summer school
Machine learning summer school in healthcare and biosciences
2nd - 9th July 2023
ETH Zürich, Zürich, Switzerland
Antonija Burcul, CEO & Founder of FOUR, writes:
Greetings visitor!
Our whole team is excited to be back in 2023 with the 2nd edition of our summer school!
Before you devour our MLSS ‘23 page, check out our participant’s experiences in our MLSS ‘22 testimonials section. I would also like to invite you to read through the lessons that we’ve learned in 2022, one of the 2023 website writing gems. In my view, we have the mandate to ask you to join us in a learning experience, only insofar we ourselves are willing to be transparent and honest about our own learning journey.
This year we are so delighted & grateful by all the incoming enthusiasm from the community about our project. This is why we had no choice but to grow, welcoming ⅓ more lecturers than the year before, while ensuring our atmosphere remains intimate, with personalised elements capable of catalysing intellectual and career leaps you aim to achieve by joining us.
Produced by Meltem Salb.
this is Bumblekite summer school
Bumblekite summer school is FOUR’s annual machine learning summer school (MLSS) in biosciences and healthcare.
MLSS aims to teach both data & software engineering skills and domain-specific knowledge, as well as skills in writing, teaching, project management and strategy development — equally important communication skills which we believe are crucial to making positive contributions in this field.
We are looking for:
last-year undergraduate students,
graduate students (MSc, PhD),
early career professionals with up to 5 years of experience
who are currently studying/working at the intersection of bio & computing and would love an opportunity to deepen their skills and knowledge within this exciting interdisciplinary area with the goal of becoming the decision-makers & leaders of tomorrow in the field that is rapidly reshaping the healthcare systems across the globe as we know them today.
visual exploration: the contrast
This year we are introducing our Bumblekite visual theme. We would love to invite you to be a part of it. You can join us, Meltem, our videographer & photographer and Frenci, our brand designer in interpreting our theme by having it in mind when you are taking your photos, creating videos, drawing doodles during our sessions & more.
We have been deeply inspired by the concept of contrast, the dichotomy between the human & the machine.
The essence of healthcare and biomedicine is intensely physical. In the very heart of healthcare lies the interaction of a human with our own body, our physicians, caregivers, family in the moments of care and decision making. This material crux has been enveloped with a digital layer of data and tools that hold many promises of enhancing our health and strengthening our healthcare systems. How do we depict this digital extension of ourselves and its interaction with the physical world?
Furthermore, as we physically gather in an ETH lecture room, though we are physically all together, a great proportion of our interactions are digital. We simultaneously occupy and share two spaces: a digital and a physical one. How do we visually capture the fullness of our experience together?
This theme is an hommage to the work of Andrew Bolton and his Manus x Machina: Fashion in an Age of Technology spring 2016 exhibition. In the book accompanying the exhibition he writes: “The mediator between the hand and the machine must be the heart.”. With the ambitious aim of our Bumblekite space embodying this mediator, the convening space for the intimate dialogue between the hands and the machines, we see our MLSS as a heartbeat, that, through our MLSS ‘23 cohort leaving the space at the end of our MLSS week and voyaging to dozens of countries across the globe, creates ripple beneficial effects across our ecosystem.
Frenci Sanna
lecturers
Meltem Salb
At Bumblekite, our priority is to expose our community of participants to a diverse network of top professionals engaged at the fourfronts (word pun intended) of machine learning, healthcare, and interdisciplinary collaboration, among others. Our invited lecturers are leaders and pioneers in their respective areas of expertise and, equally importantly, they are also amazing people.
Legend: L=lecture, T=tutorial, LC=leadership conversation series.
