Feasibility of an artificial intelligence phone call for post-operative week 1 assessment following cataract surgery in a diverse population in Canada

Theme: Cataract surgery*
What: Cataract surgery
Part of: Cataract III: Innovations and research in cataract surgery / Cataracte III: 3. Innovations et recherches en chirurgie de la cataracte
When: 6/2/2024, 11:15 AM - 12:45 PM
Where: Room | Salle 801

Abstract

Purpose:

Dora is an artificial intelligence (AI) telephone call used for clinical-grade conversations with patients, which is used in the UK in lieu of the post-operative week 4 follow-up after routine cataract surgery. We investigate the safety, acceptability and efficacy of an automated Dora call at post-operative week 1 (POW1) for patients undergoing surgery in Canada. 

Study Design: Prospective single arm study

 Methods:

Any patient having routine surgery was eligible for inclusion if they, or a nominated relative, could have a conversation in English. Patients were recruited from a high-volume surgical centre in Ontario, Canada between July-September 2023. Dora called patients the evening before their POW1 face-to-face assessment. Dora identified any clinical concern based on symptom assessment over the phone. We compared outcomes from the Dora call to the clinician assessment the following day. Dora asked every patient for a Net Promoter Score (NPS) of the likelihood they would recommend the automated service. Telehealth Usability Questionnaire (TUQ) was used to measure the quality of Dora’s interface on a scale of 1-5 with 1 being strongly disagree and 5 being strongly agree.

 Results:

A total of 198 patients were recruited. 132 (67%) successfully completed the authentication checks, and among them, 124 (94%) finished their call with Dora. The predominant reasons for not completing identity checks were linguistic differences in confirming identity, and concerns about spam calls. Dora passed 81(69%) patients, indicating they had no clinical concerns. At face-to-face POW1, 7 (9%) reported symptoms like dry eye and visual concerns, but no patient had a change in management that would have led to a serious risk of harm. Conversely, of the patients Dora did identify clinical concerns, 6 (14%) had no clinical issues identified at POW1. Patients gave Dora a mean NPS of 8/10. As for TUQ, patients stated that “It was simple to use Dora.” (mean 3.71, median 4), “It was easy to learn to use Dora” (mean 3.73, median 4), “Dora is simple and easy to understand.” (mean 3.68, median 4). While patients stated that “visits provided over Dora are the same as in-person visits.” (mean 2.76, median 3).

 Conclusion:

At POW1 Dora can safely identify patients who are recovering as expected following routine cataract surgery which may help reduce unnecessary appointments. This increases clinical capacity and can help the healthcare labor shortage. It is also convenient for patients, who can save travelling and receive calls at times that are convenient to them.

Presenter(s)

Presenting Author: Amin Hatamnejad

Additional Author(s):

Sohel Somani, Department of Ophthalmology and Vision Sciences, University of Toronto, Toronto, Ontario, Canada

Eric Tam, Department of Ophthalmology and Vision Sciences, University of Toronto, Toronto, Ontario, Canada

Aisling Higham, Ufonia Limited, Oxford, UK

Ernest Lim, Ufonia Limited, Oxford, UK, Department of Computer Science, University of York, York, UK

Sarah Khavandi, Ufonia Limited, Oxford, UK

Nick de Pennington, Ufonia Limited, Oxford, UK

Hannah Chiu, Department of Ophthalmology and Vision Sciences, University of Toronto, Toronto, Ontario, Canada, Michael G. DeGroote School of Medicine, McMaster University, Hamilton, Ontario, Canada, William Osler Health System, Brampton, Ontario, Canada

---