Reword the introduction, don’t use indispensable, use important
Put less emphasis on the eigenvector centrality, need to rephrase
the last sentence of the last paragraph
Reword the hypothesis because we are using eigenvector centrality;
align it with the objective in abstract
Methods:
Describe IADB and the access date to the database
Don’t duplicate explanation on technical parts, compare it to the
introduction, check the explanation on eigenvector centrality
Omit the terminology on knowledge graph
Results:
In the descriptive results, we need to categorize the type of
polypharmacy
Move the daily-variation figure to the supplementary file
Make the scientific annotation more consistent, e.g. when reporting
the eigenvector centrality we can limit to two digits after decimal
mark and use $c_e x 10^{-n}$
Discussion:
First paragraph: Discuss about polypharmacy
Second paragraph: Explain the concordance with other literatures
Third paragraph: Explain about the DPN
Fourth paragraph: Strength
Fifth paragraph: Limitation
Lost of individual values
Add limitation about weighting the DDD $\to$ How does it reflect
augmented polypharmacy?
2024-08-15
Discuss IADB draft by next week (2024-08-22)
Remind Prof. Mako for a meeting among supervisors
Ask around other people in the department about qualitative research
Prof. von Dijke
Sieta
Kaying Kan (now a postdoc)
Check Equator website, find quality scoring for real-world evidence
paper, e.g. GRADE for study design
GetReal EU study
Publishing the IADB work:
Journal of pharmacoepidemiology $\to$ Discuss with Prof. Eelko Hak
American/European epidemiological journal
Social science in medicine
Check other DPN publication
Check also psychiatry journals
Submit the abstract to IHEA (Bali): IADB work, SKI preprint, or
scoping review $\to$ Check the deadline
2024-07-25
Scoping review:
Add resilience as mediator
Exclude coping behavior
Consider defining the psychological outcomes to make it a bit more
specific
Psychological outcomes are defined as ICD-10 from F20-F59 and its
related signs/symptoms
2024-07-11
For DD: Needs an MoU between FSE and UI
Discuss with Prof. Mako about the DD and administration challenge
Draft an email to FSE regarding the MoU between FSE and UI $\to$ Send
this to Prof. Talitha first
Prof. Talitha and Prof. Frederike will be on holiday by the end of
June
Reschedule next week’s meeting to the week after $\to$ This will only
be with Prof. Frederike
2024-06-13
Network analysis manuscript:
Substitute the author position between Prof. Frederike and Spyros
Use common terminology: dynamic network analysis
Change title: “Eigenvector centrality as a metric of polypharmacy: a
dynamic network analysis to quantify co-prescription around
antidepressants and anxiolytics”
Look at the definition of polypharmacy
In the background: clarify about comorbidity $\to$ comorbid to
chronic diseases
In the discussion: find the link between diagnoses using high
eigenvector centrality medications
Clarify the background and objectives, state why network analysis is
relevant in this case
In the introduction: explain why we need to make a surveillance
system for polypharmacy
Discuss the method section with Spyros before finalizing the results
We need prevalence of co-morbidity in people with depression or
anxiety $\to$ Put it in the discussion section
Scoping review:
If we get too many screened articles, limit to the recent 20 years
Indicate the type of population in the included articles
2024-06-04
Email collaborators for IADB work later on this week to update about
abstract that will be submitted to ISPOR
Presentation prep:
Present slower $\to$ Adjust your breathing, plant the feet to the
floor
In example: Explain the drug, not the ATC code
Add the limitation:
Add a flow diagram of medication cleaning
Explain at first that this is a network analysis on polypharmacy
Skip the explanation of network measures, focus more on eigenvector
centrality
2024-05-23
9-mo interview form is in Prof. Frederike, I will receive it after she
signed it
Prof. Talitha will email Prof. Ralph about the DD program
Prof. Frederike will follow up on the lunch meeting ICPE
IADB data analysis:
Talk with Jens on how to remove the data with DDD outliers
One option is to remove based on the period of use
Change the table header: It’s not weighted DDD, it’s a weight to
generate edges
As a follow up: Prof. Talitha will contact a pharmacist to further
