What I did for this project:
Open Brain Platform
Summary
The Blue Brain Project, a neuroscientific endeavor by EPFL, developed the Open Brain Platform application. The software aims to explore the brain, construct models on various scales, and simulate these models. This project was completed in collaboration with scientists and engineers specializing in different aspects of the brain.
Problem framing
Conducting experiments in neuroscience requires extensive knowledge of various aspects such as brain regions, electrical behavior, morphological types, etc. This process also requires various tools located on different platforms and significant computational power. Neuroscientists often create their own workflows by piecing together incompatible components. Furthermore, pre-built models are difficult to find and are rarely validated and reviewed by a broad community of scientists.
Role
For both Fusion and the Open Brain Platform, I have been doing some UX research, prototyping, product strategy, UX and UI design. In addition that, I participated to the frontend development of the app
Approach
The app brings together various neuroscience specialities, which required us to observe and interact with different scientists. This helped us understand their tools and daily processes. We gathered insights and feedback on the existing scientific tools and began prototyping for each group. After this initial research/discovery phase, we created higher fidelity prototypes based on the discussions during workshops and interviews. We organized additional one-on-one meetings to garner more feedback and began developing the application's core concept. This led us into a cycle of developing core features and testing them with a small group of users. Over time, we progressively improved and added more features to the app.
Challenge
The complexity of each brain region and domain of speciality per lab offered us a rich panorama of the neuroscientific world. In addition to that, the combination of model-building and simulation experiments gave us quite a contrasted user proto-persona set. We had to understand the requirements behind two kinds of neuroscience. However, they are both interconnected, as a scientist working on simulation experiments requires models to launch his simulations against. Both are time-consuming activities, require a different set of tools, but need to communicate with each other in some ways. In addition to that, scientists or laboratories work on different scales. Some might focus on a single neuron or ion channel, while others work on an entire region only. The coherence between these scales and the necessity to offer the same experience was a constant challenge for us.
Solution
The app mimics the real flow of a scientific laboratory. We have virtual labs gathering different sciences. Users can create projects in which they can: explore resources brain region by brain region; build different models from ion channels to the whole brain; and finally, simulate the models they have previously built. After pinning data in explore, building models, and saving experiments, they can produce scientific papers with the help of AI. This last step takes advantage of the knowledge graph produced through the project. As each scientist requires a significant amount of customization, the AI portion of paper writing is more of an assistant than a pure automatic generator of content.
The homepage of the Open Brain platform give a sneak peek to the capabilities of the neuroscientific platform.
First prototype for the Open Brain Platform. This centered the activities of building brain models, simulation and exploration around a single user.
The exploration part of Open Brain Platform gives access to biological and reconstructed resources through specific hierarchy. The first level is through the brain region all the way through morphological type. Then resources are structured around set such as electrophysiology, density or synapses.
Browse the experimental data, all the different data related to the current model and some generated during all simulation experiment. Go deeper into knowledge discovery with our literature search AI tool.
All the way through the different iterations, the core concept has always been around the features of exploring the resources, building models at different scales and simulating those models
Configuring and launching simulation experiments such as the whole brain generate an important amount of analysis and report. Many of those are visually based and can give access to high definition videos.
Explore include both an atlas based subsection but also a literature one. In the first case, each brain region give an important amount of information concerning each region and their children.
Neuroscience is such a vast domain of science. In itself, we have many scientist specialized in different brain regions. Conducting workshop to expose the interest of each participant was extremely helpful. Each team and members listen and enriched each other process by exposing problematic faced in their daily work. Quick prototyping give the ability to establish common ground and many different solutions
A constant feature in OBP, for every scale, is the morphological-electrophysiological model (ME model)). Each neuron can benefit for that. When facing the whole brain, this is a gigantic amount of data. Three dimensional projection empower the user to understand and manipulate the parameters of each of these ME model.
The most complex and time consuming process is the connectomic behaviour between region. In order to understand the intensity, the plasticity and other feature of connectivity, an interactive matrix working in a three dimensional fashion helps the user to navigate through all the brain region layers.
Single neuron simulation can be done isolated from the region in one of the first scale tiers. While in the whole brain model construction, the user can define specific behaviour for every morphological type.
In further iteration, we integrate the concept of virtual lab. This gives the user to gather is model building experiment, his simulation ones and the different elements he gather in explore. Having these in the same place opened the ability to have an AI-assisted paper writer function.
Each virtual lab can possess a certain amount of project in which the user can isolate the perimeter of his experiment. In this context the user will be able to work on different scale such as single neuron, ion channel, microcircuit and 6 others
At many point of the application, the user can launch analysis in order to validate his parameters. As simulation can be quite massive, we've empower the user by giving him some control over the quality
