Back
MEDICAL AI RESEARCH • PUBLISHED WORK
PneumoDetect AI
Cross-operator validated deep learning model for pediatric pneumonia detection. Published research with production deployment and web interface.
Role
Lead Researcher & Full-Stack Developer
Duration
May 2024 - Jan 2025
Category
Medical AI Research
Medical AI research often stops at model validation, leaving clinicians without practical tools. Published models remain inaccessible despite proven performance.
PneumoDetect AI bridges this gap—a complete research-to-production system. Published in IJSET (Volume 13, Issue 5, 2025), the model achieves 96.4% sensitivity with rigorous cross-operator validation on 485 independent samples. Beyond academic validation, the system includes a production-ready web interface, REST API, and open science reproducibility package archived on Zenodo.
Platform Dashboard Screenshot
Published Research
IJSET Volume 13, Issue 5, 2025 with DOI: 10.5281/zenodo.17531598
96.4% Sensitivity
Catches 96 out of 100 pneumonia cases on independent test set
86% Cross-Operator Accuracy
Validated on time-separated dataset from different operators
Production Deployment
Live web application with 2.5-second inference time
Technology Stack
PythonTensorFlowResearchStreamlitFastAPIMedical ImagingBootstrap Statistics