Interfaculty Center Data processing & Statistics

Wilfried Cools, principal consultant Life Sciences and Medicine
https://www.icds.be/

Interfaculty Center Data processing & Statistics

ICDS

  • team of experts in applied statistics & methodology
    • principal consultants:
      wilfried (Life Sciences & Medicine) and lara (Human & Social Sciences)
      psychologists turned into applied statisticians for the love of research
    • coordinator, kurt, the math-stats guru at ease with real life applications
    • supporting consultants, researchers who share their expertise voluntarily
      (computer scientists, mathematicians, criminologist, ... )
  • bookable @ www.icds.be/consulting/
  • more details @ www.icds.be

methodological & statistical support

  • provide complementary support to our research community (VUB - UZ Brussel)

    • consultancy: asking the right questions the right way
    • data analysis: getting the relevant information out of it
    • tutoring: learning to do things yourself
    • automation: getting the computer to do it for you
    • ...
  • supported by: research board (UMC) & vice-rectorat research policy

  • conditions:

    • consulting free of charge for our research community
    • data-analysis implies co-authorship, includes follow-up / review
    • for longer-term commitments a suitable compensation is considered
    • check our Terms of Service at www.icds.be

consulting

asking the right questions the right way

  • discus how to

    • turn substantive questions into research aims (the right questions)
    • address these research aims (the right way)
      • with appropriate study design and sample size
      • with appropriate statistical techniques
  • includes help on

    • interpretation and communication of the results
    • writing of research proposals
    • ...

consulting - practicalities

asking the right questions the right way

  • typically involves one or a few meetings (< hour)
  • at the consultants office (1st floor, D-building)
  • booked through the booking-tool
    requesting 3 time-slots from which one is selected
  • also available for master students working on thesis
    approved by the promoter
  • does not include statistical analysis

data analysis

getting the relevant information out of it

  • perform for you

    • data analysis (descriptive, inferential)
    • data management (eg., cleaning and preparation)
  • includes

    • communication of data and analysis to the researcher
      • report and R-code (supporting reproducible research)
        • with visualizations, summaries
        • with discussions on choices and inference
      • meeting (consulting)

data analysis - practicalities

getting the relevant information out of it

  • typically time consuming: a few days to a few weeks
  • researchers are expected to participate
    • translate substantive interests into the language of statisticians
    • provide data that is easy to understand and deal with
    • show appropriate skepticism to ensure analyses are correct
    • try to understand and learn for future research
  • as co-author, consultants take responsibility
    • support in writing the statistical part of the manuscript
    • dealing with reviewer comments

data analysis - case 1

Flamée, P., Varnavas, V., Dewals, W., Carvalho, H., Cools, W., Bhutia, J. T., Beckers, S., Umbrain, V., Verborgh, C., Forget, P., Chierchia, G.-B., Brugada, P., Poelaert, J., & de Asmundis, C. (2020). Electrocardiographic Effects of Propofol versus Etomidate in Patients with Brugada Syndrome. Anesthesiology, 132, 440–451.

  • major revision, primarily due to statistics
  • removed all statistics
  • enhanced story, through visualizations
  • no clear indication of systematic change

data analysis - case 2

Smeets, S., Stangé, G., Leuckx, G., Roelants, L., Cools, W., De Paep, D. L., Ling, Z., De Leu, N., & in’t Veld, P. (2020). Evidence of Tissue Repair in Human Donor Pancreas After Prolonged Duration of Stay in Intensive Care. Diabetes, 69, 401-412.

  • refined existing analysis
  • introduced permutation tests
  • included visualizations
  • differences with time 0 exist

data analysis - case 3

Baert, Y., Ruetschle, I., Cools, W., Lorenz, A., Marx, U., Oehme, A., Goossens, E. and Maschmeyer, I. (in press). A multi-organ-chip co-culture of liver and testis equivalents: a first step toward a systemic male reprotoxicity model. Human Reproduction.

  • dealing with
    • repeated measurements
    • deviating 1st measure
    • small sample sizes
    • multiple conditions
  • modeling evolution over time with mixed models
  • various conclusions drawn
    • on differences in evolution
    • on differences at end

data analysis - case 4

Gilissen, J., Gastmans, C., Wendrich-van Dael, A., Cools, W., Vander Stichele, R., Deliens, L., Van den Block, L. & Pivodic, L. (accepted). Nurses’ self-efficacy, rather than their knowledge, is associated with their engagement in advance care planning in nursing homes: a survey study. Palliative Medicine.

  • reports for communication
  • R-code for reproducibility

tutoring

learning to do things yourself

  • workshop: sample size calculation with GPower

    • introduction to reasoning on sample sizes with exercises
    • 4 times for the doctoral schools, slides available at www.icds.be
  • presentation: key issues in communicating methodology and statistics

    • primarily intended for applicants WF Willy Gepts
    • documented on-line at www.icds.be
  • other workshops/presentations exist at ICDS, and more are planned

    • contact the coordinator to discuss what we can offer

automation

getting the computer to do it for you

  • automatic processing of data

    • into statistical results, visualizations and tables
    • into reports or slides
  • example: automatic processing of future survey data

    • build database with available data
    • writing code to process new data
      • update the database
      • generate updated slides and reports
    • in collaboration with commissioning doctors

supporting consultants

  • researchers who support ICDS as consultant:

    • expert statistics, mathematics: prof. Kurt Barbé
    • expert statistics, eHealth: prof. Ronald Buyl
    • expert pre-clinical trial design to support ECD: Susanne Blotwijk
    • expert informatics, experienced in applied statistics: Sven Van Laere
    • expert informatics, experienced in qualitative research: Iris Steenhout
  • www.icds.be/consulting/