MediRank: a lightweight, web-based pre-screening tool using Euclidean distance similarity

MediRank is a web-based pre-screening diagnosis support tool designed to assist healthcare providers in rapidly identifying probable diseases based on patient symptoms and basic clinical parameters. Targeted for use in primary care and rural settings, where laboratory resources and specialist access may be limited, MediRank enables early case prioritisation and informed referral decisions. The system applies Euclidean Distance-based similarity analysis to compare a patient’s symptom profile — including temperature, platelet count, white blood cell count, and selected clinical indicators — against a database of confirmed cases. The algorithm calculates the degree of similarity for each disease in the dataset, generating a ranked list of possible conditions from most to least likely. This process allows healthcare workers to quickly narrow down differential diagnoses before proceeding with confirmatory tests. MediRank’s novelty lies in its lightweight, explainable, and real-time analysis framework. Built with Python and deployed using Streamlit, the tool runs directly in a web browser without installation, requires minimal computational resources, and produces transparent outputs that include similarity scores, ranked diagnoses, and visual comparisons. The integrated PDF export function enables instant documentation for patient records or referral purposes. As a pre-screening tool, MediRank is not intended to replace medical judgment but to support early triaging, improve diagnostic efficiency, and provide educational value for medical students learning about case similarity analysis. Its dataset-agnostic design allows adaptation to multiple medical domains, and future developments will explore weighted similarity measures, hybrid algorithms, and integration with electronic health record (EHR) systems.