Starting in Fall 2019, the mathematical biology group is officially a mathematical biology SMART Team as part of the NSF Funded NSF Funded Data-Intensive Research And Computing (DIRAC) Research Training Group (RTG).
We meet weekly to discuss reearch involving the application of mathematical tools (such as mathematical modeling, computational simulation) to the study of biological systems. Our topics cover a range of biological processes including, but not limited to, protein aggregation, population genetics & structural variation, and dynamics of marine animals.
Our group is led by Professor Suzanne Sindi (ssindi@ucmerced.edu) and consists of graduate students, postdocs and faculty in the Applied Mathematics, Physics and Quantitative & Systems Biology graduate programs.
For Spring 2026, we will be meeting Wednesday from 3:30pm-4:20pm in ACS 362B. We will post our updated schedule on this website once it is complete. You can follow our activities at our Twitter Account.
If you are not on the slack channel but want to attend presentations, please email Prof. Erica Rutter (erutter2@ucmerced.edu) for access.
Spring 2026 Schedule
Wednesday, March 4:
Speaker: Masato Terasaki, UC Merced
Title: A Physics-Informed Framework for Glioblastoma Modeling
Abstract: Glioblastoma multiforme (GBM) remains one of the most aggressive primary brain tumors, with poor prognosis driven in part by invisible tumor infiltration beyond MRI-visible boundaries. Patient-specific biophysical modeling offers a path toward personalized treatment planning, but faces a fundamental challenge: pre-treatment clinical data is sparse, noisy, and often limited to at most two imaging timepoints. This work introduces a framework that synergistically integrates physics-based simulation and machine learning to perform robust parameter inference from sparse observations. Using the Fisher-KPP reaction-diffusion equation as a forward model, we generate a dense synthetic training library of tumor growth trajectories sampled at high temporal resolution, and train an LSTM encoder-decoder to invert sparse radial measurements into patient-specific parameters. The core methodological contribution is a three-phase curriculum learning strategy that trains the network on dense synthetic trajectories before transitioning to sparse clinical observations, bridging the domain gap between abundant simulation data and limited real measurements. Systematic evaluation demonstrates that curriculum learning provides measurable advantages under data scarcity (N 500) and maintains robustness under clinically realistic measurement noise (σ > 1%). Critically, the framework remains effective with as few as L = 10 to 16 temporal observations, retaining over 90% of full performance and suggesting practical deployability without requiring infeasible imaging frequencies. The framework is further extended to treatment modeling, simulating surgical resection and chemotherapy response using inferred patient-specific parameters
Wednesday, February 25:
Speaker: Tracey Oellerich, UNC
Title: Machine Learning–Based Prediction of Bleeding Risk in Factor XI Deficiency
Abstract: Factor XI (FXI) deficiency is a rare bleeding disorder with highly variable clinical manifestations, ranging from asymptomatic cases to severe bleeding tendencies. Predicting bleeding risk in FXI-deficient patients remains a challenge due to the complex interplay between genetic, biochemical, and environmental factors. Current clinical assays fail to reliably predict bleeding risk, highlighting the need for alternative approaches. Here, we apply machine learning models that integrate clinical characteristics with proteomic profiles from FXI‑deficient patients to predict individual bleeding phenotype. This approach identifies patterns associated with bleeding tendencies and improves predictive accuracy over traditional assays, offering a promising path toward more precise risk assessment.
Wednesday, February 18:
Speaker: Morgan Lavenstein-Bendall, UC Merced
Title: Computational Recovery and Interpolation of Age-Dependent Mortality from Short-Term Temperature Experiments
Abstract: Due to their diversity and abundance, insects play essential ecological roles, including crop pollination, nutrient cycling, and serving as a food source for other species. However, climate change is predicted to heavily impact insect populations, with some expected to decline by up to 18% globally by the end of the 2020s, raising concerns about the future health of the bioeconomy. To investigate these impacts, we conducted a controlled temperature study on Aster leafhoppers (Hemiptera: Cicadellidae: Macrosteles quadrilineatus). Using five constant temperature conditions, we collected physiological data over a month to assess the impact of temperature on survival, maturation, and egg production. We then developed an age-structured population model to examine how thermal change influences insect fitness and mortality rates. For each temperature treatment, we parameterized the mortality function separately and then interpolated across temperatures to evaluate how gradual changes in thermal environments may shape population dynamics. This approach enables us to assess how populations experiencing different thermal regimes may respond to environmental change, providing insight into the potential evolutionary and demographic consequences of sustained warming across spatial landscapes.
Wednesday, February 11:
Speaker: Suraj Sahu, UC Merced
Title: The Existential Dread and AI Agents: A Scientific Researcher's Manifesto for Sustainable, Context-Aware AI Integration.
Abstract: Despite the excitement surrounding AI integration in academic research, a lack of understanding of the capabilities and limitations of Large Language Models (LLMs) often leads to hallucinations, buggy or over-engineered code, and data privacy concerns. While AI chatbots have become remarkably powerful (and potentially dangerous), aligning user expectations with model responses can be very frustrating. In this talk/tutorial, I will show examples of context engineering and effective prompting techniques that can help you navigate large codebases and projects. We will discuss agentic capabilities that leverage tools like MCPs, skills, commands, rules, and planning modes. We will explore how custom instructions can guide AI towards giving significantly better results while maintaining your data local and private. Additionally, I will demonstrate how to create custom agents and MCPs for research-specific tasks like data analysis, image analysis, and literature organization. As LLMs consume significant energy in training and inference, understanding their strengths and limitations is essential for sustainable integration.
*Keywords: Agents, Skills, MCPs, Commands, Rules, Image Analysis, Data Analysis, Context Engineering, Custom Workflows*
Wednesday, February 4:
Speaker: Dr. Baltazar Gonzalez, UC Merced
Title: Multiple global change drivers reshape the spatiotemporal dynamics of bee--plant interactions: priorities for California’s blue orchard bee
Abstract: Rapid climate change is a primary driver of the ongoing biodiversity loss, with impacts that span both natural and anthropogenic landscapes. Pollination lies at the center of this challenge, underpinning agricultural production while sustaining native plant communities and ecosystem functioning. Understanding how climate change alters pollination systems is therefore essential for ensuring food security and ecosystem integrity. Among pollinators, bees are the most important, representing an extraordinary diversity of species. Solitary bees are particularly critical in this context: they are often more efficient pollinators than their social counterparts and account for over 80% of global bee diversity. In this talk, I will present a modeling framework combining multiple modeling strategies with different, large-scale datasets—including high-resolution climate projections, socioeconomic and land-use scenarios, species distribution models, and climate-related hazards—to assess the future of pollination services. I will focus on California, a region of exceptional biodiversity, and global agricultural importance. By integrating projections of pollinator suitability with spatial estimates of plant resource availability, I analyze how pollination systems may respond jointly to climate change. From these models, we develop a novel spatially explicit conservation priority index that integrates pollinator habitat suitability to projected changes in plant resources. This index explicitly incorporates information on ecological interaction, allowing raw model predictions to be modulated by expected species responses to shifting resource landscapes. By doing so, it provides a reproducible and operational framework for evaluating the vulnerability and persistence of pollination services across space and time, and for identifying priority areas where conservation and management actions can be most effective for the long-term persistence of both agricultural and natural areas.
If you want to be added to our mailing list please email ssindi@ucmerced.edu.
