AI and find out why implementing it will be long and difficult. 

The Benefits of AI In Healthcare 
1. Foster Preventative Medicine and New Drug Discovery 

For instance, researchers at the University of North Carolina Lineberger Comprehensive Cancer Center used IBM Watson’s Genomic product to determine specific treatments for more than 1,000 patients. Through big data analysis, the product was able to provide treatment options for “people with tumors who were showing genetic abnormalities.” 

Google’s Cloud Healthcare application programming interface (API) is bundled with CDS offerings and other AI solutions to help health professionals “make more informed clinical decisions.” Using machine learning, AI gathers data from users’ electronic health records. 

Google collaborated with the University of California, Stanford University, and the University of Chicago to create an AI system that has the ability to predict hospital visits. The system’s purpose is to prevent readmissions and reduce the amount of time patients are confined in the hospital. 

 

 

2. Task Automation and Big Data Analysis 

Administrative tasks comprise 30% of healthcare costs. With AI, it can automate some tasks such as pre-authorizing insurance, maintaining records, and prompting a follow-up on unpaid bills. AI can reduce the workload of health professionals, which helps them save money.  

On the other hand, AI can analyze big data sets by obtaining patient insights and analyzing them. With the data, healthcare ecosystems can identify “key areas of patient care” that need significant improvement. 

3. Better Service for Patients

Wearable technologies such as FitBits and smartwatches utilize AI to analyze data and notify users and health professionals “on potential health risks and issues.” Since a patient can assess their health through technology, it eases the load of hospital staff and prevents unnecessary remissions and hospital visits. 

Why Integrating AI In Healthcare Is A Challenge

 

Bias in machine learning, tracking are some of the challenges that may come in the way in the health industry of AI. / Photo credits by Panchenko Vladimir via Shutterstock


1. Unconscious Bias

Healthcare technologies are based on data humans provide. Hence, there is a possibility of “data sets containing unconscious bias.” There is also a risk for coder bias and bias in machine learning “to affect AI findings.” Healthcare is a sensitive industry. Therefore, it is essential to formulate new ethics rules to prevent and address AI bias. 

In fact, algorithmic bias is prevalent in many industries, Will Knight of MIT Technology Review wrote on Business Insider, “It’s important that we be transparent about the training data that we are using, and are looking for hidden biases in it, otherwise we are building biased systems," John Ginenndrea argued. 

2. Establishing Regulatory Frameworks

Even agencies are going to struggle with the implementation of AI in healthcare. For example, The US Food and Drug Administration (FDA) has been working to update its regulatory frameworks to be abreast with the technological advancements in the health industry.  

The agency launched its Digital Health Innovation Action Plan in 2017 to highlight its role in spearheading safe and effective digital health technologies. This September, the FDA rolled out its Policy for Device Software Functions and Mobile Medical Applications to describe how it will regulate software that aids in clinical decision support (CDS), as well as software that employs machine learning-based algorithms. 

For software developers, they must know how to design and launch their product in accordance with the FDA’s frameworks. Inevitably, software products will continue to be updated and changed over time. Hence, the product’s FDA approval may place it at risk for each update. Developers might want to resort to a version-based approach to the agency’s approval process. 

3. AI’s Black-box Nature

One such challenge is tracking. Kuan posed, “Can the root cause of the negative outcome be identified within the technology so that it can be prevented in the future?” This involves reclassifying the training data and redesigning the ML algorithms, which can be a complex process and may result in the software product being removed from the market. 

Alarmingly, someone could either intentionally or unintentionally feed incorrect data into the AI system, causing unintended outcomes such as incorrect treatment recommendations and misdiagnosis. However, detection algorithms can identify and address incorrect inputs to minimize said risks. 

There are FDA-approved software tools, but adopting them has been slow. To illustrate, patients are more likely to develop trust issues with AI technologies. They can acknowledge human errors in healthcare. However, they may have little to no tolerance for machine errors. 

Developers, regulatory agencies, and health professionals must work together for AI to be fully implemented in the healthcare system. AI should aid in revolutionizing efficiency and improving decision-making processes. AI integration will be a slow process, but it will definitely raise the bar for the health industry.