Analysis code and results for "Budget-dependent selection of risk models under measured embedded deployment cost in two cardiac test beds
Year: 2026
Type: software
Abstract: Analysis code, fold assignments, measured deployment footprint records, simulation outputs and result tables for the manuscript "Budget-dependent selection of risk models under measured embedded deployment cost in two cardiac test beds". The archive reproduces every number reported in the article. Two test beds are carried separately and are never pooled onto one axis. The first is the Cleveland cohort of the UCI heart disease collection, 303 patients and 19 arms. The second is PTB-XL 1.0.3, 21799 records from 18869 patients, myocardial infarction superclass and 6 arms. Every arm was rebuilt as a deployable artefact and measured on a Raspberry Pi 3 Model B Rev 1.2 for stored size, resident memory and single-record latency. The archive also holds the protocol audit that precedes the measurement, the graph branch and hybrid arms, the paired tests behind the two null results, and the two simulation grids that ask what effect size the protocol could have detected. Scripts are numbered in execution order and package versions are pinned. See README.md for the run order and for what each output directory holds. MANIFEST.sha256 lists every file with its SHA-256 digest. The Cleveland input files are the four centre files of the UCI Heart Disease collection (https://doi.org/10.24432/C52P4X), redistributed here under CC BY 4.0. PTB-XL is about 4 GB and is not redistributed; data/ptbxl/NOTE.md gives the download and integrity check. The archive also contains data/kaggle_heart_1025.csv, a 1025 row redistribution of the Cleveland cohort that is audited in the work. It was retrieved from https://github.com/GehadGad/Heart-disease-dataset/raw/main/heart.csv and has md5 checksum 6481ed154d5b8ff83d7c09007aaa6ed0. The fitted model artefacts are omitted to keep the archive small and are regenerated by the scripts named in README.md. No power or energy measurement was performed and no energy figure is reported anywhere in the article.
Institution: Bolu Abant İzzet Baysal University
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7203675383