Scherer Rafael

Bhatnagar Anshul, Scherer Rafael, Samico Gustavo A, Muralidhar Rohit, Gutkind Naomi E, Palazoni Vitoria et al.

A medical LLM accurately identified glaucoma medication non-adherence, visit non-adherence, and family history from clinical notes, enabling large-scale research into these crucial risk factors.

R da Costa Douglas, Scherer Rafael, Swaminathan Swarup S, Tseng Henry, A Medeiros Felipe

AI identified five most vulnerable visual field locations for glaucoma progression. These High-5 points showed significantly faster decline and better progressor discrimination, offering a sensitive functional endpoint for trials and personalized monitoring.

da Costa Douglas R, Unyi Dániel, Scherer Rafael, Muralidhar Rohit, Ribeiro Monteiro Mario Luiz, Medeiros Felipe A

This study found a graph-based AI model (GIN) significantly outperforms traditional methods and other AI in detecting glaucomatous visual field defects, offering a more accurate and interpretable diagnostic tool.

Azizi Ali, da Costa Douglas R, Scherer Rafael, Malek Davina A, Samico Gustavo A, Medeiros Felipe A

This study found a deep learning model accurately estimated RNFL thickness from fundus photos, confirming age-related thinning and thinner RNFL in glaucoma, enabling scalable population-level screening.

Liu James C, Jammal Alessandro A, Scherer Rafael, da Costa Douglas R, Kass Michael, Gordon Mae et al.

Deep learning-predicted RNFL thickness from optic disc photos identified ocular hypertension patients at higher glaucoma risk and progression, offering a non-OCT risk assessment tool.