Bio

I was born in Hubei, China, and moved to Zhuhai to obtain my BA degree in translation at School of Translation Studies, Jinan University. That was also the period when I interned in the comprehensive office of Zhuhai People’s Hospital and participated in several public health studies. After that, I pursued a master’s in theoretical and applied linguistics with a specialization in computational linguistics at Universitat Pompeu Fabra (Barcelona, Spain). During the program, I built and evaluated a neural network for metaphor detection under the supervision of Dr. Núria Bel. Later, I shifted my focus to applied natural language processing in cognitive disorders. That motivated me to join GraC at the Universitat Pompeu Fabra as a doctoral candidate, where I am studying how to model language and brain changes computationally in cognitive disorders. This study started with a focus on Alzheimer’s disease but has since expanded to other types of dementia (progressive primary aphasia for example) and schizophrenia. My doctoral study is supervised by Dr. Wolfram Hinzen and Dr. Núria Bel, and granted by China Scholarship Council (CSC).

Research

For various cognitive disorders, language serves as a clinical signal and potential mechanism, providing a cheap, convenient, and sensitive alternative to canonical cognitive tests. Automatic language analysis empowered by the impressive advances in natural language processing facilitates this utilization of language in detecting the clues of a wide range of diseases, including neurodegenerative, neurodevelopmental, and neuropsychiatric disorders. Despite these, the black-box nature of language models impedes the progress of clinical application owing to the limitations in interpretability and trustworthiness. Analyzing language model outputs in a more interpretable manner and explaining the outputs with neurological evidence could potentially remedy this problem. Additionally, although requiring more time, money, and human labor compared to language data, neuroimaging is known as sensitive to cognitive mechanisms while still cheaper and non-invasive compared to traditional biomarker collection. Language data and brain imaging could yield different but complementary results and thus bring an omnibus look at the underlying clues for pathological changes.

My interest generally falls into how language and brain function change in cognitively atypical population using automatized approaches, and further into how changes in these two domains connect to each other. My current research majorly investigates two disease spectra, dementia and schizophrenia. These studies theoretically and practically endorse the deployment of language data, as well as brain imaging, in automatic warning for early onset and disease relapse.

Key words

Natural language processing; cognitive disorders; computational MRI analysis; machine learning; dementia; schizophrenia