Bibliographic record
Abstract
If someone was to ask you to consider important advances in medicine, perhaps like me, you would think of Fleming's discovery of penicillin in the 1940s or the introduction of the portable defibrillator by Pantridge and team in the 1960s. I wonder if you would include today's health gadgets like the Fitbit.(1) Another, quieter gradual revolution is going on in Northern Ireland at the minute.(2),(3) When I started my Consultant career in 1995, it was quite commonplace for a query from a general practitioner to require access to the notes which might take about a month to arrive back with me from Medical Records (who knows what journey those notes went on). My answer to the query would then be dictated on to a cassette tape which would then join a row of similar tapes in a little plastic rack until my secretary reached the correct one. If the GP, frustrated by the delay, rang to find out what was happening, my secretary had to listen to the entire tape from the beginning to find the right segment. All of this seems faintly ludicrous some 20 years later, sitting in my home office, dictating into a desk microphone and watching these words appear a second later on my PC. The 6 Trusts in Northern Ireland are responsible for the healthcare of 1,800,000 people. Care is becoming more complex and often patients will attend more than one hospital site during their treatment. Many patients require intricate medication schedules which are changed frequently by general practitioner or hospital doctor. The “analogue era” of the 1990s can no longer meet our needs. The Northern Ireland Electronic Care Record project was initially mooted in 2005 as a solution to these problems. It was recognised that one Trust alone could not afford to develop the infrastructure required for a viable project. In 2008, site visits to innovative centres in the USA and Canada led to a regional ICT programme board approving a “proof of concept” trial involving case records in Belfast City Hospital, Ulster Hospital Dundonald and 2 large family practices. The project went “live” in January 2010. The aim of NIECR was to provide a single portal for viewing multiple sources of clinical information via a single logon to a single system which would eventually replace the multiplicity of laboratory, imaging, clinical and pharmacy applications that one must switch between to have a meaningful clinical encounter. Converting information held in isolated proprietary systems proved to be a major hurdle for NIECR but so far, more than 16 separate patient information systems have been incorporated into the dataset and there are plans to incorporate not only imaging reports but also the digital image files themselves. Consent to release of data and auditing of access formed a central part of the project with administrators able to review audit trails, especially in the setting of a “break the glass” privacy override. 5000 patient records were reviewed in the “proof of concept” phase. 78% of accesses were with full patient consent and 20% were with privacy overrides. 120 patients opted out of the system through their family practice – none opted out from a clinical setting. 74% of doctors surveyed felt that the new system led to a more rapid and correct diagnosis and 33% reported occasions when the system drew attention to a possible adverse event such as prescribing medication with a history of allergy. Following the successful evaluation phase, NIECR started rolling out across the province in 2012. The system works using a master patient index number based on a unique 10 digit “H&C” number – old hospital number prefixes like CAH, AH or RV are becoming relics of the past. Since roll-out, the system has become widely adopted throughout the province and the statistics are quite staggering:
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 | 0.003 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.006 | 0.002 |
| Scholarly communication | 0.014 | 0.008 |
| Open science | 0.001 | 0.010 |
| Research integrity | 0.005 | 0.008 |
| Insufficient payload (model declined to judge) | 0.239 | 0.161 |
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".