BenchLink

What BenchLink does

A laboratory with its own bench runs five to twenty analysers, and every number they print is read off a screen and typed into a computer by somebody who is also doing three other things. BenchLink takes that job away. A small agent runs beside the instruments, speaks their own serial or network protocol, and posts what they say; the platform matches each result to an order by the barcode on the tube and files it through the ordinary lab path, so the reference range, the flag, the delta check and the acknowledgement queue all work exactly as they did. What it will not do is guess. A barcode nobody printed, a code nobody mapped, a test the order never asked for — each of those waits in a queue with the reason written on it until a person answers it. Quality-control runs go to Quality Control, and a stored result can never name both a patient’s order and a control material — the database itself refuses that row. The mapping between an instrument’s codes and your test master is the part that makes any of this mean something, and it is yours to fill in: nothing is pre-loaded, because a code that means glucose on one machine means something else on the next one along.

How you actually use it

1. 1Register the analyserIts name, make and model, whether it speaks ASTM or HL7 v2, and whether it is on a serial cable or the network. That gives you the id the edge agent needs.
2. 2Map its codes to your testsOpen the instrument’s interface specification, find the code it prints beside each analyte, and pair it with the test in your master. Nothing is pre-filled — the codes differ between two machines from the same maker.
3. 3Label the tube at the drawMarking a specimen collected now prints its accession number. That number is the barcode the analyser quotes back, and it is how a result finds its order.
4. 4Work the queue, not the keyboardMatched results file themselves. What is left is the queue: each line says why it is waiting, and you attach it to an order or throw it away with a reason and your name on it.

What it does not do

It never attaches a result to a record it cannot prove the result belongs to: an unmatched barcode, an unmapped code, a test the order did not ask for, a value the order already carries and an order whose report has been signed are all held for a person, and the same rules apply again when that person attaches one by hand. It will finish a panel the analyser reported one test at a time — an analyte nobody has a value for yet has never been read — but it will not change a number that already stands: that is an amendment, and amendments are made by people. It cannot put a control run on a patient — a stored line may name an order or a control material, never both. The host query answers with the tests ordered on a barcode and nothing about the patient: no name, no MRN, no date of birth. It ships with no analyser codes and no unit conversions pre-filled, and it does not talk to the instrument itself — that is the edge agent, which runs in your laboratory and holds no database connection and no keys of ours. It is the laboratory bench only: bedside monitors, BP monitors, glucometers, scales and every other point-of-care device are VitalStream’s, which is where DeviceHub was folded.

Questions clinics ask