Although AI has the potential to completely transform healthcare, its increasing dependence on human supervision and expensive upkeep raise doubts about its actual effectiveness. Find out why hospitals are having trouble striking a balance between creativity and pragmatism.
With its promises of cost savings and efficiency, artificial intelligence (AI) was hailed as the future of healthcare. But the truth is turning out to be more nuanced. AI tools are far from self-sufficient; in order to operate efficiently, they frequently need expensive resources, frequent upgrades, and a great deal of human supervision.
Consider the University of Pennsylvania Health System’s algorithm, which is intended to forecast patient mortality and initiate critical care conversations.
The algorithm’s accuracy decreased throughout the pandemic, resulting in lost chances for crucial discussions. Oncologist and researcher Ravi Parikh discovered that a lack of oversight caused many AI systems to malfunction identically.
AI is currently widely used in hospitals for everything from sickness diagnosis to patient visit summarization. But as Nigam Shah, chief data scientist at Stanford, points out, maintaining them can put a strain on finances.
According to Shah, auditing the dependability of a single algorithm can take months and hundreds of hours.
Even worse, hospitals don’t have standardized ways to maintain or assess these instruments. Yale researcher Dana Edelson found significant differences in AI system performance.
In the meanwhile, the situation is made more difficult by inexplicable errors, such as algorithms that yield inconsistent outcomes.
The answer ? Experts advise spending money on human supervision to keep an eye on AI systems or even use AI to control itself, both of which are costly options.
The true cost of AI in healthcare is not only monetary; it also involves maintaining patient safety and confidence in a rapidly changing digital landscape.
