research

publications

Research across language models, computational biology, and health. The latest citation record is on Google Scholar.

selected publications

Diagram of the Leak It self-concentration training-data extraction method

Leak It: Per-Document Extraction Beyond Aggregate Membership Inference

Victor Maricato

arXiv preprint, 2026

A probabilistic method for extracting and quantifying memorized training data from black-box language models without white-box access to model internals.

arXiv code

Overview of the GLOF variant benchmark by functional class

GLOF: A large-scale expert-curated benchmark dataset of gain-of-function and loss-of-function missense variants

V Maricato, D Schlesinger, PN de Souza Moura

bioRxiv preprint, 2026

Benchmark dataset of 112,399 missense variants across 2,809 genes, each labeled as GOF, LOF, or neutral by board-certified clinical geneticists. Public on Kaggle and Hugging Face.

bioRxiv code

Suicide mortality trends and age-specific rate ratios for Indigenous people in Brazil

Suicide Mortality among Indigenous Peoples in Brazil, 2010-24: A Nationwide Registry-Based Study

Victor Maricato

SSRN preprint; poster accepted, XLIII Congresso Brasileiro de Psiquiatria (CBP) 2026, 2026

A nationwide registry-based re-estimation of Indigenous suicide mortality in Brazil through 2024, using the complete national mortality registry and the 2022 Census to characterize its age, sex, regional, and method structure.

SSRN

DispatchQA model performance measured by latency and function-calling F1 score

DispatchQA: A Benchmark for Small Function Calling Language Models in E-Commerce Applications

Joachim Daiber, Victor Maricato, Ayan Sinha, Andrew Rabinovich

EMNLP (Industry Track), 2025

A benchmark and methodology for evaluating small function-calling language models on realistic e-commerce dispatch tasks under tight latency budgets.

paper code

other publications

Prediction of Gain or Loss of Function in Missense Variants

VM Oliveira, PN de Souza Moura

Simpósio Brasileiro de Sistemas de Informação (SBSI), 2025

ML approach to predict whether a missense variant produces a gain- or loss-of-function effect, with applications in clinical variant interpretation.

PDF

Graph neural networks: A bibliometric mapping of the research landscape and applications

AMB da Silva, NCS Ferreira, LAM Braga, FB Mota, V Maricato, LA Alves

Information, 2024

A bibliometric study mapping the GNN research landscape across applications and sub-fields.

paper

Graph neural networks as a potential tool in improving virtual screening programs

LA Alves, NCS Ferreira, V Maricato, AVP Alberto, EA Dias, et al.

Frontiers in Chemistry, 2022

Review of graph neural network architectures applied to virtual screening pipelines for drug discovery.

paper

Quantification of schizophyllan directly from the fermented broth by ATR-FTIR and PLS regression

FRP Mansoldo, V da Silva Cardoso, AN Junior, SML Cedrola, V Maricato, et al.

Analytical Methods, 2020

PLS-regression model on ATR-FTIR spectra to quantify schizophyllan directly in fermented broth, avoiding costly downstream isolation.

paper