Diabetes Bayesian Network: A Probabilistic Model with 413 Nodes

413 nodes and 602 arcs form a Bayesian network modeling diabetes-related variables. The model was created by S. Andreassen, R. Hovorka, J. Benn, K. G. Olesen, and E. R. Carson for a model-based approach to insulin adjustment, as presented in a 1991 conference proceedings. It contains 429,409 parameters and has an average Markov blanket size of 3.97.

Use Cases

• Benchmarking structure learning algorithms based on the network's 413 nodes and 602 arcs.
• Studying causal relationships in diabetes management based on the probabilistic model's parameters.
• Simulating patient scenarios for insulin adjustment based on the model's inference capabilities.
• Evaluating inference speed and accuracy in large discrete Bayesian networks.

Strengths

• The network structure is explicitly defined with 413 nodes and 602 arcs.
• Model complexity is quantified with 429,409 parameters.
• Network topology metrics are provided, including an average degree of 2.92 and a maximum in-degree of 2.

Limitations

• Column-level documentation is absent; variable semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for certain machine learning tasks.
• Last update date is unknown; freshness unverified.

Provenance

Source

bnlearn Bayesian Network Repository, referenced in the description.

Collection Method

Likely constructed from expert knowledge and data for modeling diabetes and insulin adjustment.

Time Range

null

Freshness

unknown

Geography

null