Lulytica
iOS + Android · Ships winter 2026

A private logbook for ERCP

Do you know your cannulation rate?

Not a guess. The number, from your own cases, against the standards that matter. Sixty seconds to log a procedure. Ten seconds to see where you stand.

Trainee to consultant No account Works offline
In training

Your training is happening case by case. Your evidence should accumulate the same way.

In practice

You perform hundreds of ERCPs every year. You should know what they add up to.

01 · How this started

Version one was a website I built for myself.

I ran it on a machine at home, used it on my own cases, and took it to Milan and Istanbul to show colleagues.

It taught me two things and warned me of a third. It needed a connection, and the endoscopy unit is the one place that does not reliably have one. It needed an account, and signing in to a server to reach your own cases is friction at the worst possible moment. And if I had ever opened it up to anyone else, I would have become the keeper of their records. That is a job nobody should have to trust me with.

So version two is a phone app. It works in flight mode. Your thumb opens it, and nothing is checked against anyone. The record never leaves your hands.

Presented at ESGE 2026 in Milan and IAP 2026 in Istanbul.

02 · The first thing you read
The measures screen: cannulation at 92 percent, 138 of 150 cases, 95 percent confidence interval 87 to 95, marked below target; stone clearance 96 percent and subhilar stenting 97 percent at target; post-ERCP pancreatitis 4 percent at target; antibiotic prophylaxis 89 percent, below the minimum.
The measures screen, rendered from the app.

Every rate carries its denominator and its confidence interval. It tells you when it doesn’t know.

Five measures, one screen, ten seconds. A rate that clears its target says so; one that sits between the minimum and the target says that instead; one below the minimum says that plainly. Nothing is rounded in your favour and nothing is hidden because it is uncomfortable.

6 of 20 cases · still building

A gauge stays quiet until it has twenty cases behind it. Until then it says still building and nothing more. A small number pretending to be a rate helps nobody.

Wilson 95% CI · shown, not hidden

11 of 12 65% to 99% 80 of 87 84% to 96%

The same rate, twice. At twelve cases the interval is wide enough to be compatible with excellent practice and with poor practice at once. At eighty-seven it is a rate you can stand behind. That gap is why the app holds its measures quiet until twenty.

03 · Standing at the workstation
The procedure entry screen: quick presets, the when-and-who chips, papilla history, biliary cannulation and the treatment section, all answered by tapping chips.
The entry screen, rendered from the app.

Sixty seconds, from memory, one thumb.

Chips, not typing. No free text, no dictation. Everything is answerable right after the case. No timestamps, no console readings, no chart lookup.

Your own shortlist. The indication picker learns from your last ninety days: recent picks first, then your top five. All 27 stay one tab away.

Save walks you to the gaps. Save is never greyed out. Tap it with something missing and it scrolls you to the first gap and asks. Half-finished entries survive as drafts.

The save bar reading zero of seven complete, with a plus similar button and a save button.

Nothing here needs looking up, and everything recomputes the moment you save.

04 · The standards that matter

Five measures, exactly as the ESGE wrote them.

Deep biliary cannulation

of the native papilla, with standard access

≥90%Minimum · target 95%

Stone clearance

for common bile duct stones under 10 mm

≥90%Minimum · target 95%

Stent placement

for biliary obstruction below the hilum, after successful cannulation

≥95%Minimum · target 95%

Post‑ERCP pancreatitis①②

every event graded for severity

<10%Limit · target <5%

Antibiotic prophylaxis

appropriate to the indication, before the procedure

≥90%Minimum · target 95%

05 · Training to independence

The yardstick moves with your career. The data doesn’t.

Two different questions need two different instruments, so there are two, and the app never confuses them. One asks whether you have got there yet. The other asks whether you are still there.

LC‑CUSUM · while you train

Have I got there yet?

DECISION LIMIT

It accumulates evidence towards a threshold and signals once, when the run of work is good enough to be unlikely to be luck. The ESGE training curriculum’s gates sit behind it, and a glide path shows supervised work giving way to independent work. A good week is not competence.

Performance CUSUM · after that

Am I still there?

ALARM LIMIT

This one never finishes. It is held to a floor at zero and climbs only when practice drifts, so past excellence cannot bank against a present decline. Cross a career stage and your history crosses with you.

06 · On paper, when it counts

Walk in with a document, not an impression.