discuss our finding on eigenvector centrality
2024-05-14
Descriptive findings of IADB:
Distribution of DDD for each medication
Tabular description of the data
Inform Fang so that I can present my work on the 5th of June
Move the biweekly meeting to 13th of June
Prepare a slide for PEGET meeting $\to$ 20 minutes per presenter
2024-04-29
Remove action points from the revised TSP $\to$ Include the project
definition into TSP
Chapter 1: Still some rooms for revision, need to work on comments
in the document later
Chapter 2: Revise the draft
Chapter 3: Make sure that this is connected with chapter 2
Chapter 4: Outside of Prof. Talitha’s expertise, need a co-author
who knows a bit more on this concept (especially LDI, and MINI)
Check Lifelines Wiki
Check literatures using LDI and MINI
Chapter 5: Narrow down the scope to universal mental health
promotion program in workplace
Check the literatures on Dutch and international
Make sure that the model is transferrable
Revise the TSP and include the project definition as an appendix:
Edit the timeline: include preparation, execution, and reporting
phases
Y1 and Y2: detailed per quarter
Y3 and Y4: detailed per semester
Remove plan on survey
Add chapter plans
Work on the interpretation of findings
Less jargon, more verbose concept
Always include an introduction
Focus on the research question and answer it
Decide the journal for IADB manuscript $\to$ Involve Spyros for this
manuscript
For each chapter, decide who to involve and when to involve them
Involve Prof. Mako from the start for the 1st and 4th chapters
Involve Prof. Frederike more in the scoping review
2024-04-24
Contact Prof. Mako for a bimonthly meeting
Fill in and send the 9-mo interview form $\to$ Send it by EOW
Submit the IADB abstract to EUPHA and ISPOR
RIVM research group will attend EUHEA as well $\to$ Networking with
Hanneke to talk about resilience
Weighting method when creating a DPN:
Dividing and normalizing (what I did)
Multiply by dose
Multiply by dose then log
Lunch meeting with Prof. Frederike’s group on Tuesday $\to$ Many talks
about dynamic model in psychiatry
Ask Steef about who might be involved in dynamic model in psychiatry
2024-04-15
Chapter 1:
The topic is too broad
Need to include someone with an expertise on resilience
Reformulate the research question to be more narrow: “How does
resilience work as a moderator in relation with stressors and mental
disorders?”
In project scope, explain that neurobiological/physiological
mechanistic explanation is not a part of the review
In activity plan: Contact collaborators sooner
Action point: Make the question simpler, contact the librarian,
create an inclusion based on model for chapter 5
Chapter 2: Involve Spyros for this chapter
Chapter 3: Narrow down the objective to evaluate psychopharmaca uses
of anxiolytics and antidepressants
Chapter 4:
We need someone with an expertise in psychological resilience
Signifies how this chapter is required for modelling
Chapter 5:
Limit the scenario to universal intervention
Focus the intervention to workplace-based
Send the paper on about resilience
State that we only focus on depression and anxiety part of MINI
using Lifelines data
2024-03-29
Send the initial draft before the 3rd of April
Send the ABM papers to Prof. Talitha
Read more about the international guideline on polypharmacy
To learn in project management course $\to$ Delegating task and
collaborating with others
2024-03-25
Three courses of action upon the 9-mo interview:
Plan A: Re-align the research purpose, go with the current
supervisory team
Plan B: Find another suitable supervisor
Plan C: Terminate studentship
Course on science communication
Start the introductory essay by explaining ABM, limit to 3,000 words
Write with less jargon
2024-03-22
Scoping review:
Look a little bit into resilience, find some general review on
resilience and formulate a question on this
Make a clear inclusion and exclusion criteria
Create a working definition of resilience → Done, write it
explicitly
Check if the body of evidence is sufficient → Do it for plan B as
well
Find a co-author with expertise on state resilience
Plan B: Scoping review on the measurement of resilience
Concurrent psychopharmaca uses: Send a draft of introduction +
methods, deadline on the 19th of April
Trend of psychopharmaca uses:
Formulate the questions better → Why do we need to monitor it this
way?
