Early prediction of helminth infection in small ruminants with accelerometers and machine learning

Abstract

This upload (PredictionOfDHealthInSR-master.zip) contains a Python-based research pipeline for early prediction of helminth infection / health status in small ruminants (goats and sheep) using accelerometer time‑series data and machine‑learning models. The code supports preprocessing of accelerometry signals, construction of training/validation datasets (including optional frequency‑domain representations such as wavelet-based features), and training/evaluation of classifiers to distinguish healthy vs. unhealthy animals. It also includes utilities for working with associated clinical/label data (e.g., FAMACHA scoring) and optional environmental covariates (weather variables). Where to find source code in the ZIP: - Top-level scripts: main.py (run to reproduce the full pipeline) and ml.py (command-line entry point for running the ML workflow with configurable options). - Core modules: pipeline/ (data preparation and training-set generation), preprocessing/ (signal preprocessing), model/ (model training/loading code), cwt/ (wavelet/CWT routines), utils/ (helper scripts), and dataset/ (scripts used to build/assemble datasets). Where to find data in the ZIP: - datasets/ contains the dataset files used by the pipeline (e.g., accelerometry-derived CSV/JSON datasets and label/metadata JSON files, including FAMACHA-related files under datasets/famacha/). - weather_data/ contains weather covariate CSV files (and weather_data/src/ contains code used to work with those files).