Chun Liu Email & Phone Number

Cambridge, Ma, Us

Woonsocket, Ri, Us

• Defining, implementing, and refining portfolio-level impact evaluation strategy to enhance the rigor, consistency, and efficiency of existing evaluation practices• Executing robust analyses that deliver actionable insights to business stakeholders on program performance, patient behavior, and profitability• Measuring the effects and outcomes of specific programs or interventions implemented by business teams, understanding cause-and-effect relationships and providing evidence-based insights on the success or failure of projects• Conducting experimental design for A/B testing in marketing campaign; assessing the feasibility of the tests and proactively address any technical challenges that may hinder the experiment

Thousand Oaks, Ca, Us

Thousand Oaks, Ca, Us

• Worked collaboratively in a cross-functional team to prototype and industrialize enterprise product surveillance tool, which was selected as the first-place winners (4 out of 383 initiatives)• Developed strategy, analytical tools (NLP, Graph Network), and user interface for post-market surveillance of adverse events and product complaints, achieving the goal of holistic surveillance and patient safety improvement; Built adverse event profiles of drugs for similar combination products comparison with FDA Adverse Event Report System (FAERS) data• Built data pipeline for analyzing patients' feedback with natural language processing (NLP) and neural network/deep learning (BERT), saving over 50% of time for subject-matter experts on text data reading. • Analyzed thousands of text data with topic modeling (CorEx, NMF model) to classify documents into different topics, reducing analysts efforts on data reading• Applied machine leaning (XGBoost / random forest model) to predict patients' feedback

San Francisco, Ca, Us

• Built a Dash web app to visualize each individual patient’s key epigenomic data and kidney cancer diagnosis, acting as a second method to double check the clinical diagnosis result.• Tested several different combinations of dimension reduction methods (PCA, LDA, Kruskal- Wallistest) and classifiers (Logistic regression, LDA, QDA, Random forest) in Python; combinations of Kruskal-Wallis test and random forest classifier have highest accuracy in predicting tumor stage• Analyzed 208 patients with over 20,000 genes methylation data and identified 20 genes as keyfeatures for kidney tumor stage prediction; accuracy of prediction is over 70%

Philadelphia, Pa, Us

• Proposed a minimal kinetic model with differential equations to simulate cell polarization, which isimportant in directed cell migration and cancer metastasis• Built nonlinear multivariate regression models using MATLAB for interfacial enzyme kinetics datafitting; extracted key enyzme binding and catalytic features for predictive modeling• Quantitatively analyzed thousands of time-lapse fluorescence images and extracted key enzymefeatures; identified how fast the lipids are phosphorylated or dephosphorylated by enzyme

Philadelphia, Pa, Us

• Conveyed sophisticated chemistry ideas and explain detailed experimental procedures clearly.• Demoed experiments to show students how to conduct chemical reactions and analysis safely.

Taipei, Northern Taiwan, Tw

Using steady-state and ultrafast spectroscopy method to characterize the photophyiscal properties of different compounds, which have applications to developing new materials and unveiling the mechanism of enzyme function.Ultrafast spectroscopy:• Time-correlated single photon counting (TCSPC)• Fluorescence upconversion

Doctor Of Philosophy - Phd, Physical Chemistry / Biophysics

Master'S Degree, Physical Chemistry

Bachelor'S Degree, Biochemistry