Challenges Encountered and Lessons Learned when Using a Novel Anonymised Linked Dataset of Health and Social Care Records for Public Health Intelligence: The Sussex Integrated Dataset
Bibliographic record
Abstract
Background: In the United Kingdom National Health Service (NHS), digital transformation programmes have resulted in the creation of pseudonymised linked datasets of patient-level medical records across all NHS and social care services. In the Southeast England counties of East and West Sussex, public health intelligence analysts based in local authorities (LAs) aimed to use the newly created “Sussex Integrated Dataset” (SID) for identifying cohorts of patients who are at risk of early onset multiple long-term conditions (MLTCs). Analysts from the LAs were among the first to have access to this new dataset. Methods: Data access was assured as the analysts were employed within joint data controller organisations and logged into the data via virtual machines following approval of a data access request. Analysts examined the demographics and medical history of patients against multiple external sources, identifying data quality issues and developing methods to establish true values for cases with multiple conflicting entries. Service use was plotted over timelines for individual patients. Results: Early evaluation of the data revealed multiple conflicting within-patient values for age, sex, ethnicity and date of death. This was partially resolved by creating a “demographic milestones” table, capturing demographic details for each patient for each year of the data available in the SID. Older data (≥5 y) was found to be sparse in events and diagnoses. Open-source code lists for defining long-term conditions were poor at identifying the expected number of patients, and bespoke code lists were developed by hand and validated against other sources of data. At the start, the age and sex distributions of patients submitted by GP practices were substantially different from those published by NHS Digital, and errors in data processing were identified and rectified. Conclusions: While new NHS linked datasets appear a promising resource for tracking multi-service use, MLTCs and health inequalities, substantial investment in data analysis and data architect time is necessary to ensure high enough quality data for meaningful analysis. Our team made conceptual progress in identifying the skills needed for programming analyses and understanding the types of questions which can be asked and answered reliably in these datasets.
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 | 0.201 | 0.332 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.008 | 0.014 |
| Science and technology studies | 0.004 | 0.005 |
| Scholarly communication | 0.023 | 0.019 |
| Open science | 0.007 | 0.012 |
| Research integrity | 0.004 | 0.008 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
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".