BRENNEN

Industries / Health / Healthcare

Pharmacy

Pharmacy supplies medicines to patients, on a prescriber's order or over the counter, checks that each one is right and safe for the person, and explains how to use it, in shops, hospitals and by mail.

c. 775

Early Arabic sources record privately owned pharmacy shops in Baghdad in the reign of Caliph al-Mahdi (775–85), which the historian Sami Hamarneh treats as the beginning of pharmacy's independence from medicine. He rejects the often-repeated date of 754, because Baghdad had not yet been built.

doi.org

49

Milestones achieved across 7 domains.

13

To watch for: announced, not yet confirmed outside.

15

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Drug sellers kept their own shops in Baghdad by the last quarter of the eighth century, and educated pharmacists are reported there about forty years later, by Sami Hamarneh's reading of the Arabic sources. In the Kingdom of Sicily, the laws of Emperor Frederick II made physician and pharmacist separate professions with separate duties; the German Pharmacy Museum dates them to 1231 and Hamarneh to 1240. Machines came late: Time reported in 1969 that US pharmacies and hospitals filled a billion prescriptions a year and still counted the pills by hand.

Where it stands

FIP, the International Pharmaceutical Federation, counts 5.57 million licensed or registered pharmacists in 83 countries that hold 71% of the world's people, and says 77% of pharmacists work in community pharmacies. In England alone, 1.26 billion prescription items were dispensed in the community in 2024/25 at a cost of £11.2 billion, by the NHS Business Services Authority's count. Counting machines, ward cabinets, pack-picking robots and central filling plants are in routine use, and the US Department of Veterans Affairs says its seven automated mail plants fill about 80% of its outpatient prescriptions.

What is open

Many tools that now read directions, hear drug names, count pills or flag stolen doses have been tested only by their makers or on private pharmacy records, so no one can compare them. The open jobs below are shared tests and public counts built from made-up prescriptions, public label, payment and prescribing data, water-filled syringes, dummy tablets and empty cartons. None of them dispenses a medicine or advises a patient, and each can be started this month.

Domains

AchievedWatch forOpen

Narrower industries

Specialty pharmacy

Advances to build with

1 research note

Open in pharmacy

3 of 15 open deliverables

Open · Filling prescriptions

A public label-direction test

A public set of at least 2,000 made-up prescriber directions (the "sig", such as "1 po bid x10d"), each paired with the label wording two pharmacists agree on and with its parts (dose, route, how often, how long) filled in, plus a scoring script. AI models and rule-based parsers are scored on how many directions they turn into a wrong dose, frequency or duration. No real prescription or patient record is used.

Built with what exists

RxNorm drug names from the US National Library of Medicine; openFDA labeling for real strengths and dose forms; ParseRx and parsigs as baselines; general AI models and open-weight models such as Mistral Large 4 (see the Note "A trillion open parameters"); small models that can be fine-tuned on one graphics chip, such as Liquid AI's LFM3 (see the Note "One GPU to fine-tune").

Credit · unclaimed

Open · Checking for errors and interactions

Sound-alike names, spoken

A public test of speech recognition on spoken prescriptions. Take name pairs from the ISMP List of Confused Drug Names, such as hydralazine and hydroxyzine, write short made-up phone-in orders for invented patients, record them in many voices and accents, and count how often each speech model writes the other drug of the pair.

Built with what exists

ISMP's list, updated through June 2024, and the FDA's tall man lettering list; OpenAI's open Whisper model run locally through whisper.cpp (see the Note "A first stable number for local AI"); a commercial model such as ElevenLabs Scribe v3 (see the Note "Fewer word errors"); text-to-speech voices and volunteer recordings; a scoring script that counts swaps to the paired name separately from other mistakes.

Credit · unclaimed

Open · Advising patients

A shared counseling test

A public set of at least 300 scripted counseling scenarios for new prescriptions, each with a made-up patient, a medicine and a checklist of what the pharmacist should cover, drawn from the points US Medicaid law lists (name, dose and route, special directions, severe side effects, self-monitoring, storage, refills, missed doses), plus an AI-played patient for each script. AI models and pharmacy students are scored against it. Only actors or software play patients.

Built with what exists

general AI models to play patients from the scripts; open-weight models for runs on local machines (see the Note "A trillion open parameters"); speech-to-text such as ElevenLabs Scribe v3 to transcribe spoken role-plays (see the Note "Fewer word errors"); the counseling points in 42 U.S.C. 1396r-8(g)(2)(A)(ii).

Credit · unclaimed

Where it sits

Broader

Healthcare