Focus areas: statistical machine learning, deep learning, natural language processing, optimization, scientific machine learning (SciML), pharmacokinetics, precision medicine

Focus areas: machine/deep learning, natural language processing, clinical informatics, time-series predictive analysis, health equity, and real-world evidence generation.AI solutions portfolio: Medical Billing Code Categorizer using Representation Learning (Language Models), Automated Payer Policy Change Alert, Medical Error Predictor from Pharmacy Claims, Total Cost of Care Prediction, and Identification of Determinants of High Cancer Care Cost.- Develop and apply state-of-the-art machine learning and NLP algorithms to accelerate insight generation, improve operational efficiency, and support innovative pharmaceutical solutions and clinical decision-making processes.- Develop AI solutions for community oncology care practices (the US Oncology Network), advancing value-based oncology care models and promoting improved patient experiences and health outcomes.

- Led a cross-functional effort to develop applied machine learning pipelines, advanced statistical machine learning models, and state-of-the-art deep learning models. - Tackled challenges in real-world data mining, equitable clinical trial matching, automated computational phenotyping, and patient journey characterization. - Developed AI solutions to generate real-world evidence and actionable insights for HEOR studies and clinical trials to advance precision medicine and real-world care.

(Center for Artificial Intelligence, Research and Evaluation (CARE), IBM Watson Health)- Cross-functionally collaborated with IBM researchers and conducted COVID-19 related epidemiological predictive modeling and health disparity studies.- Led a group of IBM researchers to define and build cohorts from the claims database for the CMS AI Challenge tasks and supported the downstream modeling efforts.- Led and conducted claims data analysis for a range of healthcare disparity research projects leveraging the IBM Real-World Databases (MarketScan, Explorys) and produced numerous peer-reviewed conference abstracts.- Surveyed and synthesized the current hospital evaluation models in collaboration with a cross-functional team, resulting in one high-impact peer-reviewed journal article.- Developed a bioinformatics pipeline for the genomics evaluation team and collaborated with the clients on research projects, as demonstrated by two conference abstracts and one peer-reviewed journal paper submission.

- Developed bioinformatics tools and pipelines and comparatively analyzed genomic sequences of donor and recipient pairs in association with the stem cell transplant outcomes.- Developed efficient statistical inference algorithms to analyze the classical HLA gene full sequences at scale to assess the necessity of higher resolution HLA matching in donor selection.- Designed and developed a machine-learning model to statistically predict and infer the familial relationships between individuals using demographic information.

- Developed a model to mine the familial relationship for better donor selection.(Click 'View' after opening the link below to see the full presentation.)

- Designed and developed a machine learning-based R package for microbiome-wide association study (MWAS).- Collaborated with biologists and conducted microbiome composition analysis on human gut, canine and turkey metagenomics.- Mentored two undergraduate students to conduct microbiome research projects through HHMI Undergraduate Mentorship program.

- Invited talk for a bioinformatics short course at the Iowa Institute of Human GeneticsTwo topics: Introduction to Computation Metagenomics and Understanding Microbiome using QIIME- Introduced the high-level metagenomics study to non-biology background audiences - Instructed the computational and statistical analysis process in microbiome composition studies along with the step-by-step guide to the current bioinformatics tools to biologists and graduate students

- Developed a group-based classification model for accurate cancer cell recognition from histology images.- Improved the classification accuracy to the clinically acceptable level, by 8% on average compared to the traditional machine learning models.- Investigated and assessed density estimation methods on medical image analysis for the classification purposes.

- Designed and implemented a muscle electrical signal (EMG) processing and pattern recognition system for noninvasive prosthetic limb control- Developed fast and accurate motion recognition algorithms by integrating swarm intelligence and machine learning techniques and improved the recognition speed and accuracy above the clinically acceptable level - Trained and guided new graduate students joining the research group- Instructed an undergraduate biomedical signal processing lab course as a teaching assistant

- Coordinated new product development projects by translating product spec requirements from the headquarter into technical details to the local product design and development team - Researched and reported new product trends and markets and recommended the target regions and customers