About DOTS
The DemDx Ophthalmology Triage System suggests a triage priority and the conditions to consider for adults attending an ophthalmic emergency department. It was developed at Moorfields Eye Hospital, validated in the UK and Brazil, and published in peer-reviewed journals.
Designed to support the triage nurse, not to replace themThe problem
Ophthalmic emergency departments see many patients who could be cared for elsewhere, while true emergencies must never be missed.
Sources: Brandao-de-Resende et al., eClinicalMedicine 2023 [1]; Mass Eye and Ear chart review 2015–16 (Digital Journal of Ophthalmology).
How DOTS works
The nurse enters the presentation and makes their own decision first. DOTS then suggests a priority and the conditions to consider, and suggests questions that could change it.
1 · Nurse enters
- Age group
- Affected eye and duration
- Red flags
- Signs and symptoms
- Relevant history
No names or hospital numbers. A date of birth is turned into an age band on your device and is not sent.
2 · DOTS
A machine-learning model (gradient-boosted trees) trained on 11,315 Moorfields A&E attendances, with a second model ranking the likely conditions.
3 · Suggestion
Plus more urgent conditions to consider or rule out, less urgent conditions to consider, and up to 5 questions that could change the result [3].
What it covers
93 condition groups, each prioritised by three consultant ophthalmologists. Any red flag means emergency.
EMERGENCY36 groups
- Acute angle closure
- Retinal detachment or tear
- Retinal artery occlusion
- Giant cell arteritis
- Open globe injury
- Keratitis
URGENT20 groups
- Posterior vitreous detachment
- Retinal vein occlusion
- Proliferative diabetic retinopathy
- Neovascular membrane
- Proptosis
- Dislocated IOL
ROUTINE37 groups
- Dry eyes
- Blepharitis
- Chalazion
- Infective conjunctivitis
- Allergic conjunctivitis
- Subconjunctival haemorrhage
Red flags
Published evidence
Development and validation study [1]: DOTS on its own compared with triage nurses on their own, on Moorfields cases held out from training.
Validated in two countries
11,315 attendances to develop DOTS, 1,269 for internal validation at Moorfields (London) and 761 for external validation at UFMG (Belo Horizonte, Brazil) [1].
Usable and trusted by clinicians
In a mixed-methods study with 20 clinicians, over 90% found it easy to use, 85% found it safe and 95% said they would use it in their clinical workflow [2].
Emergency class. Moorfields temporal validation; differences from nurses significant at p < 0.05. Source: eClinicalMedicine 2023, Table 3 [1].
Clinical evaluation
DOTS is being evaluated with triage nurses in routine care.
Evaluation of DOTS in the Moorfields Eye Hospital A&E
- Registration
- ISRCTN25862260
- Design
- Single-centre pre-post evaluation: triage with DOTS compared with matched cases from the previous year
- Funding
- NIHR Artificial Intelligence in Health and Care Award
- Status
- Results are preliminary and not yet peer reviewed. They will be added here once published.
Publications
- A machine learning system to optimise triage in an adult ophthalmic emergency department: a model development and validation studyBrandao-de-Resende C, Melo M, Lee E, Jindal A, Neo YN, Sanghi P, Freitas JR, Castro PVIP, Rosa VOM, Valentim GFS, Higino MLO, Hay GR, Keane PA, Vasconcelos-Santos DV, Day ACeClinicalMedicine 2023;66:102331 · doi:10.1016/j.eclinm.2023.102331
- Usability of an artificially intelligence-powered triage platform for adult ophthalmic emergencies: a mixed methods studyJindal A, Sumodhee D, Brandao-de-Resende C, Melo M, Neo YN, Lee E, Day ACScientific Reports 2023;13:22490 · doi:10.1038/s41598-023-49213-y
- Enhancing ophthalmic triage: identification of new clinical features to support healthcare professionals in triageJindal A, Brandao-de-Resende C, Neo YN, Melo M, Day ACEye 2024;38(13):2536–2544 · doi:10.1038/s41433-024-03070-9
DOTS is a clinical decision support tool for triage nurses in adult ophthalmic emergency departments. It suggests a priority and conditions to consider; it does not diagnose and does not replace clinical judgement. Do not enter names, hospital numbers or other patient identifiers.