engineering keynote lectures
Judy Gichoya, assistant professor, Emory University, opening keynote (L+T+LC)
Christian Holz, assistant professor, ETH Zurich
Jonas Zierer, senior data scientist, Novartis
Lukas Widmer, associate director, statistical consultant, Novartis
Mireia Crispin, assistant professor, University of Cambridge; chief digital officer, 52North
tutorials: excellence in practice
Christoph Berns, head of machine learning engineering, Bayer (T+LC)
Julian Lechuga, research assistant, NYU Abu Dhabi
Charlotta Früchtenicht, senior data scientist, Roche
Gaetano Scebba, data scientist, Novartis
communication workshops & in practice sessions
Katie Kurz, global head, group internal communications & communications partner to group functions, Syngenta Group
Gregor Wills, head of public relations & communications, Allianz X
Mirna Šmidt, founder, Happiness Academy
leadership conversation series
Matthias Egger, professor, University of Bern; president, Swiss National Science Foundation, new financing mechanisms to fund innovation
Miriam Donaldson, ex-head of digital commercial transformation (HR), Novartis, building data science & machine learning teams
Sebastiano Caprara, head, digital medicine unit, University Hospital Balgrist, building data science & machine learning teams
Jelena Curcic, associate director & senior principal data scientist, Novartis; CEO, Gastric Imaging & Analysis, building data science & machine learning teams
Herko Coomans, international digital health coordinator, Dutch Ministry of Health, health data strategy & policy
Gian-Reto Grond, head of digital, Swiss Ministry of Health, health data strategy and policy
Jennifer Pougnet, head of data strategy, Roche, health data strategy and policy
Iris Shih, digital biomarker technical program manager, Roche
Alf Scotland, head of digital biomarkers research and development, Biogen
Claudia Mazzà, lead data scientist in gait & posture, Biogen
Valeria De Luca, associate director & senior principal data scientist, Novartis, venture capital & startups
Ralf Molitor, managing director, Helsana HealthInvest, venture capital & startups
Laura Locher, customer success manager, Microsoft, technology implementations in the hospital setting
Thomas Huggler, managing director, University Hospital Balgrist, technology implementations in the hospital setting
Antonija Burcul, CEO and founder, FOUR, building AI products in healthcare
schedule
The final schedule is available to view here (final update: July 3rd 2023), here as a pdf file and here with the accompanying lists of reading materials. It contains a whooping total of over 45 hours of sessions (excl. the social programme) out of which around 15h are dedicated to hands-on engineering work.
Meltem Salb
Social programme
Virtual get-to-know before the summer school
We are sure you are as excited as we are about getting to know so many great people! To shorten your waiting time we will give you the opportunity to meet each other via Slack. The Slack workspace will open 3 weeks before the start of the summer school. This activity will include a short introduction of yourself, if you are comfortable, as well as more informal Slack games.
A hike to Uetliberg
To ensure that you can experience the beauty of Zürich while networking with the other participants, we will organise a hiking trip to Uetliberg, which is ca. 4.2 km long and will take around 2 h. The starting point is easily accessible by tram. It will be led by a member of the summer school organising team.
A guided city tour
We hope you find some time to explore the beauty of Zürich while you’re visiting. An ideal starting point is the 90 min tour we have organised that will lead you through the heart of Zurich, led by a guide from Free Walk. The tour will include among others Paradeplatz, Fraumünster, Thermengasse, St. Peter’s church, Lindenhof, Lenin’s house and Grossmünster. The tour will start on Sunday, 2nd of July at 17:00. Please mark in the registration form if you would like to join the tour.
An evening in the heart of Zürich, at SPACES
Since some of you might not be able to attend the social programme on Sunday, we would like to give everyone the opportunity to get to know one another in a relaxed atmosphere outside of our day-to-day programme. We have teamed up with SPACES Seefeld to gather on their beautiful rooftop terrace on Monday evening, including alcohol-free drinks and snacks. We hope this enables you to share your first impressions of the summer school with your peers and to start forming long-lasting relationships.
A tour of the ETH main building
Besides its remarkable architecture, the ETH main building also has a lot to offer in terms of its rich history. To provide you with the opportunity to explore it without getting lost in the maze-like corridors ;), we have organised a campus tour for you on Thursday at 7 pm. The tour is led by two ETH students.
Bumblekite community awards
We have created the following awards to recognise the behaviours that contribute to creating a sense of community and/or general summer school atmosphere that make this MLSS the best summer school experience for all the participants and partners, including the organising team:
Best application - to award the best written application of MLSS ‘23
Biggest team player - to celebrate the act of giving, in the form of helping others, especially during the tutorials
Best question - to encourage thought-provoking and insightful questions that will keep us thinking long after the summer school ends
pre-MLSS technical onboarding
Technical onboarding is led by Dominik Steiner, customer engineer in healthcare & life science at Google Cloud.
day-to-day at Bumblekite
Meltem Salb
The school’s curriculum is designed to cover in depth pressing topics in machine learning, biosciences, and healthcare, while prioritizing practical applications of computing in healthcare, biomedical research and direct patient care.
