When Hurricane Helene hit North Carolina in late September, it caused more than $59 billion in damages.
Among those businesses damaged was one of the US’s main manufacturers of IV fluids, used to rehydrate patients and give them medicine. The resultant shortage forced hospitals to conserve and reduce their use of IV fluids, which led to canceled surgeries and treatment delays.
Such disruptions to hospital inventory have long been hard to predict and difficult for hospitals to navigate. At the same time, keeping too much of a given item on hand is wasteful. In 2019, hospitals spent about $25.7 billion on supplies that they didn’t need, the consulting firm Navigant found in a study of over 2,100 hospitals — about $12.1 million for an average hospital.
To reduce waste, while ensuring providers have the medical supplies they need, some leading hospital systems are using automation, predictive analytics, and other forms of artificial intelligence to manage inventory.
Business Insider asked supply chain managers from three systems — the Cleveland Clinic, the Mayo Clinic, and Rush University Medical Center — about how they use machine learning, generative AI, sensors, and robotics to anticipate shortages and help with contracting and ordering.
The following has been edited for length and clarity.
Business Insider: What are some of the most effective and innovative ways that you’re making use of AI?
Joe Dudas, Mayo Clinic’s division chair of supply chain strategy: We’ve deployed autonomous delivery and robotic warehouse fulfilment — robots that pick orders.
We’re advancing our algorithms for auto-replenishment to be even more accurate. We’re also using AI to explore savings opportunities and understand the sustainability of those opportunities over the length of an agreement, based, for example, on demand.
We’re doing advanced analytics in high-spend categories — we’re just getting a lot smarter about what’s happening with a little bit more precision. Even our expense management, we’re looking at profit and loss, and supply expense, to understand what’s happening from a budgetary perspective. Based on the present and what’s happened in the past, we can look forward with some degree of accuracy.
Geoff Gates, Cleveland Clinic’s senior director of supply chain management: In some of our tools, instead of having someone click lots of buttons and type data into 20 or more fields, for example, we have been able to automate that process with AI, which saves employees 20 minutes every time. Those are the tasks that are the biggest benefit from a pure efficiency standpoint — they let people focus on other things.
We also use AI for document recognition and have been using it to manage invoices through our ERP inventory-management system for the last four years. If a medical-supply rep has a bill sheet that needs to get processed — to create a purchase order — the rep submits it, and our tool automatically creates a requisition.
With distribution, our goal is to create a better view of what we have within our health system and the hospital. Our goal with key suppliers is to be able to see which supplies they have in their warehouses and to predict disruptions. For instance, if we can see that a supplier doesn’t have a shipment coming in, the system would alert us that we’ll have a problem in two weeks.
Jeremy Strong, Rush University Medical Center’s vice president of supply chain: For inventory management, we have weighted bin systems in all heavy-volume areas. When a nurse takes something out and puts something back in, we know it.
Once we implemented that, we could start to be proactive. We have a system that includes our distributor’s data about inventory coming into their distribution center. They can see where our utilization patterns are changing. Then AI reviews all that. A back-order dashboard creates alerts when automatic supply-refill levels across the system are low, inventory is low at the distribution center, or shipments from manufacturers are taking longer than anticipated. We can anticipate that we’re going to run out in a week from now or going to have a back-order problem.
We also use it in contract management. When a contract is loaded in, AI will send it to the category manager with a summary and potential clauses to review. It can also automatically send contracts to the cybersecurity team for approval. If it has patient information, it sends it to the risk lawyers. If it has indemnification, it sends it to our regular lawyers.
What are some of the advantages of automation that your system has realized?
Dudas: Our automation gives us agility. We can see things sooner and adjust faster because of our technology but also because of our talent.
Somebody asked me the other day, “Where are you advancing?” I said, “We’re not advancing. We’re keeping up with all of the curveballs we get thrown on a day-to-day basis.”
Gates: At this point, the tools have touched almost everyone in the supply chain. Even a specific process that only impacts one or two people who were doing those tasks allows us to be more efficient and accurate.
Strong: The goal was to move from being reactive and putting out fires to being more predictive, to prevent fires from happening, see things ahead of time, and be more efficient.
We’ve also sped up contract review. We cut the time it takes to review them in half and more than doubled the number of reviews each contract gets.
What advice do you have for other companies interested in implementing AI to streamline inventory?
Dudas: Recognize that you can’t do everything yourself. Even as big as our organization is, it’s not big enough. Scale is your friend in the supply chain.
Gates: Some things weren’t necessarily the biggest opportunities to start with, but they were low-risk processes that gave us the skillset needed to leverage AI.
We’re most focused on finding the right solution for the problem rather than forcing a solution.
Strong: The best tasks or processes to tackle are ones that are repetitive or require pulling and summarizing data from multiple digital sources. Tackle these, and you can gain efficiencies, improve productivity, and be proactive.
Read the full article here
