MIF Presents: 2026 Summer Webinar โ Student Presentation Competition
๐
Wednesday, July 15, 2026
๐ 1:00โ2:30 PM ET
๐ป Virtual via Zoom
โถ REGISTER NOW
About
Join the Minority Issues Forum (MIF) for its 2026 Summer Webinar, featuring short presentations from eight selected student participants in the MIF Summer Competition 2026. Presenters will share research spanning optimization, machine learning, healthcare analytics, risk management, and public policy. Each presentation is followed by audience Q&A, and competition winners will be selected through offline judging following the event.
Speakers
Gabriel Antonio Lawrence
Network-Based Dimensionality Reduction and Ensemble Inference for Inferring Reasons for Adolescent Substance Use
Using the Adolescent Brain Cognitive Development (ABCD) dataset, this research integrates Exploratory Graph Analysis (optimized with the DEFI fit index) with quartile random forests and distributional conformal inference to extract interpretable risk factors from high-dimensional data. The framework improves precision, sensitivity, and predictive stability โ and reveals that prior parental or child substance use, rather than socioeconomic indicators, is the largest driver of adolescent substance use.
Adrian Ma
From Trial-and-Error to Precision Mental Health: Evidence on Antidepressant Response
Antidepressant treatment remains highly trial-and-error, with first-line remission often only 28โ33%. This research combines information extracted from unstructured clinical notes with longitudinal EHR data to study heterogeneous treatment effects on PHQ-9 depression scores. Event-time models show symptoms peak just before treatment initiation and improve significantly after the first two weeks, with response varying by demographics, adherence, and major life events โ pointing to ways to shorten costly trial-and-error prescribing.
Valery Mfondoum
Smarter Decisions, More Relevant Signals: Density-Based Indicator Modeling and Sequential Subsampling for Risk-Aware Machine Learning in Venture Capital Project Selection
Identifying high-potential startups is one of the most critical โ and least data-driven โ decisions in venture capital. This presentation offers a unified, density-based framework that reinterprets startup valuation metrics as key risk, control, and performance indicators, maps them onto a โBowtieโ-style risk propagation model, and supports machine learning pipelines that remain robust to rare events, strategic uncertainty, and extreme market conditions.
Koushik Mondal
Channel Coexistence in the Indian Pharmaceutical Retail Market: Evidence from Unorganized, Organized, and Online Pharmacies
Why do independent unorganized pharmacies in India continue to thrive alongside organized retail chains and e-pharmacies? Combining field interviews, survey data, and conjoint analysis within an agent-based simulation, this research shows consumer choices depend on distance, urgency, trust, and service โ not just price. Channel coexistence emerges as a stable outcome, with policy implications for strengthening small pharmacies and preserving healthcare access in underserved communities.
Mohammad Pourmatin
Assessing Service System Vulnerability in U.S. Counties: A Pareto-Based Framework for Integrating Socio-Demographic, Built Environment, Energy, and Health Dimensions
This study examines how community vulnerability emerges from interacting socio-demographic, built environment, energy, and health conditions using a Service System Vulnerability Framework. Principal Component Analysis builds four subindices, combined through non-dominated Pareto ranking to classify all 3,108 U.S. counties. Results reveal strong spatial clustering โ notably in the Lower Mississippi Delta, Central Appalachia, South Texas, and Southern California โ showing vulnerability as a systemic condition that supports more equitable planning and resource allocation.
Dereje Geleta Oljira
Resilient Enterprise Investment and Innovation under Global Uncertainty: A Risk-Informed Approach
Global uncertainty โ from geopolitical instability to climate risk โ is reshaping how enterprises invest and innovate. This presentation introduces a risk-informed resilience framework that integrates systems analysis, simulation, and mixed-methods design to produce a measurable โResilience Index.โ Drawing on case insights from Ethiopian SMEs and global enterprises, the talk shows that resilience is about intelligent adaptation, not just resistance, and closes with actionable policy and practice recommendations.
Pedro Chumpitaz-Flores
A Global Optimization Algorithm for Interpretable Spectral Clustering of Millions of Samples
This work presents a global optimization algorithm for interpretable spectral clustering that identifies complex groups in graph and similarity data while representing clusters through simple decision-tree rules. An exact search procedure guarantees global optimality, accelerated by dynamic programming, valid lower bounds, streaming computation, and CUDA parallelization. Experiments solve a depth-2 problem with over 140 million observations to certified global optimality in roughly 31 minutes.
Chathuri Aththanayake
Interval-Based Pareto Pruning for Uncertain Multi-Objective Problems
Real-world problems in engineering, management, and agriculture often involve multiple conflicting objectives alongside uncertainty in input data and outcomes, producing solution sets too large for practical decision-making. This research introduces a newly defined index for interval-based uncertain solutions, applied within a constrained similarity-based clustering model, along with a new stochastic dominance rule that identifies the Stochastic Pareto Set. Computational analysis demonstrates the index's effectiveness in pruning stochastic non-dominated solutions โ giving decision-makers a more tractable and informative solution set for uncertain multi-objective environments.
Registration
๐ต REGISTER HERE
Contact
Himadri Sen Gupta
himadri.gupta@csupueblo.edu