Our goal is for each participant to be challenged, while receiving the personalized attention and support necessary to succeed in their professional aspirations.
Every dataset analysed, problem solved and a policy written - every session created and delivered is done through the lens of how that particular action helps the patient achieve their goals. Patients’ stories are at the heart of the school, creating a storyline through which we will journey together in the 8 days.
A single day at the school represents a data layer of a healthcare system that touches the patient life. We will start with a layer of data familiar to many clinical measurements that come from a visit to the hospital, emergency room, intensive care unit, a blood withdrawal, and move onto imaging, sensors, sequencing, multimodal and healthcare systems data.
A typical day at the school will include the following session types:
Meltem Salb
Engineering keynote lectures: leaders in the field provide an overview of their area(s) of expertise.
Meltem Salb
Tutorials: excellence in practice: guided exposure to e.g. relevant data sets followed by practical assignments.
The goal of the tutorials is to give participants a feeling for how a real-world data science & machine learning problem space looks like & how one navigates it. Every tutorial, in addition to being centered around the data type of the day, will also have a focus on a specific engineering step in the data/machine learning pipeline (e.g. feature design, model evaluation, results analysis).
Meltem Salb
Communication workshops: skills-building sessions designed to foster presentation of data products, data visualization, writing, project development and adjacent skills.
Communication in practice sessions: near-real time feedback discussions delivered by our communications-partner-in-residence leaders.
Ugh, not another communications-related session, why can’t they just leave us alone & let us spend more time coding? :) (paraphrased MLSS ‘22 feedback) We hear you. After all, how can one see & define in actionable detail the room for improvement for an activity so seamlessly performed on an everyday basis, such as communication with other people?
This is why, this year we’re experimenting with bringing on board our communications-partner-in residence leaders. Their tasks are the following: (1) observe you, our participants, while communicating “in the wild” during our summer school sessions: asking questions during Q&As, submitting feedback to our team at the end of the day & similar, (2) summarise the observed events & patterns and (3) present them back to you with suggestions for changes in your communication.
Meltem Salb
Office hours: opportunities for personalized advice from our lecturers and our team on participant’s own project or paper.
Office hours, as a space for a conversation whose structure & content is determined by the participant, are exactly that: room for free exploration, fueling not just the participant’s ascension in the career path of their choice, but also the next iteration of our machine learning summer school.
Meltem Salb
Leadership conversation series: talks, debates, and Q&A sessions with esteemed professionals who have pushed the boundaries of what is possible in their fields.
This conversation series was inspired by the Q&A sessions our team participated in, those conversations that no matter how long they last, are always way too short & irrespective of how many of your questions got answered, there was always this one that got away. We strongly believe that the people we have chosen to participate in this part of the curriculum have the potential to induce similar feelings in our participants. That being written, you will be the final judge of how good our curation skills were.
Why have we singled out the introductory panel on how leaders we admire have built & manage data science & engineering teams within the healthcare systems? We believe this panel will set the tone for the rest of the summer school, with the cumulative decades of experience, superb leadership skills and candidness of our panelists. What is the ideal portfolio of skills one needs to have at every step of their career, every transition e.g. from one field to another that they’re keen on making? Is there such a thing as “ideal”? Where is the room for upskilling at the job & what are the required skills one needs to bring to it? These are some of the questions we aim to tackle during the first evening of our MLSS.
Meltem Salb
Introduction to and the closing of the day: time to prepare for and reflect on the day that has passed.
Equally importantly to an excellent start of the day is the process of unwinding and closure at its end. We would love to hear your thoughts, insights, hear more about the challenges that arose, as well as to celebrate the wins of the day. We strongly believe that taking a pause, reflecting on the experience you went through during the intense day is an essential part of the process of acquiring new skills & knowledge.
partners
Bumblekite 2023 was generously supported by Google DeepMind and the Hasler Stiftung.
We are excited to be a part of Zürich’s vibrant research and business community & partner with local innovators, changemakers & businesses we admire. We believe one of the most important spaces to celebrate diversity is within our own budget: whether it is through food, legal documentation, videography or any other final product of their brilliant creative vision, we are thrilled to partner with, among others, a growing number of women-owned and managed organisations.
