Trained on both statistics and machine learning, I received a PhD on computational biology in 2009, and started a career as a data scientist in cancer research. From 2014-2018, I held a lectureship in biomarker statistics at the Institute of Cancer Research, University of Manchester. After that, I joined the National Biomarker Centre, CRUK Manchester Institute as the lead statistician. My research focuses on tackling the data science challenges in the discovery and validation of cancer biomarkers, which is vital for delivering personalized precision medicine. To date, I have published 42 peer-reviewed papers, secured fundings over 4 million pounds as an investigator, and has led the design of 5 clinical trials. 

Thanks to the advance of molecular biology, the volume and complexity of clinical and biological data available have expanded at a lightning speed in the last two decades. There is an unmet need for novel data analytical approaches to integrate multi-modality data collected serially during cancer treatment, and to subsequently identify novel biomarkers that can inform clinical decision making. In the past 10 years, I led the data science component of a research program that seek biomarkers for anti-angiogenesis cancer treatment via liquid biopsy. With synergy from data scientists, clinicians and biologists, this program successfully aggregated clinical, molecular and MRI imaging data serially generated in 5 different clinical trials, covering a spectrum of cancer types and anti-angiogenesis drugs. A novel Bayesian joint modelling approach was developed, resulting in the identification of a cytokine called Tie2 as a pan-cancer response biomarker of anti-angiogenic cancer treatment. The longitudinal changes of this biomarker can inform response and resistance status of a patient at a real time during treatment, enabling triaging the patient to the right treatment at a right time. Currently, I am leading the design of clinical trials to validate and translate this biomarker into routine clinical use. The first of these trials, VALTIVE1, started in 2020 in ovarian cancer patients, while its expansion, VALTIVE1b, started in 2023. More trials in other cancer types are being prepared, alongside the development of novel adaptive trial design methodologies for biomarker guided trials. If successful, the biomarker will change the way anti-angiogenic cancer treatment is managed in many cancer types, and will be the first biomarker that completes the CRUK precision medicine roadmap.   

Early detection is vital to the treatment and prevention of cancer. Part of a multi-disciplinary team, I helped design the pioneering community-based lung cancer screening study in 2016 which has received a huge success, and the idea has been adopted nationwide. Moving along this direction, my colleagues and I are trying to invent novel approaches to detect lung cancer via blood tests and modern high-throughput omics technologies. With the hypothesis that tumorigenesis is led by coalition and interaction of systematically dysregulated biological pathways, rather than a limited number of driver proteins/gene mutations, I am developing statistical and machine learning approaches to integrate multi-omics data, to deconvolute heterogeneous signals from early-stage tumours, and to understand the crosstalk among different -omics levels. My objective is to discover novel data pattern in the inter-regulated multi-omics system as a biomarker to inform the onset of tumorigenesis.