Medical AI / Vision

BreatheWish

AI-assisted chest radiograph screening and clinical triage platform featuring a DenseNet-121 classification backbone, Grad-CAM visual explainability, and a decoupled FastAPI clinical workflow service.

DomainMedical AI / Vision
RoleML Engineer
Timeline2026
Code AccessRepository
Why this project?

Emergency radiograph queues cause diagnostic bottlenecks for acute pulmonary consolidation. BreatheWish provides instant preliminary screening and interpretable saliency heatmaps to assist clinicians.

System Objective

Achieved 90.89% test accuracy and 0.9906 ROC-AUC with verified Grad-CAM saliency maps for clinical audibility.

End-to-end execution pipeline

01
Data Ingestion & Context

Document normalization, raw text extraction, and domain chunking

02
Embedding & Indexing

Dense vector representations via fastembed and FAISS vector indices

03
Retrieval & Grounding

Top-k semantic similarity matching with thresholded relevance verification

04
Inference & Response

Constrained LLM context injection with strict source citation formatting

Core technologies & frameworks

PyTorchDenseNet-121Grad-CAMFastAPINext.jsPostgreSQL
Impact & Validation

Open-source clinical prototype demonstrating interpretable deep learning triage with fail-closed safety contracts.

Next System Iteration

Integrated automated Grad-CAM explainability and asynchronous physician-patient communication channels.

Key engineering learnings

01

Trained and evaluated a DenseNet-121 architecture achieving 98.25% specificity on normal radiographs.

02

Engineered real-time Grad-CAM saliency heatmaps overlaid on medical DICOM/radiograph imagery.

03

Built a fail-closed API architecture with role-based access control across doctor and patient workflows.