Conducted by the Science & Innovation Desk

Mohammad Towhidul Islam is a data-analytics researcher and practitioner whose work spans predictive analytics, AI-enabled decision support, financial-risk forecasting, data governance, customer-service optimization, and supply-chain planning. He holds a PhD in Information Technology from the University of the Cumberlands and a Master of Science in Business Analytics from Trine University. His research is organized around a practical question: how can organizations turn complex, high-volume data into decisions that are accurate, timely, secure, transparent, and defensible? Rather than treating analytics as an isolated technical exercise, he examines the entire decision environment, including data quality, operational constraints, governance requirements, model uncertainty, and human accountability.
In this conversation, Mr. Islam discusses the problems his models are designed to solve, why trustworthy analytics requires more than predictive accuracy, and how he plans to extend his work across enterprise, public-sector, and infrastructure settings. The following transcript has been edited for length and clarity.

Background and Research Direction
Q: Your research spans financial forecasting, supply-chain analytics, data governance, and service optimization. What connects these different areas?

The common thread is decision support under real-world constraints. Most organizations possess more data than they have ever had, but data by itself does not make a decision. The difficult point comes when a manager, analyst, or public agency must interpret that information and act.
My work focuses on closing the distance between raw information and a trustworthy decision. I develop models that transform complex and sometimes incomplete data into forecasts, risk indicators, and practical recommendations. Just as importantly, I examine whether those recommendations can be explained, reviewed, and defended.

Q: What specific problem are you trying to solve through your research?
The central problem is that many organizations have large amounts of data but still lack a dependable way to convert that data into timely and defensible action. Information is often fragmented across different systems, collected at different levels of quality, or presented without enough context for a manager to understand what the model is recommending and why.

My research seeks to solve that problem by developing integrated, AI-enabled decision-support frameworks that bring forecasting, data governance, explainability, uncertainty monitoring, and human oversight into the same system. The goal is not simply to generate a prediction. It is to help an organization identify risk earlier, compare practical alternatives, allocate resources more effectively, and preserve a clear record of how the final decision was reached.

Q: You hold a PhD in Information Technology from the University of the Cumberlands. How did your doctoral training change the way you approach analytics research?
The PhD strengthened my ability to connect technical design with organizational implementation. Information technology is not only about building a model or selecting an algorithm; it is also about understanding the systems, people, processes, and governance structures through which that model will operate.

My doctoral training encouraged me to examine analytics as part of a larger decision architecture. That means asking how data is collected, who has access to it, how model outputs are communicated, what happens when the model is uncertain, and who remains accountable for the final decision.

Trustworthy and Auditable Analytics
Q: What does a transparent and auditable forecasting system look like in practice?
It begins with traceable data. The organization should know where the data came from, how it was transformed, which variables entered the model, and whether the information met appropriate quality standards.

The second requirement is an interpretable decision process. The model should provide more than a final score. It should identify the factors that contributed most strongly to the result, indicate the level of confidence, and make unusual or unstable conditions visible to the reviewer.
The third requirement is documented accountability. High-impact recommendations should remain subject to review by qualified professionals, and the organization should preserve a record of how the final decision was reached. In my view, explainability and auditability are not optional additions to predictive performance. They are part of performance because they determine whether the model can be used responsibly.

Q: In customer-service and hospitality-type operations, where should automation end and human judgment begin?
A considerable amount of value is lost through small but repeated inefficiencies: delayed responses, mismatched staffing, repetitive administrative work, inconsistent handling of requests, and interactions that are not directed to the appropriate employee. AI can improve speed and consistency by supporting routine, predictable parts of the service process. That allows employees to concentrate on situations that genuinely require judgment, empathy, negotiation, or specialized knowledge.

However, speed should not be confused with service quality. An automated response may be delivered quickly and still misunderstand the customer's actual problem. The system must recognize uncertainty, identify cases outside its reliable operating range, and transfer those interactions to a qualified person.

Research Impact and Future Direction
Q: Researchers naturally believe their own work is valuable. What evidence tells you that your research has been useful to others?
The more meaningful signal is whether researchers with no professional or authorship connection to me find the work useful enough to cite, discuss, or build upon. My research has been cited by independent scholars working across fields that include finance, healthcare, infrastructure, operations, and information systems. That cross-disciplinary use is important because it suggests that the underlying problems and frameworks are relevant beyond the immediate context in which they were originally developed.

Q: With your PhD in Information Technology completed, what are the next directions for your research?
I see three principal directions. The first is enterprise decision-support systems in which forecasting, governance, risk analysis, and operational reporting are integrated into a unified platform. These systems should be usable not only by large corporations with extensive data-science teams but also by small and mid-sized organizations that need advanced analytics without having the resources to build a complete infrastructure from the beginning.

The second direction is public-sector and infrastructure decision-making. Government agencies and public-service organizations manage large amounts of operational data. Reliable forecasting can help them allocate resources, identify emerging risks, and improve service delivery. In those settings, transparency and accountability are especially important because model-supported recommendations may affect communities and public resources.

The third direction is the continued development of AI-enabled decision-support frameworks with built-in governance. I want to bring predictive analytics, data protection, explainability, uncertainty monitoring, and managerial oversight into a common architecture. The system should not only generate a recommendation; it should also document how that recommendation was produced and alert decision-makers when data quality or model confidence falls outside an acceptable range.

About Mohammad Towhidul Islam
Mohammad Towhidul Islam is a data-analytics researcher and practitioner working at the intersection of predictive analytics, AI-enabled decision support, financial-risk forecasting, data governance, customer-service optimization, and supply-chain planning. He holds a PhD in Information Technology from the University of the Cumberlands and a Master of Science in Business Analytics from Trine University. His peer-reviewed research addresses transparent financial forecasting, responsible data-protection frameworks, simulation-based inventory control, and AI-supported service operations.

Email: towhidulislamshovan@gmail.com
THE DAILY MORNING VOICE | SCIENCE & INNOVATION