OHDSI Home | Forums | Wiki | Github

[Patient Level Prediction Workgroup] Meetings

Researchers
#21

Thanks all for the nice interactive TC today.

I think we will have a very interesting next year ahead of us with all
these research questions on our todo list.

If there are any research areas you like to add to what has been discussed today please let me know:

  1. Temporal data analysis
  2. Deep Learning
  3. Sparse Coding
  4. Knowledge-Data driven hybrids
  5. Leveraging textual information
  6. Studying transportability
  7. Developing Parsimonious Models
  8. Visualisation tasks and ATLAS integration

We would really like to get more people actively involved in these research areas so do get in touch with us if we need to add you to smaller focussed teams.

The recording and slides of today are available on the wiki page.

Thanks,

Peter

#22

For those that were not in our last PLP TC a quick note that there will not a PLP TC today.

The next PLP TC will be on Jan 11th and we will send out a new agenda invite to you all.

Have a great Christmas time and talk to you all in 2017!

Peter and Jenna

#23

Happy New Year to you all!!

Tomorrow we have the first PLP TC of the year in which we will discuss and demo the current version of the PLP package and we will ask those that have data to give it a try. We will definitely extend the package in the upcoming period with more algorithms, feature engineering steps and much more candy :smile: Hopefully you will contribute to that now we have made the foundation.

Furthermore, we will introduce a new validation study we are initiating to look at transportability of prediction models.

I have sent out new agenda invites, if you have not received one let me know. The gotomeeting details are the same as last year.

Note also that I like to start Eastern Hemisphere bi-weekly meetings starting next week to have our other friends join the PLP journey. I will fix the time for that with the people that join the Estimation TCs tomorrow morning. If you like to join these drop me a message.

Talk to you tomorrow.

Peter

#24

Thanks for the nice first Eastern Hemisphere meeting on Wednesday in which i introduced the workgroup goals, the work done in 2016, and the plans for 2017.

The recording and slides are added to the wiki page for those who are interested.

In the next Eastern Hemisphere meeting (Feb 1th) Seng Chan You will give an presentation on 'The experience of temporal analysis for predicting cardiovascular disease using National Sample Cohort-CDM in Korea ’ The National Sample Cohort data is composed of 1 million population from 2002-213, representing 2% of whole Korean population.

#25

Team,

Unfortunately, we have to cancel today’s Western Hemisphere meeting because I am travelling at that time to Estonia for a CDM mapping exercise.

Peter

#26

Just a reminder.

Tomorrow we will have the Eastern Hemisphere meeting (see the wiki for the connection details etc link)

Seng Chan You will give an presentation on 'The experience of temporal analysis for predicting cardiovascular disease using National Sample Cohort-CDM in Korea ’ The National Sample Cohort data is composed of 1 million population from 2002-213, representing 2% of whole Korean population.

#27

Thanks Seng Chan You for the very nice presentation of your work.

It was great to see a very clear clinical prediction question and all its challenges presented this way, e.g., dealing with unbalanced data, temporal data analysis, and optimisation under constraints.

The video and slides are added to the wiki page if anyone wants to join the fun!

1 Like
#28

Dear Western Hemisphere team,

Today in the PLP TC i like to discuss the upcoming collaborators face-to-face meeting March 17th link

In this face-to-face we are supposed to have a breakout session of one hour to discuss ongoing work etc all nice, but most importantly there is a hack-a-thon session planed on large scale statistical computing in which we like to work on the code base to increase the processing speed so we can scale up our recent depression proof-of-concept to the all-by-all challenge.

In my view this meeting is only useful and a succes if we can actually improve the code with the available people in our session, i would not like hours of presentations on possible solutions at that meeting. This however requires thinking to be done prior to the meeting next month.

In the TC today, I like to have a discussion on our options , who can contribute, how to prepare for this meeting etc.

Hope you will all be active in moving this discussion forward.

Peter

#29

Dear Eastern Hemisphere team,

Tomorrow Martijn will present the feature extraction package.

I will discuss some plans for extension of the package for temporal data analyses using deep learning.

Peter

#30

Thanks Eastern Hemisphere Team for the nice TC and Martijn for presenting the FeatureExtraction Package. Let’s work together to add temporal features to the package and do some Deep Learning on this.

The recording and slides are on the wiki page.

Peter

1 Like
#31

Hi all,

Tomorrow in the Western Hemisphere TC we will have a presentation from Shawn Dolley who is leading the Hadoop WG.

He will introduce Hadoop and we can discuss this in the context of our predictive work.

Peter

#32

Dear all,

The Eastern Hemisphere TC will be canceled tomorrow, because I am in a train in Spain at that moment. Moreover, there is a national Holliday in South Korea.

Peter

#33

Hi Western Hemisphere team,

Tomorrow Ken Jung will give an interesting overview of their prediction work at Stanford.

Connection details are on our wiki.

Peter

#34

Dear Eastern Hemisphere team,

Tomorrow in our TC i like to discuss the paper below in a bit more detail to get some feedback from the group.

https://arxiv.org/abs/1511.05942

I will meet Jimeng Sun (http://www.cc.gatech.edu/people/jimeng-sun) on Thursday and he was so kind to share some slides on this paper.

Peter

#35

The recording and slides of today’s Eastern Hemisphere meeting are on our wiki page.

Suggested topics for upcoming two Eastern Hemisphere meetings:

  1. Deep learning tutorial
  2. Discuss the proof of concept study

If you have other suggestions post them here.

Thanks

#36

Dear all,

Tomorrow in our Western Hemisphere we will update you all on the work done at the Hack-a-thon and the next steps.

It will probably be a 30 min TC.

Peter and Jenna

#37

Because of earlier Daylight Saving Time in US the TC will be 1 hour earlier today for Europe (CET 17:00) to keep it a 12:00 EST.

The agenda item for today has been updated to reflect this.

#38

The recording and slides of today’s meeting are on the wiki page.

Thanks all for joining.

Peter

#39

Dear All,

Tomorrow we have our Eastern Hemisphere PLP meeting.

Sanghyung Jin from Ajou University will give a talk on deep learning:

  • What is deep learning and why deep learning is powerful
  • Deep learning’s main architecture (CNN, RNN)
  • Issues with medical data and related work

Note, I have updated to agenda item sent to you earlier to compensate for
the day-light savings time change in Europe, thus the
time of the TC does not change for most of you.

Eastern Hemisphere time slot:

3 pm Hong Kong / Taiwan
4 pm South Korea
5.30 pm Adelaide
9 am Central European time
8 am UK time

As always we will record this session.

Peter

#40

Thanks Sanghyung Jin for the great introduction to deep learning.
It is a good starting point to start diving deep :slight_smile: into this topic.

The slides and the recording are on the wiki page for those that have missed it.

Peter

t