Data and Mathematical Modeling to Enhance Humanitarian Response

By Marco Serrato, vice president of Learning Enterprise at Arizona State University

I still remember the sound before I understood what was happening.

It was 7:19 a.m. in Mexico City in 1985. I was a child, playing on the floor with toys I had received just days earlier for my birthday, when the ground began to shake. At first, it was confusing. Then came the urgency in my parents’ voices as they moved to protect my siblings and me. In the hours that followed, the city I knew was transformed. Buildings collapsed. Streets were blocked. Sirens filled the air. A deep sense of uncertainty settled over families and neighborhoods.

That moment shaped how I understand disasters — not as statistics, but as lived human experiences.

Decades later, as a professor working at the intersection of data, mathematical modeling and global strategy at Thunderbird School of Global Management at Arizona State University, I return to that morning often.

It informs a question that now guides an important part of my work as a professor and researcher: How can we use data modeling and analytical tools to reduce human suffering when disaster strikes?

Disasters are increasing - so must our capacity to respond 

Natural disasters are rising in both frequency and severity. Analyses project that global events could reach 560 per year by 2030. In just the past five years, more than 2,300 disasters have resulted in over $1 trillion in damages, affected more than 700 million people and claimed approximately 280,000 lives, according to the United Nations Office for Disaster Risk Reduction.

These are not abstract figures. Behind every event are families whose daily routines are replaced overnight by disruption, fear and uncertainty.

In the first hours and days after a disaster, decisions matter. Humanitarian logistics — the systems that move water, food, medicine and essential goods into affected communities — often determine whether relief is timely or delayed.

Yet humanitarian response is extraordinarily complex. Infrastructure may be damaged. Roads may be blocked. Demand fluctuates unpredictably. Supplies often arrive in waves, including large volumes of in-kind donations. Multiple organizations, including public agencies, non-governmental organizations and volunteers, must coordinate quickly with limited information.

Nowhere is this complexity more visible than in the “last mile”, the Points of Distribution (PODs), where affected individuals receive essential goods.

• Where should these sites be located?
• How much capacity should each one have?
• When should capacity be expanded, reduced or closed entirely to redirect scarce resources elsewhere?

These are operational questions. But they are also moral ones.

From operational efficiency to human impact

In a recent research paper co-authored with colleagues from Universidad Iberoamericana, Rensselaer Polytechnic Institute and the University of California, Davis, we examined a central challenge in disaster response: how to dynamically adjust capacity at Points of Distribution as conditions evolve.

Traditional logistics models prioritize financial cost minimization. In commercial systems, this makes sense. In humanitarian systems, it is insufficient.

When access to water, food or medicine is delayed, the cost is not merely operational. It is human.

Our work incorporates what are known as deprivation costs, a quantifiable representation of human suffering that increases as access to essential goods is delayed. Unlike traditional penalty models, deprivation costs reflect the reality that suffering intensifies over time.

The longer a person waits for clean water, the greater the physical and psychological impact. The longer a community lacks medical supplies, the greater the health risks.

By embedding deprivation costs into the decision framework, we shift the objective. Instead of minimizing logistics expenses alone, we minimize total impact: operational cost plus human cost.

This reframes humanitarian logistics as a problem of resilience and responsibility.

Modeling decisions under uncertainty

Disaster environments are inherently uncertain. The number of people arriving at a distribution point changes from period to period. Some individuals may not receive aid in the first round and return later. Supply conditions evolve.

To address this, we developed a Markovian Decision Model for Capacity Adjustment, a structured framework that treats each phase of disaster response as a strategic decision point.

At each period, decision-makers must determine whether to:

• Maintain current capacity
• Increase or decrease capacity
• Close a distribution point and reallocate resources elsewhere

The model captures the stochastic — or random — nature of demand while incorporating both operational and deprivation costs. It evaluates how current decisions affect future system states, enabling leaders to identify policies that minimize total expected cost across the planning horizon.

Importantly, the model demonstrates the existence of a threshold policy for closure decisions. Under certain conditions, there is a clear, monotonic rule: Below a specific level of demand or shortage, it is optimal to close a POD and reallocate resources. Above that threshold, maintaining operations reduces total harm.

For practitioners, this matters. It provides clarity in environments where ambiguity is the norm.

Why this matters for global leaders

Humanitarian logistics is not only a technical challenge. It is a leadership challenge.

Leaders operating in disaster contexts must balance speed with accuracy, compassion with discipline and limited resources with immense needs. Decisions cannot be delayed, yet they must be defensible.

Data and modeling do not replace human judgment. They strengthen it.

By explicitly quantifying human suffering within decision frameworks, we provide leaders with tools that align operational efficiency with humanitarian objectives. The result is not cold optimization. It is structured compassion.

At Thunderbird, we prepare leaders to navigate complexity with clarity. Disaster response is one of the most extreme environments in which that clarity is required. The same principles apply across sectors:

• Integrate data with strategy.
• Recognize hidden costs, including human impact.
• Design systems that adapt dynamically under uncertainty.
• Make decisions that are analytically rigorous and ethically grounded.

Data as a moral instrument

When I reflect on 1985, I remember more than the shaking ground. I remember confusion. I remember delays. I remember the sense that systems were overwhelmed.

Today, we have tools that did not exist then. We have richer data. We have computational power. We have analytical frameworks capable of modeling uncertainty in real time.

The responsibility is to use them wisely.

Mathematical modeling in humanitarian response is not about replacing empathy with equations. It is about ensuring that scarce resources are allocated in ways that reduce suffering as effectively as possible.

Disasters will continue to test communities across the globe. The question is not whether they will occur. It is how prepared we are to respond.

By integrating data, mathematical modeling and strategic decision-making, we can design humanitarian systems that are more adaptive, more efficient and more humane.

For me, this work is personal. It began with the memory of a childhood morning in Mexico City. It continues with the conviction that disciplined, ethical use of data can help communities recover faster and with greater dignity.

That conviction shapes how I teach, pairing analytical rigor with human awareness, and reminding our students that every model represents real people whose lives depend on the decisions they make.

Whether in disaster zones or boardrooms, the future belongs to leaders who can turn complexity into clarity — and clarity into action.