Skip to content

MatteoPS/suicide_forecasting

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

140 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Repo Structure

.
├── config/            # Model configs, hyperparameters, YAML files
├── data/
│   ├── external/
│   ├── processed/
│   └── raw            # All data — this is gitignored
├── docs/              # Project documentation and data access instructions
├── notebooks/         # Exploratory and communicative work  
├── src/ 
│   ├── data           # Data import, cleaning, transform
│   ├── dynamic/       # Dynamic model of suicide contagion       
│   ├── ensemble/      # Multi-model combination logic; weighted averaging; scoring rules
│   ├── ml/            # LightGBM, RF, XGBoost, CatBoost, NN training and evaluation      
│   ├── nlp/           # distilBERT pipelines; NVDRS-RAD narrative processing
│   ├── statistical/   # Bayesian hierarchical models: Exponential Smoothing; ARIMA; CAR-ANOVA
│   └── utils/   
└── tests/           

Datasets

  • SEDD — ED attempt data; State Emergency Department Databases (HCUP)
  • SID — Inpatient attempt data; State Inpatient Databases (HCUP)
  • NVDRS-RAD — Violent death registry
  • Contextual — 76 contextual predictors: assembled from 26 data sources across 4 domains (demographic/environmental, socioeconomic/structural, social/community, individual-level risk factors); includes Census, BRFSS, and other public-use sources
  • Narrative — NVDRS-RAD qualitative text narratives from original medical examiner, coroner, and law enforcement reports; provide critical context

About

An ensemble statistical, machine learning and dynamical forecasting pipeline for proactive suicide cluster detection using spatiotemporal healthcare data

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors