Flow in MIP Federation
MIP Federated brings together several hospitals and clinical data centers to build a Federation.
Data are collected at each hospital (clinical data) or clinical data centers (research data), then
a user of MIP can query and do machine learning or statistical analyses but with the strong imperative that
only data aggregates are exchanged between the hospital or data centers and the central node hosting the user-facing web site.
Flow of data in MIP Federation
graph BT
subgraph Hospital #1 databases
in_data(EHR database, PACS)
end
subgraph MIP Local at Hospital #1
subgraph Data Capture
dc_anon>Depersonalisation]
end
subgraph Data Factory
df_preprocessing>Pre-processing, feature extraction]
end
subgraph Algorithm Factory and Local database
hd_af(Features database and algorithms)
end
end
in_data -->|Patients with consent| dc_anon
dc_anon --> df_preprocessing
df_preprocessing -->hd_af
subgraph Hospital #2 databases
in2_data(EHR database, PACS)
end
subgraph MIP Local at Hospital #2
subgraph Data Capture
dc2_anon>Depersonalisation]
end
subgraph Data Factory
df2_preprocessing>Pre-processing, feature extraction]
end
subgraph Algorithm Factory and Local database
hd2_af(Features database and algorithms)
end
end
in2_data -->|Patients with consent| dc2_anon
dc2_anon --> df2_preprocessing
df2_preprocessing --> hd2_af
subgraph Central node
inc_data(Research cohorts)
end
subgraph MIP Local on central node
subgraph Data Capture
dcc_anon>Depersonalisation]
end
subgraph Data Factory
dfc_preprocessing>Pre-processing, feature extraction]
end
subgraph Algorithm Factory and Local database
hdc_af(Features database and algorithms)
end
end
inc_data -->|Subjects with consent| dcc_anon
dcc_anon --> dfc_preprocessing
dfc_preprocessing --> hdc_af
subgraph Federation layer
web(Web analysis)
dist_soft(Distributed database and algorithms)
web --- dist_soft
end
hd_af -->|Aggregates|dist_soft
hd2_af -->|Aggregates|dist_soft
hdc_af -->|Aggregates|dist_soft
user((User)) --- web