What is the societal problem to solve here?
Example: The danger of polypharmacy
Other idea: Look at the duration of use → Read more literature
on this
Lifelines study: Research question $\to$ How applicable is the
residual method of measuring resilience?
ABM: Publish the model-building procedure
2024-03-21
Having a discussion with Hanneke from RIVM $\to$ PI for youth model on
mental health
Research group in Rotterdam using a group-based model
Conceptual framework:
Simplify the outcome to only include depression
Give a rationale why medication can influence resilience
IADB analysis:
Create a summary descriptive analysis for the original IADB dataset
Measure time of follow up for each person
Eigenvector centrality: Sort out pattern of co-prescription
systematically
For discussion section: Compare with other papers evaluating
co-prescriptions
Explain and discuss the plan for the chapter outlines $\to$ Then
explain how they are connected
Think about the author team
It takes about 8 month to prepare a data request for Lifelines $\to$
Consult Taichi on this
2024-03-15
IADB analysis:
Make a simulation to assess which weighting method suits the dataset
$\to$ Per discussion with Spyros:
Baseline: Unregularized matrix of number of claims
Contact dr. Frederike about the plan of joining WHO WMH survey
2023-10-19
Umbrella review:
Intended message: Raising awareness of depression in T2DM patients
Discuss about what figures to include in the main manuscript and
supplementary
World Health Bulletin
Lancet Global Health
Should focus more on the diabetes rather than the depression
Discuss which journal to submit
Results:
Create a footnote on the first table: Difference in reanalysis may
be due to difference in selection
Explaining the meta-regression:
Change the reference into Americas (region) and others
(instrument)
Add the reference to the table
Narratively explain how to read the table, give an example
Discussion:
Potentially need to discuss why the results are different
Scoping review:
Database to use: PsychLit, PsychInfo, PubMed, Scopus, Embase, Web of
Science
Need to converge the methods for all planned studies
Prepare for 6-month interview, talk about it in the first meeting with
Prof. Mako
2023-10-12
Meeting with Prof. Mako: 27 Oct 2023, 16.00 GMT+7 (09.00 GMT+2)
Create a time table for the discussion with Prof. Mako and
Dr. Frederike $\to$ Plan the minutes of meeting
Propose:
Prioritize bibliometrics analysis $\to$ Target: journal submission
by December, this will also help my introductory essay
Expect delays on analysis of IADB data $\to$ Still need one more
week to breakdown the toolbox query, will contact Jens to schedule a
meeting tomorrow EOD
Scoping review following bibliometrics analysis
Slide:
Speech is ok, but slide is too fast
Keep the point you’d like to discuss as the main focus of the slide
Hanneke from RIVM creates a scoping review on system dynamic model in
healthcare $\to$ In the later phase of my scoping review, I should
consider having an FGD with RIVM team
Umbrella review:
Preserve the outliers when analyzing the data
Create a pairplot and do descriptive analysis from data with
preserved outliers:
Variables:
IV: Prevalence
DV: Year of publication, region, and psychometric instrument
When fitting metaregression: Set region, year of publication, and
psychometric instrument as covariates
Consider extracting the dataset and perform clustering analysis
2023-10-05
Umbrella review:
Do not remove outliers
Use GBD data to calculate the relative risk of having depression in
T2DM
Complete the data analysis sub-section in the manuscript
Storytelling:
Do umbrella review on 23 studies $\to$ Fetch all the primary
studies
Redo meta-analysis, see that the variance is too high
We attempted to explain the variance $\to$ Subgroup meta-analysis
and meta-regression
The heterogeneity is 2% explained by year, meaning that overtime
the prevalence of depression is increasing among T2DM patients
Interestingly, the GBD data shows that MDD prevalence stays the
same while T2DM is increasing overtime
The increasing MDD prevalence in T2DM might be explained by T2DM
Data from GBD also shows increasing risk of depression in T2DM
patients
IADB:
Continue breaking down the source codes
Plan a meeting with Talitha and Jens
2023-09-29
Meeting with Prof. Mako should be rescheduled to 20 or 27 October
Umbrella review:
Anomaly detection:
Instead of removing the outliers based on 95% CI, it’s better to
apply outlier detection algorithm
Use IQR as a criteria
Span the results based on year, do a five-year rolling window
Metaregression:
Standardize the covariance, they should have the same range
Year: Set the smallest value of year as an index year
Experimental model: Fit a model without $\beta_0$
Addressed concerns:
Pooling of proportion from all entries:
Do we really want to pool over all studies? That is, do we
think we can find a single estimate of the prevalence of MDD
in T2DM that would be relevant worldwide and irrespective of
outcome or time.
