Research Projects
Causal Inference from Dynamic Data to Improve Environmental Health

Programme(s) to which this project applies:

☑ MPhil/PhD ☒ MRes[Med] ☒ URIS

Compelling environmental health policy concerns call for more definitive epidemiologic evidence. Causal science advances, particularly with dynamic time series data, provides us with opportunities to assess more rigorously how environmental factors affect population health and contribute to smarter environmental health policy based on more rigorous causality evidence. Frustration with the "yes, but" research findings. Both the general public and the policy makers are often frustrated with the academic research articles that never have health advice without caveats and the "more research is needed" bottom lines. The hope for definitive guidance on the threats of air pollution and extreme weather is often dampened by the "yes, but" research findings. The ambiguity in these research findings results not only from the inherent complexity of environment-health relationships, but also from the weak causal evidence in the correlation and regression analysis of observational data which are vulnerable to residual confounding.

Causal science advances. The fundamental challenge to infer causality from observational data is that we aim to derive counter-factual conclusions with only factual premises. Causality means changing X variable will change Y variable while keeping all the other variables the same; these causality findings are what the decision-makers need to know for making choices about what changes to implement. But with the real-world observational data, correlation and regression methods are still the most commonly used tools to analyze environment-health relationships, which are vulnerable to residual confounding and can lead to incorrect conclusions. Indeed, interventional experiments would produce the highest weight of causal evidence, but these experiments are either not feasible or ethically forbidden when we study environmental factors and public health. Evolving away from the 20th century fatalism understanding of causality, causal inference methods using observational data are being developed and have recently seen some applications in neuroscience, climate science, and ecology studies. Examination of the generic applicability and extension of these novel methods to other fields including environmental health in the current project is a compelling task.

Dynamic time series data advantages. Without violating any individual’s privacy or other ethical norms, we already have large amounts of real-world dynamic time series data on both environmental quality and public health status at entire city or region levels. These time series data are known for their advantages to estimate the conditional association between shortterm environmental factors and health indicators, where the compared units are days and the study question can be whether the days of high death rates correspond to those with high pollution. City-specific rates of smoking and obesity etc., that do not change at short-term time scales, would not confound the environment-health associations.

Which methods to use for causal inference? New causal analytics methods are being proposed almost monthly, but there are very few publicly available tests that are competent in assessing the new methods. Most of the new methods use simulation data for development and validation; few of them have seen applications in real-world data. One practical strategy is triangulation of evidence: to integrate results from multiple approaches with distinct assumptions and unrelated sources of potential bias.

This project aims to assess the complementary nature of distinct causal analysis methods in cases studies addressing the relationship of air pollutants/weather variables with various health outcomes, and formulate a framework to triangulate causality evidence from real-world observational data in the domain of environmental health.

Dr LW Tian, School of Public Health

Dr Linwei Tian is an environmental epidemiologist with a focus on air pollution and health. He has been conducting field epidemiology and laboratory work on indoor air pollution and lung cancer in Xuan Wei County, which has the highest lung cancer rates among women in China. Identifying the carcinogenic agents in coal and its emissions would affect local intervention policies and gain insights into the carcinogenesis mechanisms.

Meanwhile, Dr Tian gains a strong research interest in exploring the potential role of food contamination by biogenic siliceous needles in the endemic cancer of esophagus in China. The geographic patterns of esophagus cancer endemics offers a unique natural experiment in assessing environmental carcinogenesis. Based on the previous evidence on glass (silica fiber) roots of cancer, Dr. Tian hypothesizes wheat bract-derived glass fibers and needle-shaped diatoms in guts of trash fish as a major causative factor of esophagus cancer endemic in North China and southern China, respectively.

Urbanized Hong Kong provides another unique setting to study air pollution and health. Its high density of people and vehicles, high-rise buildings, a rich resource of accessible environmental measurement and healthcare data, and various air pollution control policies offers a great opportunity for valuable environmental epidemiology. Compared with static data, time series data contain far more information at our disposal for the inference of causality. Dr. Tian has been trying to examine the earlier ambiguity and enhance causal inference of the environment-health associations by contrasting the traditional time series regression models with the recent methods of causal discovery from big data.

HKU Scholars Hub
Laboratory Homepage

For more information or to express interest for this project, please email the supervisor or the specified contact point in the project description.  Interested candidates are advised to enclose with your email:

  1. your CV,
  2. a brief description of your research interest and experience, and
  3. two reference letters (not required for HKUMed UG students seeking MRes[Med]/URIS projects).

Information on the research programme, funding support and admission documentations could be referenced online at the Research Postgraduate Admissions website. General admission enquiries should be directed to

HKUMed MBBS students interested in the Master of Research in Medicine (MRes[Med]) programme may visit the programme website for more information.  

HKUMed UG students interested in the Undergraduate Research Internship Scheme (URIS) may visit the scheme’s website for more information.