Acknowledgements
We would like to recognise and thank the following individuals who supported the creation of MLSS ‘23: Andrew deMello, Vedant Joshi, Shashank Shekhar, Hatem Nawar, Julián Lechuga López and Romana Burgess.
testimonials
What impressed me the most about the Bumblekite summer school, is the active crowd of participants. They engage in meaningful discussion and don't settle with the obvious answer. By digging deeper and deeper, they challenge the speakers and the audience to be specific and to be brave in voicing opinions. A really refreshing community.
- Thomas Huggler, managing director, University Hospital Balgrist, lecturer
This summer school is truly unique in its diversity of students, high level of engagement, and balance between academic curriculum and industry perspectives. As a lecturer, I loved the opportunity to connect with peers outside my organisation and in other health subdomains, which is a source of inspiration and is shaping the next steps in my career journey. I also appreciated the opportunity to give lectures to junior associates to support their leadership development and external presence. The evolution of the summer school has been impressive, with fantastic and practical programming, and a fun variety of formats such as hands-on activities, lectures, panel discussions, social events, and office hours. I can’t wait for the next edition to (re-)connect with the other lecturers and be energised by the participants’ curiosity!
- Valeria De Luca, associate director, data science, Novartis
Machine learning has become increasingly central in healthcare and the biosciences, spanning from diagnosing a disease to predicting outcomes and choosing the optimal treatment for individual patients. Furthermore, it has emerged as an exciting research instrument, one that we increasingly use in my group at the University of Bern but also at the Swiss National Science Foundation, for example, to match proposals with suitable reviewers and panel members.
The Bumblekite MLSS provides an outstanding introduction to the methodological underpinnings and various applications. I immensely enjoyed the discussions with students in one of the leadership conversations, in tandem with Miriam Donaldson, a highly accomplished industry leader.
- Matthias Egger, professor of epidemiology, University of Bern; president, Swiss National Science Foundation, lecturer
Machine Learning Summer School - A Transformative Experience
I want to take a moment to reflect on the incredible impact of the Bumblekite MLSS experience on my career journey. One of the most significant takeaways from MLSS was the people I met. The connections I made during those intense days of learning have been invaluable. Some of my fellow attendees have become long-lasting friends, and we're still in touch. Others, I've connected with on LinkedIn, expanding my professional network and forging connections that span the globe. Bumblekite MLSS wasn't just a summer school; it is a community that continues to enrich my life.
The knowledge I gained during MLSS gave me a profound overview of what the field of machine learning represents. It was like opening a door to a world of possibilities. The curriculum was thoughtfully designed to provide a well-rounded understanding of this rapidly evolving field and to hear about personal experiences from the best specialists in the field.
But it wasn't just theoretical knowledge. Bumblekite MLSS also gave me hands-on experience and skills that I could immediately apply. The practical sessions helped me develop a skill set that was not only relevant but in demand in the job market. The school was the catalyst that ignited my passion for machine learning, expanded my network, and equipped me with the skills and knowledge to thrive in a competitive job market.
I'm immensely grateful for this experience, the wonderful people I met, and the transformative impact it had on my professional journey.
- Olga Brovkina, participant
educational outcomes measurement
We have designed & built the MLSS curriculum around a set of key measurable learning outcomes. What does this mean in practice?
After completing the MLSS, each participant is aimed to be able to make a weighted decision whether a machine learning algorithm needs to be applied in order to solve a particular problem and if yes, which ML algorithm needs to be chosen, as well as elaborate on a set of reasons for and against the decision you are taking.
- Design a data project that includes ML end-to-end : define the steps, design each step, define the expected result.
- Efficiently communicate with people involved in the ML project team regardless of their domain background: explain what needs to be done and why.
- Efficiently present the results of the data & ML project to various partners.
To ensure these learning outcomes are achieved and to gather the data for future improvements of the Bumblekite MLSS, we have built a rigorous learning success evaluation system that includes:
- self-assessment questionnaires before and after the school for each participant,
- engineering tutorials group performance rate tracking,
- reflection as both a learning & an assessment activity within the core curriculum, in late afternoon hours.
Additionally, we are aiming to evaluate if & up to which extent were the expectations of each participant met throughout the duration of the MLSS. In order to achieve this goal, we will be asking you to fill out a feedback questionnaire and have feedback interviews with us after the school has ended.