Generated on your device, from your own record. It goes where you send it and nowhere else.

While you are training

The training evidence dossier cover page, listing eight sections from procedure volume to methods and sources.

The dossier

The end of it. Every gate, every form, in one document.

Every year after

The performance report, showing the five performance measures with their denominators and confidence intervals.

The report

A year of practice, ready for appraisal. Cited, dated, and signed by you.

Once training closes, one document does the work of both: the year, the measures, the complications, the denominators, with the citations attached, so the conversation is about your practice rather than about where the numbers came from.

07 · Behind the number

Read your rate against the cases you were given.

Five measures are the headline. Underneath each one is the question someone asks next, and the arithmetic is already done.

Were the cases harder?

Your cannulation rate again, split by the predicted difficulty of the case. A grade with fewer than twenty cases behind it stays quiet, and no grade is given a target of its own, because no paper publishes one.

The cannulation-by-grade card: grade 1, 61 of 62 at 98 percent; grade 2, 51 of 55 at 93 percent; grade 3, 19 of 22 at 86 percent; grade 4 building at 11 of 20. It closes by saying no per-grade standard exists.
Was the pancreatitis avoidable?

The pancreatitis rate split by recorded risk, with the definition of high risk printed beside it. Each stratum keeps its own count, and the published ceiling still judges the whole cohort.

The pancreatitis-by-risk card: 42 of 150 reviewed cases were high-risk; 5 of 42 developed pancreatitis at 12 percent, against 1 of 108 at 1 percent in the rest. It closes by calling the strata context, not separate standards.
Would a hundred cases stand up?

The ESGE measures over at least a hundred consecutive procedures. The app counts the last hundred, names what is missing from them, and lists the cases you can still complete.

The audit-readiness card: over the last 100 cases, four without rectal-NSAID status, two with an event awaiting grading, one graded but not signed off, and seven cases you can still complete.

Three cards from the Insights screen, rendered from the app.

The context sits beside the number. It never adjusts it. A harder case mix explains a rate. It does not move the standard you are held to.

08 · Yours alone

There is no server to breach. Because there isn’t one.

There are no patient names and no hospital numbers in it. The measures never needed them. Your record is encrypted on your phone and opens with your thumb. It works in flight mode.

And the phone you lose

A logbook that lives on one handset is worthless the day the handset goes. So there is an encrypted backup you hold and restore yourself. Your key, your destination, and no account in between. Local‑only is the design, not an excuse.

09 · Built to be checked

You shouldn’t have to take my word for it.

25 source papers Every threshold traces to a sentence in one of them, quoted rather than paraphrased, one tap away behind the number.
Where no paper settles it The app says so, and the choice is written down with its reason rather than buried.

The references on this page

Performance measures for ERCP and endoscopic ultrasound: a European Society of Gastrointestinal Endoscopy quality improvement initiative
Domagk D, et al. · Endoscopy · 2018
Severity grading, one scheme per event: the revised Atlanta classification for pancreatitis, the 2010 ASGE lexicon for bleeding and perforation, Stapfer for perforation type, and the Tokyo Guidelines for cholangitis
Banks PA 2013 · Cotton PB 2010 · Stapfer M 2000 · Kiriyama S 2018, each endorsed for this use by Dumonceau J‑M, et al. 2020
Curriculum for ERCP and endoscopic ultrasound training: European Society of Gastrointestinal Endoscopy position statement
Johnson G, et al. · Endoscopy · 2021
A scale of predicted difficulty for ERCP, which the continuous monitors are anchored to; their monitoring parameters are this app’s own convention
Cotton PB, et al. · Gastrointestinal Endoscopy · 2011
Probable inference, the law of succession, and statistical inference, the interval used for every rate in the app
Wilson EB · Journal of the American Statistical Association · 1927
ERCP‑related adverse events: European Society of Gastrointestinal Endoscopy Guideline, whose two‑tier risk stratum splits the pancreatitis rate on this page
Dumonceau J‑M, et al. · Endoscopy · 2020

The logbook is yours.

I am Lukas Erhart, an interventional endoscopist in Austria. I log my own ERCPs with this, which makes me user number one and the most impatient person it has.

Built after hours, one release at a time.

Tell me when it ships

Ships winter 2026 · one address, nothing else

Lulytica © 2026 Lukas Erhart · Austria · Imprint · Privacy