December 10, 2025
Preprint: Mitigating Exposure Bias in Risk-Aware Time Series Forecasting with Soft Tokens
Autoregressive forecasting in diabetes and hemodynamic management can suffer from exposure bias under teacher forcing, yielding unstable multi-step predictions. We introduce Soft-Token Trajectory Forecasting (SoTra), which propagates continuous probability distributions ("soft tokens") to learn calibrated, uncertainty-aware trajectories, plus a risk-aware decoding module that minimizes expected clinical harm. SoTra reduces zone-based glucose risk by 18% and lowers blood-pressure clinical risk by about 15%, supporting safety-critical predictive control.
December 2025
$4.7 million support to Forlenza–Shakeri Project for Next-Gen Diabetes Tech
We are thrilled to announce that Breakthrough T1D and the Helmsley Charitable Trust have awarded a $3.9 million grant to the Forlenza–Shakeri project, a multi-center initiative co-led by Dr. Shakeri as Contact PI alongside Dr. Greg Forlenza (CU Anschutz). UVA will lead algorithm development and engineering work packages, operating alongside clinical sites at UVA, CU (Barbara Davis Center), and UCSF to advance next-generation automated insulin delivery systems. The initiative is bolstered by $800,000 in in-kind contributions from Tandem Diabetes Care and Arecor, bringing total support to nearly $4.7 million.
December 2025
LaunchPad Funding for Diabetes Foundation Model
We received $200,000 in LaunchPad funding to develop the first foundation model for Type 1 and Type 2 diabetes. Dr. Shakeri will co-PI with Boris Kovatchev and Sue Brown; the work will unify databases, pretrain, and validate on FDA-accepted simulators, targeting clinical translation within 12–18 months for initial applications.
December 2025
SDS Center Grant Development Funding
The School of Data Science awarded Dr. Shakeri $55,000 in 2025 SDS Center Grant Development funding for the project “Center Grant Development for NIH P30 Diabetes Research Center Resubmission.”
December 2025
Funding for Home-Based CGM Screening Test
We received $100,000 to build a home-based CGM screening test for early detection of Type 1 diabetes progression. Dr. Shakeri will co-PI with Leon S. Farhi and Mark DeBoer to develop algorithms, engage FDA pre-submission, and initiate industry partnerships with CGM manufacturers to identify candidates for disease-modifying therapies like Teplizumab before clinical diagnosis.
November 30, 2025
Preprint: Fast Algorithm for Minimum Weight Cycles
We released the preprint 'A Fast Algorithm for Finding Minimum Weight Cycles in Mining Cyclic Graph Topologies,' introducing a deterministic Dijkstra-inspired method with pruning heuristics to accelerate minimum weight cycle search and loop modulus computations.
November 25, 2025
Preprint: Driver-Blindness in Blood Glucose Forecasting
Dr. Shakeri's preprint 'The Driver-Blindness Phenomenon: Why Deep Sequence Models Default to Autocorrelation in Blood Glucose Forecasting' formalizes why deep models often ignore insulin, meals, and activity drivers, proposing mitigation strategies and calling for routine Δdrivers reporting.
November 2025
Preprint: Fusing Biomechanical and Spatio-Temporal Features for Fall Prediction
Vision-based fall prediction is constrained by scarce fall data. We propose the Biomechanical Spatio-Temporal Graph Convolutional Network (BioST-GCN), a dual-stream model that fuses pose and biomechanical signals via cross-attention. BioST-GCN improves F1 by 5.32% on MCF-UA and 2.91% on MUVIM over ST-GCN, and the attention mechanisms highlight critical joints and temporal phases, while a simulation-to-reality gap remains.
November 2025
Paper Accepted for Oral Presentation
Our paper 'Anomaly detection in brain MRI: a neural discrete representation learning with contrastive loss' has been accepted for oral presentation at the SPIE Medical Imaging Meeting (15-19 February 2026).
