EXPLORE
Explore the existing Hippocampus model and all its parts!
We are currently updating and improving this section. Please note that some information may be subject to change. We are working to ensure all data is up-to-date, and we appreciate your patience and understanding.
In this section, you can explore the different stages of our model build and simulation. We began by transforming sparse experimental datasets into the dense datasets necessary for building the model. Subsequently, each component of the model and the compound network model were subjected to rigorous validation, after which the model was used to make predictions. The models presented in this section are a continuation of the pioneering hippocampal research conducted during the 2014-2020 period as part of the Human Brain Project, which was published in Romani et al. (2024).
Experimental data
The first step of the reconstruction of the hippocampus involves the collection of experimental datasets from our affiliated and collaborators’ laboratories as well as from published literature sources worldwide. Available data on these brain regions are sparse and heterogeneous, and much of our work concerned the curation and organization of the data.
Reconstruction data
The second step of the reconstruction converts the sparse experimental datasets to dense datasets, required to fully reconstruct the hippocampus model. We applied several strategies to predict the missing data: algorithms, principles, and rules.
Digital reconstructions
The third step of the reconstruction generates an instance of the hippocampus model. Most of the model parameters were sampled from the distributions defined in the dense datasets of step two. For this reason, each model instance is unique within the biological ranges and could be considered to be derived from a specific individual.
Validations
We extensively validated each model component and the final network to assess the validity of the model. Successful validations support the idea that the model can go beyond the initial set of data and show emergent properties. Validation failures may indicate wrong model assumptions or incompatibility between model and validation datasets. Validation failures provide useful information for us to revise and improve the model.
Predictions
Each model parameter or behavior that has not been described experimentally could be considered a prediction. We already mentioned that most of the predicted parameters and behaviors arise during the reconstruction data step. Here, we focus on the predictions derived from the simulation experiments we carried out.