Predicting bilingual aphasia treatment outcomes using digital twins: a double-blind randomized controlled trial

Area of Science:

• Neurolinguistics
• Computational linguistics
• Speech-language pathology

Background:

• Bilingual aphasia rehabilitation presents challenges in selecting the optimal language for therapy to enhance recovery across both languages.
• The BiLex computational model offers a potential solution for predicting the best language for intervention.

Purpose of the Study:

• To evaluate the efficacy of the BiLex computational model in predicting the optimal language for aphasia therapy in Spanish-English bilinguals.
• To compare treatment outcomes when therapy targets the BiLex-recommended language versus the non-recommended language.

Main Methods:

• A double-blind randomized controlled trial involving 48 Spanish-English bilinguals with chronic aphasia.
• Participants received 40 hours of semantic feature-based treatment in either the BiLex-recommended language or the opposite language.
• Outcomes measured included treated-language naming and cross-language generalization, with subgroup analyses based on performance.

Main Results:

• Both groups showed similar improvements in treated-language naming.
• The group receiving therapy in the model-opposite language demonstrated significantly greater cross-language generalization, albeit with higher variability.
• The BiLex-assigned group showed an advantage in most performance subgroups, aligning with model predictions.

Conclusions:

• The BiLex model's predictions were largely consistent with observed treatment outcomes.
• While targeting the recommended language aids naming, targeting the opposite language may promote broader cross-language generalization in bilingual aphasia.
• The BiLex model shows promise for guiding personalized language therapy choices in bilingual aphasia rehabilitation.