FDA Collaborations

The Center collaborates with the Food and Drug administration through our Fellowship Program and through jointly developed research collaborations with our faculty and students. A few of these research collaborations are featured below.

High Yield Data Refinement and Analytics (HYDRA): A Machine Learning Platform for Regulatory Decision-Making

Ben Reis, Assistant Professor of Pediatrics at Harvard Medical School
Collaborating with the FDA Center for Devices and Radiological Health (CDRH)

CDRH is increasingly managing large volumes of complex, novel, and often disparate datasets across siloed and outdated systems that are difficult to connect for regulatory decision making. We are investigating whether new computational techniques, including machine learning and natural language processing, can enable CDRH to integrate data across databases and RWE data sources to analyze device performance, identify performance trends across devices and manufacturers, and detect signals faster and earlier in the life cycle of the device. This pilot project is assessing the feasibility of linking similar medical device submissions and adverse events to enable signal detection across medical devices.

Characterizing Disease Evolution in Non-Alcoholic Steatohepatitis (NASH)

Massimiliano Russo, Harvard-FDA Regulatory Science Fellow
Collaborating with the FDA Center for Drug Evaluation (CDER)

The complete natural history of Non-Alcoholic Fatty Liver Disease (NAFLD) and Non-Alcoholic Steatohepatitis (NASH) remains poorly understood, primarily due to a lack of studies with long term follow‐up periods. For NASH there are additional complications due to the absence of non-invasive techniques for diagnosis, which is currently based on liver biopsy. We are analyzing data from clinical trials and other sources to improve current understanding of NASH disease evolution. This project aims to characterize the natural variability of liver activity in NAFLD, analyzing variations of bio-chemical markers across disease stages (e.g., early NAFLD, early NASH, advanced NASH). The characterization of this variability is of primary importance in the evaluation of drug induced liver injuries (DILI), since it is currently difficult to discriminate DILI from liver injury incurred due to disease progression.

Predicting Drug-Drug Interactions Contributing to Drug Induced Liver Injury (DILI)

Kenichi Shimada, Postdoctoral Fellow at Harvard Medical School
Collaborating with the FDA Center for Drug Evaluation (CDER)

The primary goal of this project is to identify drug-drug interactions that induce liver injury using clinical trial data, and to elucidate the mechanisms of injury. In collaboration with the FDA, we are analyzing submitted clinical trials to find combinations of drugs and/or biologics that exhibit liver toxicity. We are particularly interested in tyrosine kinase inhibitor (TKI) chemotherapeutic combinations, and we aim to interpret the mechanisms of liver toxicity through pharmacokinetic or pharmacodynamic mechanisms.

Safety Signal Discernment and Biostatistics (SANEST) for Paclitaxel

Alejandra Avalos Pacheco, Harvard-FDA Regulatory Science Fellow
Collaborating with the FDA CDRH & the National Evaluation Systems for health Technology (NEST)

This project aims to provide recommendations on the analyses and necessary regulatory actions for the integration of randomized controlled trials (RCT) and real world data (RWD) for patients treated with paclitaxel-coated devices, including drug-coated balloons and drug-eluting stents. The work investigates the impact of misclassified devices on paclitaxel-coated device data as well as the limitations of data integration. Outputs include the development of guidelines for misclassification methods, data source quality assessment, and data integration methods.

Assessing Targets on the Pediatric Molecular Target List Under the RACE Act

Florence Bourgeois, Associate Professor of Pediatrics at Harvard Medical School
Collaborating with the FDA Office of Hematology and Oncology Products, CDER

Under the newly enacted RACE Act, the FDA is authorized to require pediatric studies for oncology drugs developed for adult populations if the drug targets a molecular target relevant to a pediatric cancer. This project aims to apply methods in natural language processing to identify emerging biomarkers that may be relevant to pediatric cancers, and to further characterize the scientific evidence underlying inclusion or exclusion of molecular targets on the Pediatric Molecular Target List. The project also involves a systematic review of pediatric oncology trial and drug development activities to establish benchmarks for assessing the impact of the RACE Act.