Artificial intelligence (AI) is a technology that has, essentially, automated learning. With enough data, it can identify disease and, now, according to new research, it can even predict which patients are likely to require additional treatment in some cases.
Investigators collected 12,242 corneal OCT images from 3,162 corneal surgery subjects and included 3,318 measurements collected at the baseline visit of each patient. A total of 333 eyes had some form of corneal surgery, i.e., penetrating keratoplasty, lamellar keratoplasty, deep anterior keratoplasty, Descemet’s stripping automated endothelial keratoplasty or Descemet’s membrane endothelial keratoplasty. Using the information culled from those images, the AI found a way to identify patients who may be at higher risk for future keratoplasty.
The automated learning system under investigation can make this determination using comprehensive corneal shape, thickness and elevation parameters, the researchers say. The AI system found clustering eyes into five different groups based on these corneal features identified five different percentages ranging from a 2.2% change of the need for future keratoplasty to 31.0%.
Future research that uses independent datasets will be necessary to validate this proposed system.
|Yousefiab S, Takahashic H, Hayashi T, et al. Predicting the likelihood of need for future keratoplasty intervention using artificial intelligence. Ocular Surf. February 28, 2020. [Epub ahead of print].