Discharge information is a vital aspect of physician-patient communication, and a patient’s poor understanding of their instructions can adversely influence outcomes once they leave the hospital. Upon discharge, patients are given pertinent written information about managing their diagnosis, including medications and follow-up appointments, often through English-language materials at high literacy levels, regardless of preferred language. In some cases, generic discharge information handouts are generated for common languages and diagnoses, but cannot be tailored specifically to the patient. A study from the University of California, San Francisco found that limited English proficiency (LEP) patients demonstrated worse understanding of their home medications based on discharge instructions than their English-speaking counterparts. This illustrates a gaping hole in healthcare communication: once a patient and family has left the hospital, they need clear, specific instructions for their personal care plan in their preferred language.

The current options for discharge information translation all have relatively significant shortcomings. The first is a direct translation algorithm, similar to Google Translate. Though the translation is instantaneous, the lack of linguistic understanding in this system could easily lead to dangerous information mistranslations. The second, mentioned above, is premade discharge handouts for common diagnoses in a few popular languages. This approach does not allow for customization, therefore limiting the physician’s ability to illustrate a specific home care plan. Another option is remote patient document translation by a certified interpreter. Though this approach creates understandable documents for a LEP patient in the language of their choice, it is expensive and time consuming, which can pass inequitable costs along to patients and delay discharge and bed turnover for the hospital. Many hospitals opt for an in-person or telephone interpreter for the physician to explain the home care plan, which is sent home with the family written in English. This experience can be overwhelming for families because they cannot refer back to an understandable document once they arrive home, and must rely on phone translation services or the help of others with higher English proficiency.

An artificial intelligence-based immediate translation platform for patient discharge information is a solution to this problem while addressing the shortcomings of its alternatives. The proposed platform will utilize human translators during its development to improve document translations made by the computer, thus ‘training’ the AI-based application to translate in an accurate, understandable manner. Once the translations are consistently correct, the application will instantly create patient-specific discharge information for the physician’s tailored care plans without consulting a translator. This will allow for easier, faster discharge because LEP patients will be able to read and review their instructions with an interpreter at the hospital in their preferred language, before leaving with written instructions the family can understand without assistance. Because the application runs without human interference once it has been ‘trained’ in a language, it will be cost effective and can provide services for as many languages as needed. With this platform, physicians can provide patient-specific discharge information materials in an effective, understandable manner to all families, regardless of preferred language.



Author: Sarah Woodworth

Status: Project Concept