This month at Pistoia Alliance’s Centre of Excellence for AI/ML in Life Sciences was defined by the AI/ML Workshop at the Pistoia Alliance conference in Boston, MA on October 9th and10th and the start of the Datathon.
Boston AL and ML Workshop Summary:
- The Workshop was oversubscribed and with over 100 attendees from biotech, pharmaceutical & technology/data industries. With good representation covering larger Pharma, biotech, start-ups, tech/data and academic scientific community.
- Speakers and presentations in the workshop
- 5 full-length research presentations (slide decks with speakers’ permission),
- 3 highly engaging panel discussions, and two brainstorming events: one during the workshop itself, and another in the break-out meetings during the main Pistoia conference on the next day.
- Over 50 Pistoia conference attendees joined our break-out session on October 10th, making it even stronger and more diverse.
- The key topic highlighted in brainstorming and break-outs was data & data quality:
- Can we deploy AI solutions to characterize data quality?
- Can we standardize formats and normalize data sets for computation across disparate sources?
- Can we use AI to cleanse the data according to defined quality criteria and to enforce fairness and privacy in data sharing?
- On the other hand, can we create fully annotated computational models that would exemplify the FAIR principles in application to methods, and not just data alone? All of these is essential for reproducibility of research.
- The second key topic that emerged in our discussions was AI use in clinical trials. Can we use AI to predict hypo- and hyper- responders in human trials? Can we use AI to improve the translational animal models? Can we use AI to detect fraud at remote trial sites?
- The third topic mentioned in the break-outs and in panel discussion was development of methods for explaining the results of use of AI models for scientific and regulatory review, Model interpretability.
Pistoia Alliance AI/ML Workshop Summary
- Run in collaboration with Elsevier, Cures Within Reach& Mission-Curefrom October to December with judging from mid-January 2019
- Help cure rare diseases and win awards!
- Scientific challenges: predict likely drug candidates, any possible side effects (adverse events) and any drug-drug interactions that might arise from treating comorbidities
- Announced at the Boston Pistoia conference, registration open during October:
More information on Datathon and to register
- Upcoming talk: Maximizing Value from Healthcare Data Using Machine Learningin collaboration with the PRISME Forumon October 18th (re-scheduled from September) – sign up to attend
- Plans for the future webinars– welcome ideas for autumn series
- Catch up with recordingsof our earlier webinars
How you can contribute right now:
- Can you supply additional data mining, quality improvement, community development project ideas? Please visit our idea brainstorming center at https://ip3.pistoiaalliance.org/
- Projects listed there already include BioAssay Data Annotation and Imaging idea.
- Propose ideas for future webinars?
- Ideas & Planning for London 2019
- Volunteer to be a speaker or a discussion panel member at our next event in London in the spring of 2019? If so, please contact Vladimir Makarov by e-mail (email@example.com)