glossary¶

Glossary¶

chi¶

The model rate for excess mortality as a function of age and time. This is also denoted by \(\chi(a, t)\).

child jobs¶

If refit_split is false, there is one child job for each child node of the current fit_node .
If refit_split is true, and the current fit_node is not in the node_split_table , there is one child job for each child node of the current fit_node .
If refit_split is true, and the current fit_node is in the node_split_table , the child jobs have the same fit_node and there is one job for each split_reference_value that is not equal to the current job’s split_reference value.

dage¶

The abbreviation dage is used for differences of a function value with rest to age values. These are differences and not derivative approximations.

dtime¶

The abbreviation dtime is used for finite differences of a function value with rest to time values. These are differences and not derivative approximations.

eta¶

This is an offset in a log transformation; i.e., a mapping of the form \(x \rightarrow \log(x + \eta )\). Note that this transformation is finite valued at \(x\) equal to zero.

fit_goal_set¶

This is the set of nodes that we need a fit for. It must be a subset (which could be all) of the fit_goal_table .

fit_node¶

For a dismod_at database, fit_node is the parent node in the dismod_at option table.

fit_node_set¶

This is the set of nodes that need to be fit in order to fit the fit_goal_set. This includes all the nodes between the root node and the goal nodes (including the root node and he goal nodes).

fit_database¶

This is a dismod_at database with the following properties:

Constant Tables¶

see at_cascade.constant_table_list.

Option Table¶

The fit_database option table is the same as in the root_database except for the value of parent_node_name. This node is called the fit_node for the database. The value of parent_node_id does not appear in the option table.

Covariate Table¶

For each Relative Covariate, the covariate table reference value is the average returned by com_cov_reference for this node and split_reference_id. The split covariate reference value is one of the split_reference_value in the split_reference table. The other columns in the covariate table are the same as in the root_database.

Fixed Effects Smoothings¶

The age and time points corresponding to the smoothing for the parent rates (except omega) and the covariate multipliers are the same as in the root_database. In addition, the age and time difference prior corresponding to each age and time point is the same as in the root_database.

Random Effects Smoothing¶

The smoothing for the child rate effects (except omega) are the same as in the root_database.

Rate Table¶

The rate table smoothing for pini must be null or the corresponding smooth_grid entries must use the age_id corresponding to the smallest age in the age table.

Integrand Table¶

For the list of rates [ pini, iota, rho, chi ] the corresponding integrand names are [ prevalence, Sincidence, remission, mtexcess ]. If the parent_smooth_id in the rate table for one of these rates is not null, the corresponding integrand name must be in the integrand table. For each value covariate multiplier index mulcov_id, the integrand name mulcov_mulcov_id must be in the integrand table.

input_node_database¶

This is a fit_database with the following extra properties:

Input Tables¶

Only the dismod_at input tables are significant in a fit_databases; e.g., an init command should be executed before any other dismod_at commands (except possibly a set command).

nslist Table¶

The nslist and nslist_pair tables must be empty.

avgint Table¶

The avgint table must be empty.

Rate Table¶

The rate table must have null entries for the omega parent and child smoothings.

iota¶

The model rate for incidence, relative to the susceptible population, as a function of age and time. This is also denoted by \(\iota(a, t)\).

job¶

A job corresponds to one row of the job_table; i.e., one fit node and split_reference_id. The results for a job is its fit and the value priors for its child jobs ; see start_child_job_id and end_child_job_id.

meas_noise¶

The abbreviation used for measurement noise covariate multiplier.

mtall¶

The abbreviation used for all cause mortality data.

mtexcess¶

The abbreviation used for excess mortality data.

mtother¶

The abbreviation used for other cause mortality data.

mtspecific¶

The abbreviation used for cause specific mortality data.

node_name¶

Is the node_name for a node in the root_database node table.

node_split_set¶

This is the set of nodes at which the cascade is split by the value of the splitting covariate. No node in this set can be a descendant (or ancestor) of another node in this set. (Because the splitting can only occur once in the path from the root_node to any other node.)

ode_integrand¶

The following integrands requires solving the dismod_at ode to model and hence each of them is called an ode_integrand: susceptible, withC, prevalence, Tprevalence, mtspecific, mtall, mtstandard.

omega¶

The model rate for other cause mortality as a function of age and time. This is also denoted by \(\omega(a, t)\). Note that omega a a special rate because it is know ahead of time; see omega_all .

omega_grid¶

A single age-time grid used for the omega constraints.

pini¶

The initial prevalence as a function of time. If the initial age is zero, time is t , and prevalence is p, then pini is the function \(p(0,t)\) . This often referred to as a rate because it can be estimated and is an input to the dismod_at ordinary differential equation, like the other rates in the dismod_at model.

prevalence¶

This is the ratio the number of people with the condition divided by the number of people currently alive.

rate_value¶

The abbreviation used for rate value covariate multiplier.

Relative Covariate¶

A relative covariate is any covariate that is not an absolute covariate and is not the splitting covariate (which is specified by the split_covariate_name). The reference value for a relative covariate can depend on the node and the value of the splitting covariate.

remission¶

The abbreviation used for the remission integrand. This corresponds to a directly measurement of rho

rho¶

The mode rate for remission as a function of age an time. This is also denoted by \(\rho(a, t)\).

root_node¶

The top level node for this cascade; see root_node_name.

root_database¶

This is a dismod_at database where the root_node is the parent node in the dismod_at option table. It must satisfy all the conditions for an input_node_database plus the following:

The subgroup table must have one row; i.e., there is one subgroup, one group and they correspond to all the nodes.
The option table must not have an other_database or other_input_table entry.

Sincidence¶

The abbreviation used for the susceptible incidence integrand. This corresponds to a directly measurement of iota.

top_directory¶

Directory where the input data is located.