Caitlin Chen

01  ·  AI / Retrieval

AI-Powered Medical Diagnosis

Addressing medical conditions with AI-powered diagnoses given symptoms, using academic research reports and FDA Adverse Event Reporting System data to augment real-time data with academic knowledge.

The system takes a set of presenting symptoms and returns candidate conditions, grounding each suggestion in two very different kinds of evidence: peer-reviewed academic research, and the FDA’s Adverse Event Reporting System.

The FDA’s Adverse Event Reports reflect rapid updating of events and cover a wider surface area of patients and medical symptoms compared to the slow, measured process of peer-reviewed academic research. Reading them together lets a real-time signal be checked against established knowledge before it is offered as a diagnosis.

Category
AI / Retrieval
Methods
Python, FAERS, Academic literature
A force-directed knowledge graph in blue: drug nodes linked to the adverse reactions reported alongside them, and on to the reported outcomes. Entities are drawn from the project's own FAERS extract; the layout is illustrative.
A force-directed knowledge graph in blue: drug nodes linked to the adverse reactions reported alongside them, and on to the reported outcomes. Entities are drawn from the project's own FAERS extract; the layout is illustrative.