Register Desk, a Jambo Labs demo

Overnight, North ValeBehind the scenesExtracts arrive from GP clinical systems, community services, the hospital, death registrations and practice lists. Each is loaded, checked for completeness and matched on the NHS number, which never leaves the secure platform.What is automatedCollecting, loading and checking every extract, every night.Could go furtherAlert the practice when its extract fails, before the numbers go out.

02:10GP recordsclinical events, 24 practicesLoaded
02:45Community servicescontacts and rehabLoaded
03:30HospitalA&E and admissionsLoaded
04:05DeathsregistrationsLoaded
04:40Practice listswho is registered whereLoaded

23 of 24 practice extracts in. Practice 17's is missing, so its last good extract is used and every figure for it is marked.

Today An analyst downloads extracts and rebuilds the register in spreadsheets each quarter.

Patient 4172, 68Behind the scenesYearly snapshots only show who is on a practice's register today. Rebuilding from every event keeps the first diagnosis, corrects codes entered in error, and follows the patient when they move practice.In this demoInvented patient and dates. The rebuild rules are the kind we have built for real registers.What it achievesNew diagnoses are counted once, trends stay true across years, and nobody falls off the register by moving.

20182019202020212022202320242025

2018 COPD diagnosed at Practice 07
2020 COPD marked resolved in error
2021 Re-coded as COPD after spirometry
2023 Moved to Practice 12
2025 Two admissions for exacerbation

Yearly snapshots saya new diagnosis in 2023

Events rebuildon the register since 2018

436 patients the snapshots counted as new this year had simply moved practice.

573 were truly new diagnoses.

Achievement by practice, % of patients, first 12 of 24Behind the scenesEach measure is a rule over the rebuilt register: who is in the denominator, what counts, and in which window. 40 measures in 8 groups, each split by deprivation, ethnicity, age, sex and neighbourhood.AssumptionThe measure set is ours, of the kind primary care reports on COPD. For a real build it is checked line by line against the current national indicators.Could go furtherThe same pipeline carries asthma, diabetes, heart failure and frailty registers.

	Diagnosis	Severity	Review	Vaccination	Smoking	Treatment	Rehabilitation	Outcomes
Practice 01	85	78	82	85	75	89	87	85
Practice 02	81	79	85	86	83	81	90	77
Practice 03	68	63	65	72	63	61	63	54
Practice 04	90	88	93	78	83	83	92	90
Practice 05	80	73	89	91	74	77	88	84
Practice 06	53	55	71	52	59	59	68	58
Practice 07	55	66	53	73	61	66	57	68
Practice 08	71	76	65	59	69	75	64	68
Practice 09	56	43	58	44	37	46	50	45
Practice 10	87	67	68	69	78	73	69	88
Practice 11	66	84	86	87	81	70	78	68
Practice 12	84	90	72	83	88	76	88	78

Practice 09 averages 47% against 72% across the area.

0

2 or more flare-ups in 90 days and no review in 12 months

0

Breathless, never referred to rehab

0

Discharged, no review within 6 weeks

323 patients this month, once 44 who meet more than one rule are counted onceBehind the scenesPlain rules find who is due something. A risk score, trained on past admissions in the area, puts first the patients most likely to be admitted in the next 90 days. Every row shows its reasons.AssumptionThe score here is invented. A real one is tested on past data before anyone relies on it.What it achievesClinicians see who to call first and why, instead of a dashboard to interpret.

Patient 992580, Practice 05Breathless, no rehab2 flare-ups, no review
Patient 112578, Practice 122 flare-ups, no reviewA&E twice
Patient 394985, Practice 09No flu vaccineA&E twice
Patient 782968, Practice 093 steroid coursesDischarged 9 days ago
Patient 542486, Practice 09Discharged 9 days ago
Patient 138677, Practice 03A&E twice2 flare-ups, no review

To Practice 09, drafted for the clinical lead

43 of your patients with COPD need something this month.

Ordered by who to see first, with the reason on every row
Links open each patient in your clinical system
Your practice is behind the area on most measures

Behind the scenesThe worklist goes through the channel the practice already uses. A clinician decides who is contacted and how; the system only orders the list and shows the reasons.Could go furtherBook reviews straight into practice clinics, and track who was seen at the next run.

Four weeks later

61% of the area's list reviewed (modelled). The next run removes them and adds the newly due.

0

patients on the register

0

measures, current every night

0

patients on this month's worklist

0%

reviewed within four weeks (modelled)

Board paper, drafted from the same numbersBehind the scenesA language model writes the narrative from the measures and the worklist only, and every figure in it is checked against the tables before it is shown. The lead edits and approves it.AssumptionThe paper replaces about 30 hours of analyst work a quarter with about 2 hours of review. Modelled.

Across 40 measures, COPD care in North Vale averages 72%. Practice 09 is furthest behind, at 47%, across most groups. This month 323 patients were identified for action.

A disease register, rebuilt and put to work.