Maybe here the proportion as a measure of prevalence differs
from an effect measure, where you may reasonably assume it to
be independent of setting in principle.
Maybe when we combine all the 634 studies, this gives us the
global average, which of course reflects a lot of
heterogeneity (as we may expres by presenting a histogram). We
should then ensure not to take more than one proportion from
each study. (so it will be less than 634).
Maybe that is more relevant than taking only the studies that
are closest together (to avoid heterogeneity). Since we do
expect and understand heterogeneity.
Subgroup analysis:
Do we indeed expect that once we concentrate on country, we
can expect to pool? I would be inclined to think so, but as
you show, also time, outcome measure and maybe more do matter.
So question: Once you stratify by country: can we combine
studies into a setting specific average? Is homogeneity
sufficient? If not, does it make sense to select studies based
on being no outlier? Or would it be more appropriate to
further stratify for factors that will explain heterogeneity,
like time of study and outcome measure?
Next approach to explore:
About the pooling. Another factor that could explain
heterogeneity is underlying prevalence of MDD and of T2DM.
Next to outcome measure, that may also be a matter of risk
profile of the population (e.g. obesitas).
So could you and Nora also find the expected prevalence (by
country/by time), based on published prevalence estimates of
MDD and of T2DM (using GBD, WHO, OECD or IHME published
numbers, whatever is most clear and consistent); calculate the
expected proportion of MDD in T2DM based on such prevalences
and then start doing the meta-regression on the remainder (So
the additional prevalence of MDD over what you may expect) or
maybe on the RR (the crude proportion divided by the expected
proportion) or Odds Ratio.
Possibly this latter measure is one that you may expect to be
the same irrespective of time or setting more reasonably,
since it is more of an effect measure (the effect of T2DM on
MDD risk). So then it becomes more reasonable to use
established methods of meta-regression packages, which were
initially developed for pooling effect measures.
For the analysis of the observed proportions, I think it is
mostly the heterogeneity that you need to show in the results:
SO how this varies indeed by country (a map would be great),
by outcome measure etc.
If you take more than one outcome per study, how could you
take account of the dependence these will have? Multilevel
analysis an option? Or using clustered standard errors??
Analyzing RR:
Get the prevalence of MDD in the general population based on year
and/or country $\to$ Check GBD or IHME
RR = Prevalence of MDD in T2DM, as reported, divided by the
prevalence of MDD in the general population
Compare the RR for each subgroup
Bayesian meta-analysis:
Use half-cauchy distribution as a prior for SE
Use beta distribution as a prior for the proportion
IADB:
Breakdown the toolbox, check what each lines do
Formulate questions to discuss with Jens
Check courses for meta-regression
2023-09-23
Umbrella review:
Might need to redo the analysis
To discuss with Katja and Nora:
Pros and cons of the current approach
Discuss about multiple entries
Discuss with Xinyu regarding meta-regression done by the
mathematician in RUG:
Comparison of methods and packages to conduct metaregression
How to handle logit transformation
IADB data:
Recheck the use of persistence toolbox $\to$ Combine parts of the
code to my query
Change the overlap query into weekly basis
Check what we know about concurrent medication uses before and
during the pandemic
Scoping review:
Use multiple search to address:
Stressors, self perception, and resilience
Medication/policy and resilience
Resilience and psychological outcomes
Check PRISMA guideline on selecting articles for a scoping review
Perform bibliometrics analysis to narrow down the search
Need to create a clear selection criteria
Have a discussion with Frederike Jorg/librarian about the search
strategy
Check on published articles using scoping review to investigate
psychological disorders:
Align the research questions with IADB data request protocol:
1: Add the phrase: “In relation with drug’s use”
2: Check the data request protocol
3: How does an individual perceive their health in relation with
environmental stressors, such as the COVID-19 pandemic?
