v1.2.4

Fixes

• Restricts tasks per child to one, preventing processing from drifting
• Makes population for syringe services test all PWID to better test if the feature is working.

v1.2.3

Fixes

• Restricts tasks per child to one, preventing processing from drifting
• Makes population for syringe services test all PWID to better test if the feature is working.

v1.2.2

v1.2.2

Fixes

• Floating point error on input parameters by rounding all incoming params to the nearest millionth
• Only mark agents as random trial treated if they are actually enrolled in PrEP

v1.2.1

v1.2.1

Fixes

• When an agent dies in HIVModel.die_and_replace, they are now replaced with an agent of the same drug_type - previously the drug_type was determined at random from the parameters on replacement.

doi_generate

This is a dummy release to link with zenodo for DOI generation

v1.2.0

v1.2.0

License and repository status

• License file updated to reflect GNU GPL v3
• Repository now public

Documentation

• Updates docstrings to the Google style
• Adds documentation and deploys github page at marshall-lab.github.io/TITAN
• Adds action for continuous deployment of docs on push to main

Features

• Adds location to the model, enabling scaling or override of parameters depending on agent's assigned location
• Agents now store their (location-specific) parameters
• Adds location edges, currently unused
• Standardized terminology and use of start times/stop times for agent attributes
• Moves creation of weights for agent initiation to a method

PrEP

• Allows array of PrEP models, enabling more flexible eligibility criteria
• Adds method to determine if agent is a man who has sex with men, per CDC criteria
• Removes unused "prep reason" feature
• Updates criteria for the CDC prep target
• Adds prep targets for PWID

v1.1.0

v1.1.0

Settings

• Scott County setting now accurately reflects the real-world setting in the time period from 2010-2020

Parameters

• Distributions defined in classes and can take a variable number of arguments
• Classes can now be used as validation items for other parameters
• See params app for parameter definitions and documentation

Partnering

• Agents can now attempt to partner with only the agents in their network component with params.partnership.network.same_component.prob

Features

• Parameters can now be scaled based on the timestep in the model with params.timeline_scaling
• HIV interaction and assignment can be set to start at any time step
• Reports can now aggregate by customizable classes with params.outputs.classes
• Population can be saved to files and also uploaded from files and re-used across runs
• Syringe services program may now be implemented flexibly in time
• Partner tracing now allows user-defined probabilities for tracing separately from the probability of PrEP initiation, as well as user-defined tracing period length
• prints_components report added fields for random trials. Now prints component status for no intervention, intervention components, and intervention components with no eligible agents status.
• the agent_zero criterion can now be defined based on number of partnerships that share a user-defined interaction type

Batch Submission

• --savepop and --poppath added as arguments to run_titan.py to respectively allow saving of a population after creation (but before run), or loading a saved population to run (the equivalent flags in subTitan.sh are -p and -P)

Technical Notes

• population added as an attribute of agent - lightly used currently
• Output writing abstracted/reorganized to make common report building steps re-usable
• Distributions can be any numpy distribution (just add to params.classes.distributions) or a distribution in distributions.py
• Added interaction type to edge list
• Renamed initialize_graph to trim_graph

v1.0.0

New

Settings

• All settings except Chicago, Scott County, and Atlanta have been deprecated
• Scott County is uncalibrated in its current form. It is meant to be used with a CSV calibration file

Parameters

• Parameters now given as .yml files separated by setting and intervention
• Injection transmission now given as flexible distribution instead of bins
• Race is now a definition, not restricted to black/white
• Sexual orientation, sex, and gender now user-defined, allowing cis/trans agents and combinations of possible sex partners
• Drug use (Inj, NonInj, None) now defined as percentage of race/orientation
• See params app for parameter definitions and documentation

Partnering

• Loop through all agents to find partners with customizable "break point"
• Allow customizable distributions for target partner number and duration, separated by bond types
• Bond types are now flexible and user-defined. Interactions are determined per bond type.

Features

• Agent 0 abstracted to allow user-defined bond type and number of bonds
• Condom use is definable by agent race, orientation/gender, and PWID status
• Death by risk factors now works as a scale on a base rate (defined on orientation, race, and PWID status)
• Syringe exchange overhauled to allow implementation and discontinuation and linear increase/decrease over time period
• Assortative mixing abstracted to allow variable definitions, based on any agent attribute

Batch Submission

• Parameters can now be swept as command-line inputs or given by CSV files
• Setting and intervention given separately as command-line inputs
• Distributed sweeps/runs

Technical Notes

• Now pypy compatible (removed scipy dependency)
• Agent sets now sets
• Integration tests split into stochastic and deterministic, stochastic allowed to fail
• Parameter error checking implemented
  * run_titan runs models in parallel if more than 1 core available (-c flag in batch script)
• Results folder can now be any arbitrary directory
• Various language updates to better reflect epi usage

v0.2.0

New And Improved

• HIV vaccination and boosters added
• Peer-change agent feature added
• improved testing

Settings

• Chicago created
• Deprecated locations deleted

v0.1.0

New And Improved

• PrEP allocation scenarios cleaned up, allocation by race added
• Vaccine added; allows for co-occurring PrEP and vaccines with boosters for HVTN702 and RV144
• Partial unit test coverage added

Fixes

• Removed hardcoded probabilities to a separate file
• Dead code removed
• Certain non-working features removed