In a hospital room, a nurse scans an IV bag. A computer instantly logs the moment, creating a timestamp in a patient’s electronic record. In the data trail, it now appears that treatment has begun.
But the drug hasn’t reached the patient yet.
In a humanitarian group’s warehouse, another system is logging data. A shipment of supplies is recorded as received. But the system doesn’t say which donor funded those supplies, or which population they were meant to serve. To the system, inventory is inventory. In the field, those distinctions matter.
In both cases, the data is technically correct. And in both cases, it may be telling the wrong story.
That’s the problem Yuehwern Yih is trying to solve.
“We are collecting more data than ever before,” Yih says. “But does that data truly represent what happened in the physical system?”
That question anchored the fourth annual Douglas C. Montgomery Distinguished Lecture at Arizona State University, where Yih, a National Academy of Engineering member and Tompkins Professor of Industrial Engineering at Purdue University, examined how digital systems can drift out of sync with reality and what that means for health care and humanitarian aid.
When the system misses what matters
Held in April at the Paul C. Helmick Center on ASU’s Tempe campus, the lecture brought together students, faculty members and researchers to examine a growing disconnect: As data systems and automation advance, they don’t always reflect the complex, variable conditions of the real world.
In one example from her talk, “Bridging the Cyber-Physical Gaps in Healthcare and Humanitarian Assistance,” Yih described how hospitals track drug delivery through electronic health records. The system logs when medication is scanned, but not when it enters a patient’s bloodstream, or when treatment is paused, delayed or interrupted.
That discrepancy can ripple outward. A mistimed blood draw can mislead clinicians into adjusting a dosage incorrectly, all while the data appears reliable.
Yih pointed to similar challenges in humanitarian aid, especially in tracking resources.
In many operations, identical supplies are funded by different donors, each with strict rules about how they can be used. A system may show that bottled water is in a warehouse, but it may not indicate which donor funded it.
Even though the items are the same, they can’t be treated interchangeably. Some may be designated for specific groups, such as pregnant women or low-income households, while others are restricted to certain programs or regions. Without that detail, decision-makers can’t reliably determine what inventory remains available or how it can be used.
When data breaks down
This is not a failure of technology, Yih argues. It’s a failure of design.
“What we must always investigate is how the data was collected, including when, why and by whom,” she says.
For Yih, the solution isn’t more data or better algorithms. It’s rethinking the systems that produce and interpret that data. Industrial engineers, she says, are trained to design entire systems by examining how workflows, measurements and human decisions interact.
That includes questioning whether data collected for one purpose, such as billing in health care, can be reliably used for another, like real-time decision-making.
“If we don’t challenge our assumptions about data,” she says, “the conclusions can be very misleading.”
For industrial engineers, that means designing systems that account for human behavior and ensuring the data reflects what is happening on the ground.
“Where we put the sensor, what we decide to collect, those are choices that must be interrogated,” Yih says.
As artificial intelligence, or AI, accelerates the ability to analyze massive datasets, those decisions matter more than ever. Without scrutiny, flawed assumptions can scale as quickly as insights.
A growing forum for big questions
Now in its fourth year, the Douglas C. Montgomery Distinguished Lecture Series has become a cornerstone event for the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering at ASU. The series brings leading experts to the school to tackle the field’s most pressing challenges.
Past speakers have explored topics ranging from AI-driven optimization to global infrastructure problems. In 2025, Georgia Tech’s Pascal Van Hentenryck examined how AI can be applied responsibly for social good. Yih’s lecture builds on that momentum, adding a warning: Even the most sophisticated tools can fail if the underlying data and assumptions are flawed.
Yih’s message resonated with attendees navigating an era increasingly shaped by AI. Feng Ju, a Fulton Schools associate professor of industrial engineering and program chair, says the series introduces students and researchers to new ways of thinking.
“It was exciting to see students and faculty actively engaging with these ideas,” Ju says. “Lectures like these are important because they encourage people to look beyond theory and assumptions and consider how their work can help address real-world challenges across industries.”
Why industrial engineers matter
Following the lecture, attendees gathered at Foch’s Café for a reception, where they continued conversations and connected with ASU Regents Professor Douglas C. Montgomery, a seminal figure in industrial and systems engineering.
Montgomery emphasized that the challenges Yih described highlight why industrial engineers are in high demand. As industries grow more complex and data-driven, organizations need people who can understand entire systems and translate that understanding into better decisions.
Even as AI reshapes many fields, Montgomery says industrial engineers are well positioned to adapt because they focus on the intersection of systems, data and human decision-making.
“Industrial engineers will remain essential because they understand systems,” Montgomery says. “As Dr. Yih’s lecture shows, that kind of thinking is more important than ever.”