In scoping review: Include the role of medicine to resilience
Find a co-author with primary domain in neuropsychiatry or
neurpsychology $\to$ Contact dr. Khamelia
Simplify the idea about the model
2023-07-20
Contact Prof. Mako regarding monthly or bi-monthly regular meeting
$\to$ Output: Brief MoM
Protocol draft:
Table of population estimate: Check the box to include this data
Title: Add in adults
Overview of literature: Add other study on the effect of pandemic on
the use of psychopharmaca $\to$ Tell how many studies have been
done, and how they were investigated
In methods: Add comparison during the pandemic because different
policy might have different impact $\to$ Add to the exploratory
question
Clarify the research question, create a footnote to explain what a
daily epoch is
Specify what the pattern implies in the secondary question
Specify “the past five years” as 2018-2022
Exclude patients whose prescribed duration is less than two weeks
Prescription table: Add variable request for the prescriber
Rewrite he effect measures $\to$ Clarify the point in a layman term
In statistical methods:
First paragraph should be the summary of how I plan my analysis
Explain how to analyze the differences before and during COVID-19
Link each method with the research questions
Add references to the graph knowledge embedding
Clarify the aggregation on weekly, monthly, and quarterly basis
(first paragraph in methods)
Specify what $N$ means $\to$ It is the total number of people
making medication claim on daily basis
In the fourth paragraph
Explain how to handle patients with multiple drugs (use examples,
as explained in the email)
Use subheadings to clarify which paragraph belongs to which ideas:
Executive summary (in bullet point)
Data management and feature engineering
Concurrent medication use as a matrix
Data analysis plan
Show the formula for eigenvector centrality $\to$ Refers to the R
package being used and its function
Explain that the node is the medication
Last paragraph: Refer subgroup to the “Stratification of data”
section
Stratification on age:
WHO/ILO: 18-25, 25-35, 35-45, 45-55, >56
Theory: 18-25, 25-45, 45-55, >56
Report: Interim analysis should be discussed and reported to IADB as
well
Discuss with Nynke Bosman regarding the medication and policy in the
Netherlands
Ask Fang about Dutch’s guideline on MDD and anxiety disorder $\to$ Use
this as a baseline to select the medication
There’s a table describing zip code into socioeconomic status
Ask Frederick about courses for time-series analysis
2023-07-13
Complete the data request form by next week
Draft a SQL query for data request
Proposal revision:
Choice of graph metrics $\to$ Include this as a literature review
$\to$ Mention in subsection 4.2
Last paragraph in 4.2 is redundant, make it more compact
Add about the time-series analysis
Clarify the overall purpose: Is it to evaluate seasonality (pattern)
or trend?
Add about the age range: 18-65
In the introduction of 4.2, explain why concurrent drug use is
important to evaluate
Guideline to use when selecting the medication
Master’s course of time-series analysis in Autumn $\to$ In mathematics
and economics department:
Mathematics: Fitting dynamical model to data
Economics: (Needs further searching)
2023-07-06
Complete the SQL course from IADB, then submit it right away
Read the book: Going beyond cost-effectiveness by Kaying Kan
Consider doing:
Interrupted time-series analysis
Moving average
List all medications needed $\to$ Will discuss it later