As in many fields, artificial intelligence (AI) is all the rage in the health care industry. Hundreds of startups are pitching AI-themed products and services to providers like Henry Ford Health and Corewell Health, two of Michigan鈥檚 largest hospital systems. And leaders at those providers are sorting through the offerings and piloting what seem to be the most useful applications.
Seeing Patterns Humans Cannot See
AI鈥檚 ability to mine huge data sets can help radiologists and pathologists draw insights from X-rays, MRI readouts, and other medical imagery technology.
鈥淢achines are really good at seeing patterns in data that we can鈥檛 see,鈥� said Jason Joseph, Chief Digital and Information Officer for Corewell Health.
That can mean spotting a problem in an image that even a trained human eye might miss. It can also mean spotting a problem when the human reader was looking for something else, like highlighting a bone or heart problem when checking an X-ray for cancer.
As David Allard, Chief Medical Information Officer for Henry Ford Health said, AI is 鈥済reat at picking out details especially when it鈥檚 something
maybe we weren鈥檛 looking for.鈥�
AI tools can also be useful in summarizing a doctor/patient conversation. AI-aided programs can pull out useful information from a recording 鈥� patient histories, follow-up questions, next appointments or prescriptions 鈥� and sort it into patient summaries that save doctors countless hours of paperwork.
Predictive Analytics Spots Problems Early
And using “predictive analytics” 鈥搈ining vast data fields for what may happen next 鈥� can help spot a problem before it gets worse, like seeing the first telltale signs of respiratory distress or blood pressure falling.
鈥淥ften the signs that the patient is deteriorating are very, very subtle,鈥� said Joseph. 鈥淪ometimes by the time you see that as a human it鈥檚 late in the process. And so predictive analytics allows us to get ahead of that.鈥�
Joseph and Allard agree that among the questions providers like Henry Ford and Corewell are asking: Will any given AI advance improve patient outcomes? Is the cost prohibitive? Does an AI-themed product fit into the workflow, or will it be too burdensome to use daily? Only time and practice will help sort out the truly useful advances.
‘Nobody’s Worried 有料盒子 Their Job’
An important point: In no way is AI intended to eliminate the 鈥渉uman in the loop鈥� in diagnosing and treating patients.
鈥淣obody鈥檚 worried about their job. They鈥檙e worried about getting their job done,鈥� said Allard. 鈥淚 get a lot more questions about when can (they) get more AI鈥� than otherwise.
Joseph echoed that. 鈥淪o far, I don鈥檛 think we鈥檝e seen a lot of AI taking peoples鈥� jobs. What I think we will see is AI will take peoples鈥� jobs who don鈥檛 use AI.鈥�
AI Technology Likely to Become Routine
Certainly, it will take time to sort through all the possible uses, drawbacks, and tradeoffs for the industry. But nobody doubts that more AI-assisted products and services are coming.
鈥淚 think we鈥檙e in this messy period of figuring it out as an industry,鈥� said Joseph. 鈥淲e鈥檙e trying a lot of different things, and we鈥檒l keep the things that stick, and we鈥檒l probably not keep the things that don鈥檛 stick.鈥�
And at some point, Allard said, today鈥檚 intense conversation about AI will fade into the background as the technology becomes more routine.
鈥淎I is going to be built into all sorts of tools and processes that we use every day,鈥� he said. 鈥淚t鈥檚 going to be part of health care quality and assurance. It鈥檚 going to seep into almost everything over time.鈥�
John Gallagher is a freelance writer and author in Detroit, and formerly of the Detroit Free Press.