Dr Darsy Darssan

Lecturer - Biostatistics

School of Public Health
Faculty of Medicine
d.darssan@uq.edu.au
+61 7 336 55272

Overview

Dr. Darsy Darssan obtained three degrees in Statistics at mathematical sciences schools of three different universities. A Bachelor of Science with Honours in 2005 at University of Jaffna, a Master of Applied Science in 2008 at RMIT University and a Doctor of Philosophy in 2014 at Queensland University of Technology.

While doing his two years full time traditional face to face master degree, Darsy worked as a part time Statistician at Australian Council for Educational Research for a year.

Between the two bouts of postgraduate studies Darsy worked for two years. As a Statistician at University of New South Wales for an year and as an Associate Research Fellow in Applied Statistics at University of Wollongong.

While doing the highest degree in Statistics Darsy worked as a sessional academic, contributed to teaching introductory statistics to various cohorts of first year undergraduate students. Upon completion of the doctoral degree Darsy moved to University of Liverpool in UK to do his Postdoctoral research in Biostatistics. Darsy returned home in late 2015 and worked as a Biostatistician at The University of Queensland for three years prior taking the current position.

Dr. Darsy Darssan is appointed as a full time continuing teaching and research academic.

Teaching:

Current teaching: 2019 Sem 2

Introduction to Biostatistics (PUBH7630, for medicine - PUBH7290)

  • Postgraduate level course
  • Face-to-Face and Online mode
  • 120 Students
  • Role of Statistics in Health/Data/Measurements
  • Qualtiy data management
  • Sample size calculation
  • Significance tests
  • Statistical analysis plan

Past teaching: 2019 Sem 1

Introduction to Biostatistics (PUBH7630)

  • Postgraduate level course
  • Face-to-Face and Online mode
  • 142 Students
  • Role of Statistics in Health/Data/Measurements
  • Qualtiy data management
  • Sample size calculation
  • Significance tests
  • Statistical analysis plan

Career Statistician:

As a career statistician Darsy is interested in developing novel statistical methodologies and publishes in Statistics Journals. The following three research streams currently flow smooth: time to event outcomes and dynamic prediction, rigid strategies for pooling study results, and design methodologies for phase I/II trials.

Service Statistician:

Darsy has long experience working as a service statistician, mainly on clinical trials, designing studies, randomisation, preparation of protocols, statistical analysis plans, final statistical reports, and participating in data safety monitoring boards. Darsy successfully provided statistical service to Biologists, Rheumatologists, Ophthalmologists, Nephrologist, Endocrinologist and Health Service Researchers. Service statistician role is currently limited to those full time academic researchers who formally recognise Darsy as an investigator in their projects at the time of preparing the research proposals.

Research Interests

  • Experimetal design methodology
    Statistical methods in designing clinical and lab based studies. Adaptive design methodologies in early phase clinical trial design. Continual reassessment method and its variations.
  • Personalized Predictive Modeling
    Dynamic prediction using statistical machine learning methods
  • Bayesian modelling
    Using statistical models under Bayesian paradigm

Qualifications

  • Bachelor of Science, University of Jaffna
  • Master of Applied Science, Royal Melbourne Institute of Technology
  • Doctor of Philosophy, Queensland University of Technology

Publications

View all Publications

Supervision

  • Doctor Philosophy

  • Doctor Philosophy

View all Supervision

Available Projects

  • This PhD project will investigate the possibilities of linking Statistical modelling techniques with machine learning approaches to make realistic predictions. The project will develop systematic data-based dynamic modelling framework using supervised and un-supervised statistical machine learning techniques and integrate fractional polynomial, regression spline, and/or joint models. The student will have the opportunity to use big data collected on life course epidemiology and women’s health.

    The student must have first degree in Statistics with honours or master in Statistics or Mathematics. Experience with statistical software program R would be highly valued.

    The supervisory team will include school's lead biostatisticians.

    -------------------------------------------------------------------------------------------

  • Generally, randomised control trials (RCTs) are analysed using a number of simple statistical techniques such as t-test, ANOVA, ANCOVA, LME models, Cox regression. This PhD project will investigate the use of advanced statistical techniques in interpreting the RCT data. This include Bayesian techniques, false discovery-based approaches, bootstrapping, and AFT models.

    In the first year the student will do a systematic review on statistical methods in RCT and possible classification of methods involved. The second year of the candidature will identify possible alternative methods and apply them on simulated data under several scenarios. The final year will be further exploration of methodologies using real data. Although this project is an applied research, quick learning of advanced statistical methodologies and exploration of computationally challenging techniques require a student with first degree in statistics.

    Experience using statistical software programs R, SAS and STATA is must. Experience in analysing RCT data will be an advantage.

    The supervisory team will include biostatisticians and clinicians.

    -------------------------------------------------------------------------------------------

  • The ICH E9 guideline for clinical trials suggest the use of dynamic randomisation techniques to achieve balance across several factors. Multicentre trials suffer from imbalance, loss of medication packs, and operational bias when a dynamic randomisation is used. This minor research project will investigate the feasibility of implementing covariate adaptive randomisation techniques in multicentre trials.

    The project suit to a statistics student who would like to carry out an honours or master thesis in biostatistics for a year. The first part of the research will review all the variations of covariate adaptive randomisation techniques including the algorithm based minimisation, model based CAR, and distributional balance techniques. The second part of the research will estimate imbalance and logistical costs under several scenarios using simulated data. Real multicentre RCTs will be used to retrospectively estimate the logistical costs due to the choice of CAR.