November 2025
Dr. Shakeri Joins NDIF External Scientific Advisory Board
Dr. Shakeri has joined the National Deep Inference Fabric (NDIF) External Scientific Advisory Board, supporting efforts to advance AI interpretability, interdisciplinary research, and responsible technology development.
October 2025
Invited Speaker at SIAM Central States Section Meeting
Dr. Shakeri will present "Operator-Theoretic Meal Detection: Windowed DMD, Stability Cues, and Translation to Closed-Loop Care" within the mini-symposium "Interactions among analysis, optimization and network science" at the 10th Annual Meeting of the SIAM Central States Section.
October 2025
Paper Published in Applied Soft Computing
Our paper 'Graph-embedded reinforcement learning for dynamic pricing and advertising under network effects' has been published in Applied Soft Computing.
July 2025
Paper Accepted to ICML 2025
Our paper 'Multi-Marginal Stochastic Flow Matching for High-Dimensional Snapshot Data at Irregular Time Points' has been accepted to ICML 2025.
March 22, 2025
Paper Accepted as Spotlight at ICLR MLGenX 2025 Workshop
Our paper 'Learning Non-Equilibrium Signaling Dynamics in Single-Cell Perturbation Dynamics' has been accepted as a spotlight paper at the ICLR MLGenX 2025 Workshop.
March 18, 2025
Forbes Features UVA's AI-Driven Diabetes Management Research
UVA professor Heman Shakeri's AI-driven insulin delivery system using reinforcement learning was featured in Forbes magazine, highlighting benefits for children with Type 1 diabetes.
March 3, 2025
Paper Accepted at Engineering Diabetes Technology 2025
Our paper 'Online Meal Detection Based on CGM Data Dynamics' has been accepted for oral presentation at the IFAC EDT 2025, to be held in Valencia, Spain, May 8-9, 2025.
February 20, 2025
Invited Speaker at SIAM Conference on Dynamical Systems
Dr. Shakeri will be an invited speaker in the 'Koopman Operator in Control - Part II of II' session at the 2025 SIAM Conference on Dynamical Systems, presenting 'Control of Complex Network Dynamics with Koopman: Tweaking Local Interactions for Desired Emergent Dynamics'.
January 29, 2025
Capital One Data Design Fellowship
Dr. Shakeri has been appointed a School of Data Science Capital One Data Design Fellow for 2025-2026. The project will use Multi-Marginal Stochastic Flow Matching (MMSFM) to capture the dynamics of high frequency financial transactions, with applications to risk management, trading strategy development, and market stability analysis. The fellowship comes with a $55,000 discretionary fund.
January 15, 2025
Computational Genomics Pilot Award
Our lab received a $25,000 Computational Genomics and Data Science Pilot Award (FY2025) for the project 'Sex differences in the melanoma tumor microenvironment and clinical outcomes after immune checkpoint inhibitor therapy'.
January 2025
Two Abstracts Accepted at 2025 ARVO Annual Meeting
The lab will present two abstracts at the 2025 ARVO Annual Meeting (May 4-8): 'An Automated Pipeline for Converting Unstructured Clinical Text to Standardized Measurements,' and 'Identification of Visual Impairment Risk Factors in Southwest Virginia'.
December 20, 2024
Paper Accepted at ATTD 2025
Our work on 'Deep Reinforcement Learning Bolus Priming for Intelligent Prandial Management' has been accepted for presentation at the 18th International Conference on Advanced Technologies & Treatments for Diabetes (ATTD 2025) in Amsterdam, The Netherlands, March 19-22, 2025.
December 20, 2024
LaunchPad Grant Award
Shakeri Lab received a $200k grant from LaunchPad for Diabetes to pilot trial the use of advanced ML tools in Artificial Pancreas technology. The program supports innovative solutions for Type 1 or Type 2 diabetes treatment, focusing on translational research projects that address unmet clinical needs.
November 2024
Paper Accepted at NeurIPS Workshop
Our paper was accepted at NeurIPS Workshop on Responsibly Building the Next Generation of Multimodal Foundational Models.
