ShortageRadar turns medicine-shortage management from reactive into predictive. Today, national systems issue an alert only once a supplier reports a disruption. ShortageRadar applies a trained artificial neural network to historical supply data and scores each medicine-and-country pairing for shortage risk over a forward horizon — surfacing the medicines most likely to enter shortage while there is still time to act.

Methodology

The model was developed on World Health Organization Medicines Shortage Reporting System data spanning the United Kingdom, United States, Italy and Belgium — learning the patterns that historically preceded shortages: vulnerable drug categories and forms, recurring root causes, seasonality and cross-border signals. On held-out evaluation it achieved 97.91% accuracy with 1.00 recall on shortage events — meaning it missed none of the true shortages in the test data.

Provenance

Built on original MSc dissertation research — Predictive Big Data Analytics for the Prediction and Prevention of Medicine Shortages — conducted at the University of Bradford School of Management by Adeshina Akande, Flintweb Global Resources Ltd, supervised by Dr Emilia Vann Yaroson.

Current build is a research demonstrator scoring historical WHO MSRS data. Roadmap: live regulatory-feed ingestion (NHS, DHSC, FDA, AIFA, FAMHP) with continuous re-scoring and prospective validation. Independent product by Flintweb Global Resources Ltd; not affiliated with or endorsed by the NHS.