Selected Publications

This paper considers the problem of undertaking fully Bayesian inference for both the parameters and structure of a vector autoregressive model on the basis of time course data in the ‘p >> n scenarioa’. The autoregressive matrix is assumed to be sparse, but of unknown structure. The resulting algorithm for dynamic Bayesian network inference is shown to be highly effective, and is applied to the problem of dynamic network inference from time course microarray data using a dataset concerned with the transient response of budding yeast to telomere damage.
In Journal of Biometrics & Biostatistics, 2011
Source Document

Recent Publications

More Publications

BCL2 Expression Identifies a Population with Unmet Medical Need in Previously Untreated (1L) Patients with DLBCL

Source Document

Prognostic impact of BCL2 and MYC expression and translocation in untreated DLBCL: results from the phase III Goya study

Source Document

Safety And Efficacy Of Obinutuzumab With CHOP Or Bendamustine In Previously Untreated Follicular Lymphoma

Source Document

Recent & Upcoming Talks

Predict survival for cancer patients using real world data: simple model or advanced machine learning?
4 Jun 2018 3:00 PM
PSI Conference 2018
Slides

Recent Posts

More Posts

4 min read Cancer

“Dying to Survive” is a 2018 Chinese comedy-drama film based on a real-life story of Yong Lu, a Leukemia patient who imported generic cancer drugs from India to China to supply fellow patients suffering from Chronic Myeloid Leukemia (CML) who could not afford the cost of the domestically available medication. When Yong was arrested in 2013, the charge against Yong was eventually dropped following a public outcry by over 1,000 patients whose lives had been extended by his efforts.

CONTINUE READING

2 min read Machine Learning

This week I presented “Introduction to Machine Learning with XGBoost” to my stats colleagues. Why do I want to learn XGBoost and introduce to other statisticians? XGBoost is one of the most popular machine learning algorithms these days. It has won the RAAD (Roche Advanced Analytics Data) challenge among 81 teams across 19 Roche sites in early 2018! Among the 81 teams, different methods (including logistic regression) have been tried, the top 3 teams all used XGBoost!

CONTINUE READING

2 min read R

I just returned from PSI conference 2018. I got a chance to give an oral present at session “Step into the Real World”. I shared my experience from taking part in Roche Advanced Analytics Data (RAAD) Challenge. The take-home message is “with available wealthy real-world data and the power of machine learning for such data, statisticians would be benefit to know machine learning methods”. You can read my slides “Predict survival for cancer patients using real world data: simple model or advanced machine learning?

CONTINUE READING

Professional Experience

Statistical Scientist, Product Development, Roche, UK (2012 – Present)

As a statistician working in drug development, I work at study level, project level and franchise level. I now work on clinical trials of three breast cancer drugs. I have been Hematology Franchise Biomarker Biostats Lead during 2013 - 2017 working closely with biomarker scientists to develop the biomarker strategy and leading some statistical analyses. I have also led project level safety reporting from 2012 to 2014 for one molecular, including first RMP (Risk Management Plan) and SCS (Summary of Clinical Safety) for regulatory filing.

Experimental Statistician, Product Development, Roche, UK (2011)

Experimental Statistician’s work focused on PHC (Personalized Healthcare) aspect, but also with opportunity to work on other areas. I have conducted Meta-analysis for serotype biomarker. I have also implemented the first Comparative benefit risk (CBR) analysis within company for regulatory filing.

Senior Statistical Programmer, Product Development, Roche, UK (2008 – 2010)

I started programmer role when I joined Roche. My work was mainly on PHC (Personalized Healthcare). I worked closely with biomarker scientists, senior experimental statistician to perform biomarker analyses and intepret the results. I also collaborated with a member from Statistics Methods & Research Group on safety work. I have applied methods such as Propensity Scores, IPTW (Inverse Probability of Treatment Weighting), SSM (Sample Selection Models), IVA (Instrumental Variable Analysis) and CART (Classification and Regression Tree) to safety data analyses.

Academia Experience (2003 – 2008)

Before I joined Roche, I have worked for five years in Academia, as Post-doc/Research Associate, Statistical Geneticist at Universities in Sweden and UK. My research focused on Bayesian Inference in Computation Biology.

Training

I provide training for my colleagues:

  • Introduction to Machine Learning with XGBoost (Slides)
  • R Markdown
  • Bayesian Methods
  • Introduction to Github
  • Statistics for Non-Statistician
  • Working with R in the BEE (Biometrics Exploratory Environment)
  • R Shiny app “Predictive Probability” and “Data Driven Conditional Probability” (DDCP)

Tags

Contact