Developing a prototype for federated analysis to enhance privacy and enable trustworthy access to COVID-19 research data
Bibliographic record
Abstract
BACKGROUND: The use of federated networks can reduce the risk of disclosure for sensitive datasets by removing the requirement to physically transfer data. Federated networks support federated analytics, a type of privacy-enhancing technology, enabling trustworthy data analysis without the movement of source data. OBJECTIVES: To set out the methodology used by the International COVID-19 Data Alliance (ICODA) and its partners, the Secure Anonymised Information Linkage (SAIL) Databank and Aridhia Informatics in piloting a federated network infrastructure and consequently testing federated analytics using test data provided from an ICODA project, the International Perinatal Outcome in the Pandemic (iPOP) Study. To share the challenges and benefits of using a federated network infrastructure to enable trustworthy analysis of health-related data from multiple countries and sources. RESULTS: This project successfully developed a federated network between the SAIL Databank and the ICODA Workbench and piloted the use of federated analysis using aggregate-level model outputs as test data from the iPOP Study, a one-year, multi-country COVID-19 research project. This integration is a first step in implementing the necessary technical, governance and user experiences for future research studies to build upon, including those using individual-level datasets from multiple data nodes. CONCLUSIONS: Creating federated networks requires extensive investment from a data governance, technology, training, resources, timing and funding perspective. For future initiatives, the establishment of a federated network should be built into medium to long term plans to provide researchers with a secure and robust data analysis platform to perform joint multi-site collaboration. Federated networks can unlock the enormous potential of national and international health datasets through enabling collaborative research that addresses critical public health challenges, whilst maintaining privacy and trustworthiness by preventing direct access to the source data.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.010 | 0.025 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.005 | 0.008 |
| Open science | 0.004 | 0.006 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.009 | 0.005 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".