Shreya Parchure, an Indian American doctoral student, is pioneering an AI tool to personalize speech therapy for stroke survivors, enhancing recovery prospects for those affected by post-stroke aphasia.
Shreya Parchure, an Indian American researcher and doctoral student at the University of Pennsylvania, is making significant strides in the field of speech therapy for stroke survivors. Her innovative approach utilizes artificial intelligence (AI) to personalize treatment for individuals suffering from post-stroke aphasia, a condition that impairs the ability to understand or produce speech and affects approximately one-third of stroke survivors.
Growing up across two continents, Parchure developed a deep appreciation for the importance of language in enhancing quality of life. Her clinical rotations in a neurocritical care unit further solidified her commitment to advancing research and care for patients with aphasia. During her interactions with patients, she witnessed firsthand the profound impact that speech therapy can have on recovery. One patient, who initially struggled to speak, gradually regained her ability to communicate through dedicated therapy. “She was overjoyed,” Parchure recalls, highlighting how progress in speech therapy can instill hope in patients.
Traditional speech therapies for post-stroke aphasia often follow standardized protocols. However, Parchure and her team at the Laboratory for Cognition and Neural Stimulation (LCNS) are exploring the potential of “explainable AI.” This set of machine learning methods focuses on providing clear rationales behind AI-generated results, enabling healthcare providers to interpret and trust the recommendations made by the technology.
While some AI models have utilized neuroimaging and the duration since a stroke to assess aphasia severity, Parchure’s research expands on these methods by incorporating how language is formed and processed in the brain. “Explainable AI can integrate clinically available data—such as age, education, or the size of a stroke—with the linguistic difficulty of words,” she explains. This multifaceted approach allows the AI model to predict recovery timelines and suggest tailored treatments based on individual patient circumstances.
“When we have an AI making a prediction, we really want to know why,” Parchure emphasizes. She has leveraged speech samples from patients with post-stroke aphasia to train an explainable AI algorithm, testing its ability to account for various language tasks and make recovery predictions based on a diverse array of clinically relevant information. The tool also considers personal attributes, such as the size of the stroke and the level of social support available to the patient.
“Incorporating language into the fold adds a new layer of considering human and brain complexity,” Parchure notes. The explainable AI tool can predict speech performance on a word-by-word basis, which can help clinicians identify the underlying factors affecting a patient’s speech abilities. This granularity informs more nuanced treatment plans and recovery predictions.
“It’ll help tailor speech therapy for where exactly people are having trouble,” Parchure states. “We can really meet patients where they are in a more personalized manner.” To facilitate this, Parchure and her colleagues have developed an AI-powered application for use in both clinical and research settings. A particularly innovative aspect of this research is the creation of a “digital twin” for each patient, which serves as a predictive tool for language recovery.
The simulated “twin” allows for a comparative analysis of how a patient may respond to different treatments, enhancing the efficiency of clinical trials by enabling researchers to compare projected outcomes with actual recovery results. “The goal of my MD-PhD training has been to translate advances in research in a way that will benefit patients,” Parchure explains. Her work has already garnered recognition, including the Best Poster award in Translational Research at the 2025 PSOM Student Research Symposium.
Looking ahead, Parchure envisions a future where AI plays a crucial role in personalizing speech therapy, ultimately helping stroke survivors with aphasia reconnect with the joy of language. “Over the next decade, I believe we will see significant advancements in this area,” she concludes.
According to Penn Today, Parchure’s research represents a promising development in the intersection of technology and healthcare, offering hope to countless individuals affected by stroke.