The scarcity of human ocular samples with short postmortem intervals (PMIs) is a significant issue in ophthalmic research and drug discovery. A contributing factor is that eye banks must manually match donor data to prospective research project criteria, which is time-consuming, inefficient, and error-prone. We have previously reported on the successful use of a semi-automated matching system, ReSync. The barrier to full autonomy is that donor medical data is often provided as unstructured data in free text fields, which prevents interoperability with matching databases. Herein, we report on a small retrospective study, in which artificial intelligence (AI) is incorporated into ReSync (ReSyncAI) to test AI's ability to structure donor data for subsequent matching. From a set of historical cases, medical data was securely sent to a large language model with natural language processing. After structuring and standardizing, data was returned to ReSync for analysis and match testing. A 94.2% success rate in medical terminology keyword extraction in concert with correcting and standardizing medical data was achieved. Structured data was fully interoperable with ReSync. In a subset of cases, ReSyncAI properly matched donors to the standardized term of "age-related macular degeneration" from donor data, including instances of abbreviations, misspellings, and incomplete designations.
Keywords: AI; AMD; Artificial intelligence; Biobanking; Eye banking; NLP; Ophthalmic research; Ophthalmology; Postmortem; Retina.
© 2025. The Author(s), under exclusive license to Springer Nature Switzerland AG.