DOSAGE: Personalized Antibiotic Medication Dataset

The DOSAGE dataset provides structured, clinically verified antibiotic dosing recommendations tailored to individual patient characteristics. It is designed for use in clinical decision support systems (CDSS), AI-driven auditing tools, and pharmacological research. DOSAGE integrates key variables including disease-specific regimens, standard dosing guidelines, renal function adjustments, and pregnancy-related risk classifications. The dataset supports evidence-based prescribing decisions across 98 commonly used generic antibiotics and has been reviewed and validated by practicing clinicians.<br>The dataset is organized into five primary CSV files:<br><b>d_dose.csv</b>Contains disease-specific dosing regimens. Doses are stratified by patient age, weight, disease condition, and route of administration (e.g., oral, intravenous, intramuscular). This table enables highly specific recommendations based on patient demographics and indication.<b>s_dose.csv</b>Provides standard dosing information without linking to any particular disease. Doses are still stratified by age, weight, and administration route, serving as a baseline reference for general antibiotic use.<b>r_dose.csv</b>Focuses on renal-adjusted dosing recommendations. Entries are stratified by creatinine clearance (CrCl), age, and sometimes disease, with optional flags indicating when a drug is not recommended for patients with impaired renal function.<b>preg_risk.csv</b>Lists FDA/WHO pregnancy risk categories for each antibiotic generic. This supports safer prescribing decisions for pregnant individuals by identifying drugs that are contraindicated or require caution.<b>generic_disease_map.csv</b>Acts as a supplementary index linking each generic antibiotic to the diseases it is used to treat. This file enables efficient querying and dataset integration when filtering or mapping drug-disease relationships.<br><br>Each dataset entry includes structured attributes such as:<br>generic: Antibiotic namedisease: Applicable infection (for disease-specific or renal tables)min_age / max_age: Age range expressed in days, months, or yearsmin_weight / max_weight: Weight range when dose-per-weight appliesmin_dd_mg / max_dd_mg: Direct daily dose range in milligramsmin_dw_mg / max_dw_mg: Dose per kilogram for weight-based regimenslim_mg / lim_iu: Maximum safe daily dose when applicablemin_crcl / max_crcl: Creatinine clearance range for renal dosingroute: Administration mode (PO, IV, IM)flag: Special instructions for renal adjustment (e.g., “not recommended”)r_category: Pregnancy risk class<br><br>To support implementation and reproducibility, two key support files are included:<br><b>README.md</b>This document explains the structure and purpose of each dataset file, provides definitions for all field names and units, and outlines the intended use in clinical and computational workflows.<b>dosage_demo.ipynb</b>A Jupyter Notebook that demonstrates how to filter the dataset based on patient-specific inputs such as age, weight, renal function, disease, and route. This practical example is intended to help users quickly integrate the dataset into research or application environments.The entire dataset is formatted in CSV and curated using Microsoft Excel for accessibility. It contains no personally identifiable information and focuses on standardized dosing logic derived from publicly available medical references. By organizing complex clinical guidance into structured data, DOSAGE enables safer prescribing, facilitates rule-based validation, and supports the development of next-generation digital health tools.