A data privacy and deep learning based AMR dashboard for rural and regional veterinary practices in Texas

Antimicrobial resistance (AMR) grows as a major worldwide health threat which affects treatment of both human and veterinary medicine. Practitioners working in rural and regional veterinary offices throughout Texas face difficulties obtaining real-time data tools which support making antimicrobial treatment decisions. This study introduces an AI-driven dashboard to address veterinary medicine needs by utilizing real-world AMR data collected over 14 years from veterinary labs throughout Texas. The dashboard employs deep learning models along with Long Short-Term Memory (LSTM) and Prophet with Recurrent Neural Networks (RNN) for prediction tasks and data imputation so practitioners can access insights utilizing visual elements such as heatmaps, Sankey plots, and MIC distributions and susceptibility tables. The dashboard empowers veterinarians with predictive analytics to perform empirical treatment selection and monitor resistance patterns to improve antimicrobial stewardship. Additionally, the dashboard integrates privacy-preserving fingerprinting techniques using steganographic marks, ensuring secure data sharing without compromising utility. Our novel approach addresses critical gaps in veterinary AMR data analysis, supporting antimicrobial stewardship and public health efforts through One Health frameworks. The findings demonstrate AI has proven its capacity to transform evidence-based veterinary medicine through data integrity and privacy.