NHash: Randomized N-Gram Hashing for Distributed Generation of Validatable Unique Study Identifiers in Multicenter Research
Bibliographic record
Abstract
BACKGROUND: A unique study identifier serves as a key for linking research data about a study subject without revealing protected health information in the identifier. While sufficient for single-site and limited-scale studies, the use of common unique study identifiers has several drawbacks for large multicenter studies, where thousands of research participants may be recruited from multiple sites. An important property of study identifiers is error tolerance (or validatable), in that inadvertent editing mistakes during their transmission and use will most likely result in invalid study identifiers. OBJECTIVE: This paper introduces a novel method called "Randomized N-gram Hashing (NHash)," for generating unique study identifiers in a distributed and validatable fashion, in multicenter research. NHash has a unique set of properties: (1) it is a pseudonym serving the purpose of linking research data about a study participant for research purposes; (2) it can be generated automatically in a completely distributed fashion with virtually no risk for identifier collision; (3) it incorporates a set of cryptographic hash functions based on N-grams, with a combination of additional encryption techniques such as a shift cipher; (d) it is validatable (error tolerant) in the sense that inadvertent edit errors will mostly result in invalid identifiers. METHODS: NHash consists of 2 phases. First, an intermediate string using randomized N-gram hashing is generated. This string consists of a collection of N-gram hashes f1, f2, ..., fk. The input for each function fi has 3 components: a random number r, an integer n, and input data m. The result, fi(r, n, m), is an n-gram of m with a starting position s, which is computed as (r mod |m|), where |m| represents the length of m. The output for Step 1 is the concatenation of the sequence f1(r1, n1, m1), f2(r2, n2, m2), ..., fk(rk, nk, mk). In the second phase, the intermediate string generated in Phase 1 is encrypted using techniques such as shift cipher. The result of the encryption, concatenated with the random number r, is the final NHash study identifier. RESULTS: We performed experiments using a large synthesized dataset comparing NHash with random strings, and demonstrated neglegible probability for collision. We implemented NHash for the Center for SUDEP Research (CSR), a National Institute for Neurological Disorders and Stroke-funded Center Without Walls for Collaborative Research in the Epilepsies. This multicenter collaboration involves 14 institutions across the United States and Europe, bringing together extensive and diverse expertise to understand sudden unexpected death in epilepsy patients (SUDEP). CONCLUSIONS: The CSR Data Repository has successfully used NHash to link deidentified multimodal clinical data collected in participating CSR institutions, meeting all desired objectives of NHash.
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 | 0.012 | 0.039 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.003 | 0.006 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.008 | 0.004 |
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".