Site icon Connected World

Unpacking the Implications of AI in Healthcare

We all know AI (artificial intelligence) offers huge opportunities and huge risks for various industries, many of which we have discussed here on the blog. Today, let’s unpack what this will mean for one specific vertical market: healthcare. I have great hope for AI. But with great technology comes great responsibility. And as much as I do not want to encourage more regulation when it comes to human health, sometimes the risk outweighs the rewards. But let’s continue to talk and unpack this hefty issue a little more.

Healthcare is a critical, vital industry that requires research, hands-on care, and, yes, tedious, and time-consuming notetaking tasks. What if AI could step in and help? What if it could support research and development of new medicines? Indeed, this appears to be a perfect opportunity to lift the burden off the healthcare community. What if it could detect and treat current and emerging diseases faster? What if it could relieve the workload of healthcare providers? But like anything we must raise a red flag for the sake of humankind.

All of these what-ifs are appealing to those in the healthcare profession, but the reality is AI is here and it can help, but of course alongside these benefits are also some potential risks. This is precisely what a new whitepaper from the Senate Health, Education, Labor, and Pensions Committee explores.

The Exploring Congress’ Framework for the Future of AI whitepaper looks at the oversight and legislative role of Congress in the integration of artificial intelligence in health, education, and labor. Looking specifically at healthcare, the whitepaper suggests AI can help research and develop new medicines, diagnose and treat disease, support patients and providers, and other administrative tasks. Research shows physicians spend about 8.7 hours a week on administrative tasks. Certainly, AI can help here.

Still, the challenge is the current regulatory framework, security and trust, and potential liability, just to name a few. Steps should also be taken to ensure that AI is not overriding, or even altering, clinical judgement. Some patients have been unable to receive a provider’s opinion due to algorithms automatically deciding a treatment plan. The bottomline is that each use case is going to have different benefits and challenges.

Case Study: AI and Rare Diseases

Let’s spend a few minutes digging into a very specific example to weigh the pros and cons of such technology in healthcare. As we have seen here, AI can help detect patterns and do research and treatment of rare diseases.

More specifically, AI can assist by analyzing vast amounts of data to identify characteristic patterns and markers of specific rare diseases, reducing diagnosis time and costs. Typical rare disease diagnoses involve medical histories, physical examinations, and genetic testing. AI-powered diagnostic tools can streamline this process by identifying potential diagnoses faster and more accurately. Various machine learning techniques are being developed to standardize and share medical information, improving interoperability in the field of rare disease.

However, there are challenges. In the U.S., a disease is considered rare if it affects fewer than 200,000 people. The challenge is this is a limited patient sampling, which prevents statistically significant parameters for research.

Dr. Harsha Rajasimha, founder and executive chairman of IndoUSrare, also advocates for diversity in clinical trials in order to take full advantage of AI’s reach for rare diseases. This requires a keen eye and a clear understanding that not all data can provide.

The organization suggests genomic studies have primarily used samples from people of European ancestry, which means the AI doesn’t have the opportunity to learn from larger and more genetically diverse populations, particularly in Asia, which comprises 59.5% of the world’s population.

This exclusion hinders the unbiassed advancement of genomic medicine, leading to persistent uncertainty concerning the genetic basis and epidemiology of diseases across various populations and disparities in drug reactions, treatment outcomes, and overall health. Similarly, in immunogenomics research, a lack of diversity hinders the discovery of novel genetic traits associated with immune system phenotypes, both common and distinct across populations.

Certainly, this is only one very specific example in a very specific vertical market. The point here is that each industry is going to face its own challenges as AI becomes more widespread. Each company will need to examine the inherent value adds and the hurdles that will need to be overcome to have AI reach its full potential, working alongside humans. There is still so much more as a society we can do and certainly so much more to gain, risk, and share.

Want to tweet about this article? Use hashtags #IoT #sustainability #AI #5G #cloud #edge #futureofwork #digitaltransformation #green #ecosystem #environmental #circularworld #healthcare

Exit mobile version