THERAPICK: An Integrated Approach for ARTI Antibiotic Recommendations Using Rule-Based Logic, Tanimoto Similarity, and Multidimensional Data

Antibiotic resistance poses a critical global health crisis, exacerbated by inappropriate prescribing for Acute Respiratory Tract Infections (ARTI). This study presents THERAPICK, a system integrating clinical data (WHO Guidelines, CARD, ATC, KEGG, SMILES) to recommend antibiotics. A rule-based approach assesses antibiotic need from patient history and weighted symptoms, while filtering with ATC codes (J01) generates clinically relevant alternatives. Molecular similarity, calculated via the Tanimoto coefficient and validated with KEGG classification, identifies different pharmacological classes to mitigate cross-resistance risk. Unlike prior systems relying on culture-based testing, THERAPICK leverages molecular and pharmacological profiling for safer recommendations. The system successfully filtered 57.21% of irrelevant antibiotics from CARD, with structural analysis showing 85.6% of low-similarity alternatives (0.1-0.4 Tanimoto coefficient) belong to different pharmacological classes. Securely implemented with a FastAPI backend and web frontend, THERAPICK provides rational, data-driven recommendations, supporting safer and more efficient clinical decision-making for ARTI treatment.