OHBM 2019

June 19, 2019

My first experience getting off the plane in Rome was running into Ross & Emily in the airport and we all run to get on the train together. Note to future self - leave more than 13 minutes to get to the train from the airport. Though I am proud of myself for how I operate in stressful situations, remembering to laugh and just go with the flow. My AirBnB was just a 5 minute walk from the Termini station. I walked 10 miles on the first day, mostly touring the Villa Bourghese gardens. Found a pair of nice sunglasses on a park bench, which seemed like karma since I lost my Ray Bans the last time I was in Italy hiking Cinque Terra. Wander around looking for pizza and miraculously manage to stay awake until 11pm with no naps during the day.

 

Hackathon begins! We are in the upper floors of Mercato Centrale which is within Termini Station. What an amazing space, many props to Katja and Roberto for their hard work organizing. Meet lots of internet friends in real life and reunite with many Neurohackademy friends. I decide to work on implementing regression into NeuroPredict package. I've been meaning to re-write our predictive model in Python, so this seems like a good opportunity to cross with off my "deep" to-do list. Hackathon is naturally fueled by espresso, carbonara, wine, and gelato. Another note to self - you are suppose to get 3 flavors of gelato at once! My favorite combo is hazelnut, carmel, pistachio.

 

 

I move to new airbnb near the Spanish steps on Via Margutta. Living room is big enough to hang a hammock in! There is also the most beautiful garden where I put my yoga mat.

 

Day 1: Educational courses. Deep Learning course is in the main auditorium. "Wow" moment to be on this stage. My talk went really well, and I got lots of questions. My preparation for this talk was focused on the experience I wanted the people in the audience to have, and less focused on the information I wanted to convey. I think this perspective made such a difference.

Opening reception where I talk to so many new and old friends - my networking skills are taken to the next level. End the day drinking a massive bottle of champagne (Thanks Todd) in Piazza Navona.

 

Day 2: Keynote by Dani Bassett (lots of Philosophy inspo). I had so many questions which I wrote down during her talk, but sadly because of time, no questions were asked at the end. Poster session where I presented our recent ABCD work predicting neuro-cognition using resting state. It was the first time that I had zero down time during a poster session. I talked to so many people about brain-behavior predictive modeling and accessing, processing, and analyzing the ABCD study data. Gael Varoquaux stopped by my poster and gave some excellent feedback: Always report MSE and variance explained along with correlations and we should share our pre-trained models for people who may not have access to ABCD data but want to test the model on other datasets. The next session I attended was on heterogeneity in psychiatry. Lots of good talks. The one that stood out most was by Andre Marquand from Donders. He happily took a skeptical perspective, raising many excellent points about clustering methods. He showed a scree plot where the components representing psychiatric dimensions were further down because they explain less variance than components for things like site/scanner/age/sex. This made me think about how we select components in our Brain Basis Set modeling, and how we many need to be more strategic in our selection.

Day 3: My presentations are done and over with. Now it is time to focus on everyone else's work, yay! I talked to so many awesome people in the poster session. Despite the heat, I was in the poster session for the full two hours. I spoke with: Vyacheslav Karolis (from Imperial College I think?) who is also working on fetal fMRI processing, Emma Bluemke from Oxford working on brain age prediction using 3D CNNs in UKBiobank data, Vlad Drobinin working on reliability of multimodal developmental data, Kristina Rapuano - cognitive control in ABCD data, a scientist at Columbia working on how attention systems help overcome visual deficits in schizophrenia (can't wait to talk more about this with Ivy Tso), a researcher at Harvard doing pre/post scanning during mindfulness training for patients with anxiety & depression, Kevin Anderson about genetics and schizophrenia risk, a scientist from Zurich about CCA, MEG brain networks, and behavior prediction, and many more! Thomas Yeo gave an incredible keynote on so much of his work on parcellation, behavior prediction, and dynamics. He is such a humble guy, and he made sure to give every student and post-doc credit for their work.

 

Day 4: Took the day off to recuperate, think about all I had learned so far, and do a few work things (but followed along on Twitter). I went out to dinner with a group of University of Michigan scientists. It was so much fun to connect with people from Ann Arbor in Rome, several of whom I had actually never met before! Thank you Scott & Robert for organizing. Hung out with friends drinking wine on the Spanish steps after dinner.

 

Day 5: Made it to the 8am session by Gael Varoquaux "Predictive models on functional connectivity: do we meet clinician's need" despite being out until 2am. It was well worth waking up early for. Roshan Cools gave a phenomenal keynote that made me think a lot about some contrasting results I’ve seen in our within-subject on/off methylphenidate data. The population neuroscience session was a really great way to end the conference. Lots of UKBiobank and ABCD results. These kinds of (large, open) datasets are the future! I had lunch in the hanging gardens, visited the brain art section, then ended by listening to my new friend play the piano. I walked home smiling and reflecting on how much I have grown as a scientist since my last OHBM two years ago in Vancouver. I love this community so much, and am so full of inspiration, gratitude, and joy. Now time to head to the Amalfi coast with my mother to relax in the sun and integrate all of these new experiences!

 

Share on Twitter
Please reload