November 2024
Brain Institute Grant Award
Brain Institute awarded us $20k to develop ML methods for Characterizing Neural Dynamics of Auditory Reconstruction in the Central Auditory System.
October 2024
Invited Talk at SIAM
Dr. Shakeri presented 'Enhancing Network Design and Dynamics through Spectral and Topological Analysis' at the 9th SIAM Annual Meeting of Central States Section, Kansas City.
October 2024
Presentation at NIH Workshop
Our work 'BPS-RL: Reinforcement Learning Trained Bolus Priming System' was presented at the NIDDK AI Workshop on Artificial Intelligence in Precision Medicine of Diabetes.
October 2024
Cancer Center Grant Award
UVA Comprehensive Cancer Center awarded us $42,500 to develop Multiscale Computational and Experimental Framework for Analyzing Melanoma Cell Drug Responses through Stochastic Dynamics.
September 2024
Paper Published in Network Science
Our paper 'The art of interconnections: Achieving maximum algebraic connectivity in multilayer networks' was published.
July 2024
ACC 2024 Session Chair
Dr. Shakeri served as the session chair for Machine Learning at the 2024 American Control Conference in Toronto.
July 2024
Paper Published at ACC 2024
Our paper 'Operator-Based Detecting, Learning, and Stabilizing Unstable Periodic Orbits of Chaotic Attractors' was published.
June 2024
International Talk
Dr. Shakeri presented at the 44th International Symposium on Forecasting in Dijon, France.
May 2024
Paper Published in PLOS ONE
Our paper 'MAD-FC: A fold change visualization with readability, proportionality, and symmetry' was published.
March 2024
Paper Published in Frontiers
Our paper 'Biophysical modulation and robustness of itinerant complexity in neuronal networks' was published in Frontiers in Network Physiology.
2024
UVA Research Communications Fellow
Dr. Shakeri is named UVA Research Communications Fellow. The six-month program provides media training to enhance faculty's ability to discuss research with lay audiences.
July 2023
NIH NCBI R01 Grant Award
Our team received an NIH R01 grant for 'Optimizing Treatment Decision Making for Patients with Localized Renal Masses'. This $1,637,195 grant runs through June 2027. Dr. Shakeri serves as Co-Investigator on this project.
July 2023
Paper Published in ISA Transactions
Our paper 'A purely data-driven framework for prediction, optimization, and control of networked processes' has been published.
April 2023
Oracle for Research Award
Received $50k Cloud credits from Oracle for Research to cover Cloud resources.
March 2023
Preprint Available: Contra-Analysis
Our paper 'Contra-Analysis for Determining Negligible Effect Size in Scientific Research' is now available on arXiv.
2022
Paper Published in SIEDS
Our paper 'GeoTyper: Automated Pipeline from Raw scRNA-Seq Data to Cell Type Identification' was presented at SIEDS 2022.
2022
Preprint: Cortical Metastability
Our paper 'A Simple Model of Cortical Intraregional Metastability' is available on bioRxiv.
May 2021
3 CAVALIERS RAPID SEED GRANT Awarded
Received $60,000 for our project 'Dissecting the origins of heterogeneous cancer cellular interactions and responses to therapeutic perturbation'. Dr. Shakeri is the Principal Investigator on this grant.
August 2020
NCI COVID-19 Grant
Received $426,972 two-year National Institutes of Health grant for 'Risk Prediction for COVID-19: Vibrent Health/UVA'. Dr. Shakeri serves as Co-Investigator.
July 2020
Ivy Foundation Grant Awarded
Received $100,000 from Ivy Foundation for 'Epidemiologic Modeling, Public Health Surveillance and Sewershed Monitoring to Predict Surges in the COVID-19 Pandemic'. Dr. Shakeri serves as Principal Investigator.
2020
Paper Published in Physical Review E
Our paper 'Designing optimal multiplex networks for certain Laplacian spectral properties' has been published.
