Will artificial intelligence (AI) become the panacea for all the difficulties healthcare supply chain faces? This is a question that many supply chain leaders are pondering.
But before health systems and hospitals can harness the power of AI in healthcare supply chain operations, they need to establish an effective data strategy. Data is the foundation — the raw material — that is necessary to produce AI models and algorithms. Without a large quantity of high-fidelity data, AI will not be able to learn and adapt to provide meaningful information.
Quantity of data is not an issue; data is everywhere. In the healthcare supply chain, it comes from several sources; the enterprise resource planning (ERP) system, electronic health record (EHR) system, manufacturers, distributors, group purchasing organizations (GPOs) and other resources.
In the world of AI-enabled insights, it is quality, not quantity, that is essential for success. And data quality remains a major challenge in healthcare.
I attended the AHRMM24 Annual Conference in September, which featured multiple sessions on the topic of AI in healthcare supply chain. In one session presented by Deloitte, 44% of attendees believed an AI solution would “help address or improve their pain points,” 50% said it was “too early to tell,” and only 6% believed AI would not help.
Turning to specific healthcare supply chain pain points, “lack of data quality and interoperability” ranked highest in the session polling.
If a health system or hospital cannot collate all their disparate data in a way that maintains its reliability, then AI will fall short in delivering on its promise. Turning this data into useful information to optimize the delivery of healthcare services is the end goal.
Bad data is limiting strategic use of AI in healthcare supply chain processes
While retail and banking industries are using AI effectively to improve their operational efficiency, healthcare is still struggling to realize the benefits because of data quality and integration issues.
Here are three factors that hinder health systems and hospital supply chains from creating a complete and accurate data source for effective use of AI in healthcare supply chain.
1. Dirty data in the item master
For any insights to be meaningful and actionable, we need clean data. The item master contains data on many of the products used in patient care. It is, therefore, critical to cleanse this data source. Data points such as a product’s description, unit of measure (UOM), conversion factor and item price must all be correctly configured. This allows AI algorithms to surface accurate and actionable insights on vendor and product spend, inventory levels, item usage, etc.
To address discrepancies in their ERP systems, I’ve seen hospitals engage in a one-time deep cleanse project of their item masters. While that can provide an immediate fix, the problem with this approach is that healthcare supply chain data is constantly changing. Manufacturers change product data, GPOs change contract pricing data, the hospital procures new products and ceasing purchasing others, etc. The supply chain team that paid for a one-time data cleanse soon finds themselves again with “dirty data.”
Any AI-enabled insights derived from erroneous supply chain data cannot be trusted. Furthermore, insights will become more and more inaccurate and inapplicable over time as the algorithms will learn from continuously degrading data quality.
2. Lack of data for products outside of the item master
Another major roadblock to establishing an accurate and complete data source for the application of AI in healthcare supply chain is the inability to integrate data on products that fall outside of the item master.
Even if a hospital has cleansed its item master and is continuously maintaining this data, there are many products used in patient care that are not contained within the ERP system; most notably Bill Only products.
If a supply chain team is training its AI algorithms on item master data alone, they are missing out on insights related to other significant areas of spend.
3. Data discrepancies at the point of use
Let’s say a health system or hospital supply chain team has established a single repository of data for all spend categories and has a mechanism for continuously cleansing and updating this data. Are they positioned for successful use of AI in healthcare supply chain optimization work? Not if they haven’t addressed the challenges of point of use (POU) data capture.
Enabling clinicians to capture accurate and complete data on the products used in patient care remains a significant challenge for many supply chain teams. Even if they have a pristine data source on products that have been procured by their health system or hospital, including those managed by clinical departments, if they don’t have an automated means to collect clean data on product usage at the patient bedside, their data integrity is at risk.
Many organizations still rely on either manual or outdated point solutions for POU product data capture that clinicians often find difficult and/or time consuming to use, resulting in inaccurate documentation in patient record.
Pressed for time and focused on their patients, clinicians will do their best to document items in the EHR, especially when it requires manual intervention, such as keying information from the product's packaging into the system.
Some solutions will provide drop-down lists for vendor and product options so the clinician can select this information instead of keying it in. When a barcode fails to scan, clinicians may inadvertently select the wrong vendor or item from a long list of names and descriptions.
Whether a clinician manually keys inaccurate or incomplete data into the patient record or selects the wrong option from a drop-down list, the result is the same. The healthcare organization is left with erroneous or missing data that can impact the accuracy and reliability of insights derived from AI in healthcare supply chain analytics.
How to bridge supply chain data quality gaps for AI success
What most health systems and hospitals are missing in their data strategy for AI in healthcare supply chain applications is an enterprise-wide supply chain management (SCM) solution that:
- Integrates seamlessly with ERP and EHR systems
- Synthesizes end-to-end item level data, from procurement through use
- Encompasses data on products both inside and outside of the item master
- Continuously maintains the accuracy and integrity of this data
- Facilitates the flow of clean and complete data from the ERP to EHR system and vice-versa
- Automates POU product data capture so clinicians can focus on patient care
- Features AI-driven analytics tools that continuously learn from this data and present meaningful and actionable insights to decision makers
An enterprise platform that empowers health systems and hospitals in their clean data strategies will be successful in realizing the promise of AI in healthcare supply chain operations. By integrating data from disparate systems, it effectively addresses the challenge of capturing accurate and complete data.
The path to AI value realization
While AI holds great promise for transforming healthcare supply chain operations, its success heavily depends on establishing a robust and reliable data infrastructure. The healthcare sector faces significant challenges with data quality, integration and the ability to collect accurate and comprehensive information at critical points.
Without addressing these issues, AI's potential to deliver meaningful and actionable insights will be limited, and hospitals may struggle to fully optimize their supply chain processes.
By bridging data gaps with an enterprise-wide SCM solution, health systems can unlock the full potential of insights derived from AI in healthcare supply chain applications. This approach ultimately enhances decision-making and improves operational efficiency in the delivery of patient care.