Can Supply Chain Big Data Solutions Be Leveraged in Healthcare?
on May 21, 2013
The tidal wave of healthcare data is upon us and will only continue to grow rapidly. What can be learned from how Big Data is done in supply chains to get real value?
Full Article Below -
In “Connected Healthcare's Mountains of Data,” we looked at a definition of the five major elements of connected healthcare and the types of data they generate. Here we look at three broad categories of big data value realization: 1) Alerting and real-time decisions, 2) Performance optimization, 3) Serving the customer/patient. Specifically, what lessons might healthcare take from the application of these in supply chain?
Monitoring and Alerting / Managing by Exception / Real-time Decisions and Plan Adjustments
Figure 1 - Applications of Big Data
Companies have developed analytics that take in huge amounts of real-time or near-real-time data and generate alerts and/or make quick decisions based on a set of algorithms. A well-known example in finance is algorithmic electronic trading of securities. Relevant examples exist in supply chain where companies have begun to instrument their supply chains, tagging goods (with RFID or barcode) and tracking their movement across the chain. RFID combined with GPS provides real-time location of shipments. Temperature sensors provide monitoring for temperature-sensitive products. Once the supply chain is instrumented, companies are doing things like:
Early warning of delivery issues—Enables on-the-fly changes to plans and finding alternates for delivery, production, distribution, and allocation.
Geo-fencing—Alerts that a vehicle is traveling outside the expected route, which may indicate a theft in progress or driver error.
Real-time Routing—The final destination for goods in transit (to deconsolidation) can be decided based on current/changing demand. Another example: by tracking temperature, the remaining life for produce in a DC can be used to determine which cases ship first and to where.
Store shelf replenishment—Using a combination of RFID and/or video, shelves are monitored to instruct store associates to restock them the moment they fall below certain levels.
Early warning of supply issues—Predicting supplier or supply chain failures.
Real-time monitoring already happens in intensive care healthcare settings, where patients’ vital signs are constantly monitored. Monitoring is being extended to less intensive settings, like remote at-home monitoring of the elderly or patients with chronic conditions. An instrumented hospital, with real-time locating and sensing/monitoring, allows more automated and efficient allocation and ‘routing’ of scarce resources (doctors, nurses, equipment), based on the ever-changing needs and circumstances.
Using big data to monitor supply risk provides an example that might be relevant to monitoring patient risk. Firms have been able to better predict in advance when suppliers or their supply chain are at risk by monitoring and correlating a wider array of risk-correlated data, such as the supplier’s performance, quality issues, lawsuits against suppliers, late payments to the supplier’s suppliers, reduced shipment volumes, and so forth (in addition to monitoring the supplier’s financial statements). By taking in a wide range of structured and unstructured data, and creating algorithms that look for and learn about combinations, patterns, and predictive correlations, firms can get much earlier warning of potential supply issues, giving them the time to address them. Perhaps there are similar opportunities to look beyond the traditional patient risk indicators, to diagnose problems earlier within patient populations, identify at-risk individuals, and intervene sooner and more effectively.
Performance Optimization and Improvement
The mountains of data generated by instrumented supply chains, factories, warehouses, and stores are also being analyzed to find performance improvement opportunities. Performance can be continually monitored and outliers identified and investigated. Outliers on the plus side may indicate that a particular worker or facility has some best practice that other sites or individuals can emulate. When sub-par performance is identified, then root cause analysis can be done to find and fix the issue.
Increasing the granularity and richness of data can also improve optimization algorithms, to make smarter decisions about how much inventory to buy and when, where it should be held, more efficient routing of trucks, more optimal placement of facilities, smarter utilization of labor, and so forth. This data can also help track costs more accurately and granularly, to aide in understanding the true costs to serve.
In healthcare, big data offers promise in measuring outcomes. This can be used for pay-for-performance, or more sophisticated accountable care payment schemes, to improve outcomes while lowering costs.
Understanding/Serving the Customer (Patient)
Big data has also become critical in understanding and serving customers. Certainly in forecasting, as well as price optimization, the ability to take in more granular data, as well as richer variety of relevant data can improve the performance of those algorithms. Many products have casual factors that impact demand as well, such as weather or the timing, location, and outcome of sports events or myriad other factors. Companies are also starting to experiment with analyzing social media, not just for forecasting (like ferreting out which colors and styles are popular), but also to get early warning of problems with products (people complaining online). The goal of improving the customer experience can be aided by monitoring a variety of data, including wait times in queues, customer behaviors and movements (in-store or online), returns, online comments, as well as explicit surveys. Analyzing all that data together in a cohesive way can be daunting, but offers great potential.
Some Big Data Lessons Learned in Supply Chain
These lessons can be useful in healthcare as well:
Data quality is critical—The best companies put a huge amount of continuous effort into the mundane task of monitoring and cleansing their data (de-duplicating, identifying missing data/empty fields, fixing formatting or incorrect units, normalizing, classifying, cleansing). Big garbage data can be worse than no data.
Think big, start small—It is easy to get carried away with big data grand ideas. Yes, architectures must be designed to support ambitious goals in the long run, but projects should be bite-sized (to the extent possible) for early, achievable wins.
Intuitive visualization and ease of use—These are critical to success. The crux of big data is for the computer to do all the hard work of absorbing massive complexity and make it really simple for the end user.
Leverage machine learning—Programs that continually improve their own algorithms have proven to be very effective for many problems.
So, back to our original question: Can the systems used to manage supply chain big data be repurposed for healthcare big data? These tools and algorithms are quite specific to particular objectives and tasks. It seems highly unlikely they can be directly transplanted for use in healthcare. However, the skills and knowledge of the people who created them, working side-by-side with healthcare domain experts could prove to be a powerful combination. Having just skimmed the surface on this topic, there will be much more to say about big data in healthcare in future articles.