We strive to design and build natural language processing and machine learning frameworks that are portable across medical and public health problems. We put particular attention to ensuring that technological innovations in data science, NLP, machine learning and artificial intelligence comply to the specific needs of the medical domain. These needs include, but are not limited to, interpretability, simplicity, reliability and timeliness. The following are some of our current specific focus areas.
Despite the ongoing opioid crisis, current strategies for close-to-real-time monitoring and characterizing drug-related health problems are laggy. We are building methods and tools that leverage data from sources such as social media and electronic health records to generate statistics in close to real time, predict potential future problems, and empower domain experts (e.g., toxicologists and public health professionals) who are fighting the crisis.
Portable Natural Language Processing
We are continuously developing BioNLP software that are portable across medical domain problems, and don’t live and die with narrow-scope studies. We implement methods for text classification, information detection and extraction, text representation and normalization, topic analyses and visualization.
Social Media Mining
Social networks contain abundant information on every topic. Adoption of social media is now at an all-time high, and the number of people using social media continues to grow. We are innovating strategies for curating and utilizing social media data for medical and public health tasks. We are also continuously exploring new uses for social media data.
We are striving to answer the question: What is patient-centeredness from the perspective of NLP? Our patient-centered NLP research focuses on enabling NLP-driven evidence-based medicine practice.
NLP for Cancer
We are developing social media based NLP pipelines to help fight cancer. We are specifically interested in studying patient reported outcomes (PROs), so that we can better understand the outcomes that matter to cancer patients.