Back

An Evaluation of the MedSnap Medication Authentication System

Hymel, P. A.; Brossette, S. E.; Wong, D. Y.; Zheng, N.

2020-05-30 public and global health
10.1101/2020.05.06.20093427 medRxiv
Show abstract

Falsified and substandard (SF) medications are a large and growing global public health threat affecting developed and developing markets. The need for scalable, easy-to-use, accurate methods to detect SF medications continues to grow. We conducted a proof-of-principle laboratory and field sample evaluation of MedSnap in Vientiane, Lao PDR. Over three days, MedSnap models were created from trusted and authentic medications and tested against samples of authentic/trusted or falsified artesunate, artemether-lumefantrine, azithromycin, and ciprofloxacin. Across 48 tests of authentic/trusted and falsified samples, MedSnap was 100% sensitive and specific. On the final day of the study, a convenience sample of ciprofloxacin and azithromycin was collected from local pharmacies and tested. Each field sample matched an authentic/trusted model. The speed and ease of data collection, reference model building, and sample testing suggest scalable use cases for the system.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.