ESCRS - PPE1.02 - Using An Artificial Intelligence Telephone Call To Automate Cataract Pre-Assessment Appointments

Using An Artificial Intelligence Telephone Call To Automate Cataract Pre-Assessment Appointments

Published 2024 - 42nd Congress of the ESCRS

Reference: PPE1.02 | Type: ESONT Abstract | DOI: 10.82333/g44j-7y55

Authors: Aisling Higham* 1 , Ernest Lim 2 , Daiserie Vinluan 3 , Kanmin Xue 3 , Kikkeri Arun 3 , Louisa Stacey 4 , Vera He 2 , Rory McKinnon 2 , Sarah Khavandi 2 , James Godwin 3 , Rebecca Turner 3

1Ophthalmology,Oxford Eye Hospital,Oxford,United Kingdom;Ufonia,Oxford,United Kingdom, 2Ufonia,Oxford,United Kingdom, 3Ophthalmology,Oxford Eye Hospital,Oxford,United Kingdom, 4Ophthalmology,Ufonia,Oxford,United Kingdom

Purpose

Pre-operative assessment is often a bottleneck for surgical waiting list throughput. We evaluate the real-world effectiveness of an artificial intelligence powered telephone conversation agent (Dora R3.9, Ufonia Limited, Oxford, UK) for performing automated cataract surgery pre-operative assessment history taking prior to proceeding to an in-person cataract assessment.

Setting

All patients referred for routine cataract surgery at a large teaching hospital had a call from Dora before their in-person one-stop (biometry, clinical assessment, consenting and pre-assessment within the same appointment) cataract assessment . 

Methods

Dora assesses 18 domains in order to determine case complexity (for example current medications and ability to lie flat) and undertakes a shared decision making discussion. A Net-Promoter-Score (NPS) of the likelihood of recommending Dora to other patients was captured (score 1-10). Effectiveness was assessed through time in motion studies.  Accuracy of clinical assessment was compared to an expert human listening to the calls as well as ground-truth from face-to-face assessments.

Results

Between March 2023- March 2024, 1504 patients completed pre-assessment telephone calls with Dora. Patients had a mean age of 76 years (SD=10) and gave Dora a median NPS of 8/10 (IQR= 7-10). When comparing Dora to the expert call reviewer, the mean agreement was 93% across 20,950 questions. Comparing Dora decisions to face-to-face cataract assessment showed agreement of 86% across 654 values collected over a 3-week audit. Time in motion studies suggested 12 minutes time saving per consultation. 5.4% of patients were unsure or did not want cataract surgery, thus unnecessary in-person appointments were avoided. 

Conclusions

In this population, automated cataract pre-operative assessment reliably collects key clinical information from patients. This information can identify patients who are unsure or do not want surgery prior to attending the cataract clinic. The information also shortens appointments, and can free clinician time within the ophthalmology department.