Hackathon Planning - European Conference 2016

Owned by David Proudlock (Unlicensed)
Last updated: Feb 10, 2017
4 min read


Hi Innovation Committee, everything here is up for debate / discussion and improvement!

Thanks for your support! (David)


Still in development, I will keep adding thoughts and ideas here...

Theme: 

Machine / Deep Learning for R&D Innovation


Purpose Statement

Why a hackathon? Why Deep Learning?

The hackathon allows us to bring a diverse mix of machine learning experts and Pistoia members together to test the potential of machine learning for life sciences. It will demonstrate that the Bio / Pharma / Life / Agri Sciences arena wishes to be at the forefront of this technology. Our intention is that the ideas developed over the session will help Pistoia Alliance shape a new collaborative project in this space for the benefit of our members.

Why bring your data / your challenge to the hackathon?

Hackathons bring new ways of looking at existing problems or challenges. If you bring your datasets to the event, you are setting the agenda to meet your current needs, by providing your data the teams will be doing work for your organisation, your science. It can also bring comms possibilities for your organisation.


We believe that an event can this will help you as a member of the Pistoia Alliance and / or as a participant in the hackathon because:


1.       Biopharma has a poor reputation in this space:

a.       Demonstrate that biopharma is keen to invest in the next generation  - in this particular instance the next generation of information scientists to support life science and healthcare R&D

2.       Biopharma is short of data scientists:

a.       Encouraging interaction between nascent big data informaticians to develop their own individual capabilities in the understanding and manipulation of biopharma RWD

b.       Opportunity for biopharma talent scouting (you might find the next bright young data scientist for your company)

3.       Biopharma is always looking for new ideas – not least in data management and analysis:

a.       Bringing ‘fresh’ and perhaps challenging ideas to seasoned RWD professionals

4.       Open Source software (e.g. R) is increasingly important to biopharma:

a.       Exploring and exploiting novel open source software tools in the biopharma RWD domain

5.       Pistoia Alliance is always seeking relevant new projects to pursue so:

a.       Identifying new opportunities for the Pistoia Alliance to exploit its unique, cross-company, open innovation platform to develop a project in the RWD space

6.       It provides an opportunity for networking and collaboration:

a.       We will be bringing in universities, machine learning experts into one space, you may see opportunities spring from the team you work with or the people you meet.



Objective:

Hackathon for Pistoia to bring ideas forward to show potential of deep learning for R&D in life sciences / healthcare?

Aim, demonstrate the potential of AI / machine learning for health ecosystem of the future. Use conference 1 hour to summarise findings / ideas.

Ideas:

Options:

  1. Pistoia educational event, they participate bringing data sets with them. We provide vendors / experts / platform(s) to demonstrate what they can do with datasets
  2. Show and tell, invite vendors to come in and build solutions with our datasets, end of day we vote for winning idea / vendor
  3. Classic hackathon:
    1. Open just to Pistoia members / data teams, plus bring in some guest experts
    2. Open to students / general AI communities + Pistoia members (could do something like 50:50 split)
  4. University Competition:
    1. Invite universities (European but predominantly UK (travel reasons)) to bring crack team for machine learning in R&D, we'll provide the data on the day. Series of Pistoia and non - Pistoia judges. 5 finalists selected. They pitch at the conference (Name "Universally Challenged", Pistoia President's Academia Challenge ???)

Audience / Attendees:

Traditional Hackathon:

Build teams on the day (or night before at social event)

  • University / students + Pistoia members - build teams on day
  • Pistoia members, their data teams

Sponsorship / advertising / Sales opportunity

  • Bring provider(s) who have the deep learning toolsets
  • Pistoia members bring data sets
  • Source external data sets

Pistoia members get to use the tools / learn from experts and see their ideas developed

Recruitment:

Could I build virtual teams and let them collaborate prior to the event (e.g. one week before, receive data, teams can start planning)?

Deep Learning Tools:

Open source / available


Amazon just released their AI platform: (Image recognition suite + "Lex" chatbot (Alexa API) - Carmen, amazon meeting...

http://tcrn.ch/2gGrQSj


Not sure if available, would need partnership / collaboration to bring the tools on board:

Could we engage Microsoft's cure cancer through computing group? (Richard)

Google / Amazon / Apple

Google might be best opportunity due to health focus


Data sets?

What open source data can we provide?

Clinical data from pharma:

https://www.clinicalstudydatarequest.com

http://www.gsk.com/en-gb/research/sharing-our-research/patient-level-data/

Health data (NHS, Dr Foster, other?)


John:

Lars Griefenberg, Abvie (potential dataset)

Cathy Critchlow (Amgen?)

Martin Leitch?


Data:

What about the ontology project? 

The chemical safety data

Address HELM....


Look up Real World Data story in IP3 (Merck)...


Challenges:

Terms of use for the data sets could be a challenge - don't forget the legal stuff... (TransMART) - some restrictions of use...

Access to Data:

We might need something like the Michael J Fox foundation legal agreement for each entrant to sign before participating:

http://www.ppmi-info.org/documents/ppmi-data-use-agreement.pdf


Format / Operational

Strawman:

Session runs the Monday before the conference. Suggested timings:

8am start (Breakfast provided)

Welcome

Admin / format for the day

8.30-9.30 Team creation

9.30 -12 First session

12-1 Pizza delivery!

1-6 Execution

6-7: Final hour, prepare presentations

7-8 Elevator pitches and awards...


Get some coloured over-jackets like Health and Safety ones for the judges / mentors. (Nick)


Question:

What to present at the conference, we'll have one hour at the conference to tell the attendees about what happened. Present the winners? Pitches from the top 3 or all of the teams (challenge is that if all hackathon members come to the conference session we probably break the numbers allowance of the venue)


Idea: each team does 90 second elevator pitch. All video's. At conference we repeat the audience vote use voxvote or similar

Risks:
Can you execute a new machine learning solution in a day?

Venue Requirements:

Large room for around 100 people.

Banquet style layout, it needs a lot of tables, preferably round for 5-10 people at a table.

Each table will need power sockets for all the laptops

High quality Wifi (is it fast and reliable? can it connect all of your participants? does it block any ports?)

Projector

A microphone, at least in large rooms

Accessible entrances and wheelchair-friendly seating space (and if there is a stage, check if it is accessible, if applicable)

Toilets!


Depending on who is coming, should we have a drinks / team creation night the evening before?


PR

Get someone to video the day and build a short item to showcase the day for the conference and the website

Utilise SPARK...


Prizes?

Vary it:

Assuming a team of 5-6 people.

1st and 2nd prizes

Winners:

Students:

Quality laptop / desktop / xbox / oculus rift? etc.

Pistoia members:

ipad

Runners Up:

Smaller gadget, something novel and interesting


If we did a university challenge..

€5000 to the winning faculty

xbox for each team member?