Bedside infusion calculator
A bedside infusion tool that shows its arithmetic. You enter the patient's weight and the ordered dose; it shows you the equation, substitutes the numbers in front of you, and gives you a pump rate you can check by hand.
A clear line is both a patent IV and clear reasoning. Most dosing tools hand you a number and ask you to trust it — which means an error in the input produces a confident, wrong answer that nobody can spot. Clearline is built on the opposite assumption: the nurse should be able to see where the number came from, and the tool should refuse to proceed when the inputs do not make sense.
Infusion errors are rarely someone forgetting how to multiply. They come from the inputs — a weight that was never entered, a bag concentration that differs from the one assumed, a dose read in the wrong unit. So the design problem is not "calculate faster", it is make the inputs and the arithmetic visible enough that a wrong one looks wrong.
Vasopressors, heparin, insulin and vancomycin are all dosed per kilogram. A tool that never asks for weight cannot be checking anything — it is just scaling a number. Clearline makes weight the first required field and refuses to continue without it.
The same drug arrives in different bag strengths. Norepinephrine 4 mg/250 mL and 8 mg/250 mL look identical on the pole and differ by 2× at the pump. The concentration is shown as an editable input, not baked into a constant.
0.5 typed as 5 is a tenfold overdose that still looks like a plausible number. Clearline flags any dose that lands an order of magnitude off the usual starting dose, and names it as a suspected decimal error rather than a generic "out of range".
Safety bands come from each drug's own clinical range, not from where the slider happens to sit. A rate can be near the top of its slider and be perfectly routine, or barely off the bottom and be dangerous — only the drug knows which.
Four steps, gated: each one unlocks only when the previous is genuinely complete, and a blocked button tells you what is missing. Try entering a dose ten times the starting value in step 3 and watch what the tool does with it.
Every number above is computed from the inputs you entered — nothing is hard-coded. The drug ranges and concentrations are realistic and drawn from standard adult references, but this is a design concept: do not use it to give a patient anything.
Part of what makes bedside dosing confusing is that different drugs are ordered in genuinely different units, and each needs a different rearrangement to reach mL/hr. Clearline shows which form it is using rather than hiding all three behind one box.
Ordered in mcg/kg/min but pumps run in mL/hr, so the ×60 converting minutes to hours is where errors hide. Rate = dose × weight × 60 ÷ concentration.
Already ordered per hour, so no time conversion — but the concentration is in units/mL rather than mg/mL, and mixing the two up is a classic error. Rate = dose × weight ÷ concentration.
A total dose, not a rate — so the answer depends on how long you run it, and there is a hard ceiling (10 mg/min) below which you must stay. Rate = volume ÷ duration, with duration itself constrained by the dose.
The earlier version of this project described safety gating in its copy while letting you click straight to the monitor. Claimed constraints that do not exist are worse than none, because they train the user to believe the tool is watching. Here every step button is genuinely disabled until its precondition holds, and the blocked control states the missing condition in words.
When the dose is an order of magnitude off the usual starting value, the honest response is not to clamp it silently or render it in red and continue. Clearline names the specific suspicion — a decimal error — and blocks the start button, because that is the failure it is actually shaped to catch.
Bag strength is the input most likely to differ from what a tool assumes, and the one least likely to be questioned. Making it editable and deriving the concentration on screen costs one field and removes an entire class of silent error.
Not a clinical device, and not pretending to be one. Real deployment would mean pharmacy-maintained drug libraries, integration with the order and the pump, hard limits enforced in the pump firmware rather than the UI, and validation far beyond a portfolio piece. What this demonstrates is the interaction design and the argument for visible arithmetic.