    The student must have experience with R or SAS-IML or Mata-Stata.

    Reference:

    1. Hu F, Hu Y, Ma Z, Rosenberger WF. Adaptive randomization for balancing over covariates. WIREs Comput Stat. 2014; 6:288–303

    2. Taves DR. The use of minimization in clinical trials. Contemp Clin Trials. 2010 Mar; 31(2):180-4

    3. Weng H, Bateman R, Morris JC, Xiong C, Validity and power of minimization algorithm in longitudinal analysis of clinical trials. Biostat Epidemiol. 2017; 1(1): 59–77.

    -------------------------------------------------------------------------------------------

View all Available Projects

Publications

Featured Publications

Journal Article

Conference Publication

PhD and MPhil Supervision

Current Supervision

  • Doctor Philosophy — Associate Advisor

  • Doctor Philosophy — Associate Advisor

Possible Research Projects

Note for students: The possible research projects listed on this page may not be comprehensive or up to date. Always feel free to contact the staff for more information, and also with your own research ideas.

  • This PhD project will investigate the possibilities of linking Statistical modelling techniques with machine learning approaches to make realistic predictions. The project will develop systematic data-based dynamic modelling framework using supervised and un-supervised statistical machine learning techniques and integrate fractional polynomial, regression spline, and/or joint models. The student will have the opportunity to use big data collected on life course epidemiology and women’s health.

    The student must have first degree in Statistics with honours or master in Statistics or Mathematics. Experience with statistical software program R would be highly valued.

    The supervisory team will include school's lead biostatisticians.

    -------------------------------------------------------------------------------------------

  • Generally, randomised control trials (RCTs) are analysed using a number of simple statistical techniques such as t-test, ANOVA, ANCOVA, LME models, Cox regression. This PhD project will investigate the use of advanced statistical techniques in interpreting the RCT data. This include Bayesian techniques, false discovery-based approaches, bootstrapping, and AFT models.

    In the first year the student will do a systematic review on statistical methods in RCT and possible classification of methods involved. The second year of the candidature will identify possible alternative methods and apply them on simulated data under several scenarios. The final year will be further exploration of methodologies using real data. Although this project is an applied research, quick learning of advanced statistical methodologies and exploration of computationally challenging techniques require a student with first degree in statistics.

    Experience using statistical software programs R, SAS and STATA is must. Experience in analysing RCT data will be an advantage.

    The supervisory team will include biostatisticians and clinicians.

    -------------------------------------------------------------------------------------------

  • The ICH E9 guideline for clinical trials suggest the use of dynamic randomisation techniques to achieve balance across several factors. Multicentre trials suffer from imbalance, loss of medication packs, and operational bias when a dynamic randomisation is used. This minor research project will investigate the feasibility of implementing covariate adaptive randomisation techniques in multicentre trials.

    The project suit to a statistics student who would like to carry out an honours or master thesis in biostatistics for a year. The first part of the research will review all the variations of covariate adaptive randomisation techniques including the algorithm based minimisation, model based CAR, and distributional balance techniques. The second part of the research will estimate imbalance and logistical costs under several scenarios using simulated data. Real multicentre RCTs will be used to retrospectively estimate the logistical costs due to the choice of CAR.

    The student must have experience with R or SAS-IML or Mata-Stata.

    Reference:

    1. Hu F, Hu Y, Ma Z, Rosenberger WF. Adaptive randomization for balancing over covariates. WIREs Comput Stat. 2014; 6:288–303

    2. Taves DR. The use of minimization in clinical trials. Contemp Clin Trials. 2010 Mar; 31(2):180-4

    3. Weng H, Bateman R, Morris JC, Xiong C, Validity and power of minimization algorithm in longitudinal analysis of clinical trials. Biostat Epidemiol. 2017; 1(1): 59–77.

    -------------------------------------------------------------------------------------------

  • Fractional polynomials have been used for assessing within-study non-linear relationship and pooling via inverse-variance weighted method. This project will explore the use of smoothing and penalized splines for estimating the within-study functional form of exposure-outcome relationships. Comparison will be made to the use of fractional polynomial and categorisation of continuous covariates. Sensitivity analysis will be done for knots choices in Splines.

    The project suit to a statistics student who would like to carry out an honours or master thesis in biostatistics for a year. The first part of the research will involve a reading course in fractional polynomial, splines and meta-analysis of individual participant data (IPD). The second part will involve fitting splines to 11 studies with IPD. The 11 studies with IPD comes from InterLACE. The last part of the research involve comparison and sensitivity analyses.

    The student must have experience with Stata or R.

    This project suit to a Master of Biostatistics (BCA) student (STAT7620 / STAT7621).

    Reference:

    1. Sauerbrei W, Royston P. A new strategy for meta‐analysis of continuous covariates in observational studies. Statist Med. 2011;30(28):3341‐3360.

    2. Burke DL, Ensor J, Riley RD: Meta-analysis using individual participant data: one-stage and two-stage approaches, and why they may differ. Stat Med 2017; 36:855–875.

    3. Mishra GD, Anderson D, Schoenaker DA, Adami H-O, Avis NE, Brown D, et al. InterLACE: a new international collaboration for a life course approach to women’s reproductive health and chronic disease events. Maturitas. 2013;74(3):235–40.

    -------------------------------------------------------------------------------------------