Deep Learning-Driven Glaucoma Medication Bottle Recognition: A Multilingual Clinical Validation Study in Patients...

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Mobile Health

Background:

• Glaucoma is a leading cause of vision loss, necessitating consistent topical medication use.
• Patients with glaucoma often experience impaired vision, complicating medication management and adherence.
• Accurate identification of glaucoma medications is crucial for effective treatment and preventing disease progression.

Purpose of the Study:

• To clinically validate a convolutional neural network (CNN)-based Android smartphone application for identifying topical glaucoma medications.
• To assess the app's effectiveness in patients with glaucoma and impaired vision, including non-English speakers.
• To evaluate the impact of the app on medication identification accuracy and time.

Main Methods:

• A nonrandomized prospective crossover study involving 41 patients (20 non-English-speaking, 21 English-speaking) with visual acuity of 20/70 or worse.
• Patients identified 6 topical glaucoma medications in a randomized order, first without the app, then with the app.
• Ease-of-use surveys were administered before and after app usage to collect qualitative data.

Main Results:

• The CNN-based app achieved high precision (98.8%) and recall (97.2%).
• Medication identification accuracy significantly improved from 27.6% without the app to 99.2% with the app (P < 0.001).
• Improvement in accuracy was consistent across English- and non-English-speaking groups, with no significant change in identification time.

Conclusions:

• The CNN-based mobile app substantially enhances topical glaucoma medication identification accuracy in patients with glaucomatous vision loss.
• The app is a practical tool that can improve medication adherence for both English- and non-English-speaking populations.
• This technology offers a valuable adjunct for daily medication management in patients with glaucoma and low visual acuity.