Enter the clinic manifest refraction and age. The engine reasons through
myopia vs hyperopia and age (presbyopia) separately — learned from real operated outcomes.
88 / 99%
myope sphere / cyl within one click*
15
nomogram signatures isolated
Nomogram Signatures — the surgeon-specific adjustments that turn a manifest
refraction into the programmed treatment. Each is graded by how strongly the evidence backs it.
Strong · well-confirmed
Moderate
Our data only
Diverges from convention
Contradicted by our data
Not yet testable here
Applied & well-supported
Presbyopia Sphere Hedge
Strong
In myopes, leave progressively less minus (toward slight plus) as age rises —
a buffer against age-related overcorrection. Applied only above ~45 yr.
Finding: corr(age, sphere-adjustment) = +0.41 in myopes; >50 yr averaged +0.17 D
vs −0.27 D under 40.
Hyperope Age-Neutrality
Strong
Do not apply the age hedge to hyperopes — treat sphere close to full regardless of age.
Finding: corr(age, sphere-adjustment) = +0.08 in hyperopes (essentially none).
Sphere −0.25 D Base Bias
Strong
A small baseline reduction of programmed sphere relative to the manifest value.
Finding: median sphere adjustment −0.25 D across the dataset.
Slight-Myopia Target
Moderate
Aim for a hair of residual myopia (~−0.1 D), not overcorrected plano.
Finding: mean recorded target −0.10 D.
Manifest-Refraction Ceiling
Strong
Don't over-tune sphere — the manifest refraction itself is the noise floor, so
adjustments finer than ~0.25 D are not meaningful.
Finding: ~0.2 D irreducible sphere scatter remains after modeling.
Backed by our data — our practice, not the convention
Cylinder Pullback Law
Diverges
Treat ~80% of measured cylinder, and a higher fraction of larger cylinders
(≈77% in myopes, ≈85% in hyperopes).
Finding: strong and consistent — corr(|cyl|, pullback) = +0.48; predicted to
within one click 99% of the time. Conventional practice treats cylinder in full, so this pullback
may under-correct astigmatism — the open question our post-op data must settle.
Gentler-Than-Factory Sphere Slope
Our data
Sphere is adjusted less aggressively than the generic factory nomogram —
about a −2% slope versus the typical −5%.
Finding: myopic sphere slope −0.023 per D — a milder correction than the factory curve.
Axis As Measured
Our data
Program the astigmatic axis exactly as refracted — no cyclotorsion or vector shift.
Finding: mean axis change ≈ 0°. A cyclotorsion compensation (~6°) could in
principle shave residual astigmatism — a refinement we do not currently apply.
Beliefs our data does NOT support
“Young → leave +0.50 hyperopic”
Contradicted
The common heuristic that young patients should be left slightly hyperopic.
Finding: the opposite — young myopes receive more minus (fuller /
slight over-correction), not left plus. Not present in these cases; would have to be imposed by rule.
“Pull cylinder back more in hyperopes”
Contradicted
The intuition that hyperopes need a larger cylinder pullback than myopes.
Finding: reversed — hyperopes are treated closer to full cyl (≈85%)
than myopes (≈77%).
“Higher myopia needs a bigger sphere adjustment”
Not seen
The intuition that greater myopia should get a larger correction.
Finding: high myopes (<−4 D) received ~0 sphere adjustment vs −0.09 D for low
myopes — if anything the reverse.
Not yet testable in our data
Steep-Cornea Retreatment Risk
Untested
Steeper preop corneas may carry higher retreatment risk → could warrant adjustment.
Finding: keratometry present in only ~23 eyes — insufficient to test.
Optical-Zone Overcorrection Interaction
Untested
Overcorrection may increase with both age and optical-zone diameter.
Finding: age effect confirmed; the optical-zone interaction is not yet analysed.
Cyclotorsion Axis Compensation
Untested
Shifting the treated axis by mean residual cyclotorsion (~6°) to reduce residual astigmatism.
Finding: not applied — no cyclotorsion capture in the current data.
Pachymetry & Preop-SE Predictors
Untested
Corneal thickness and preop spherical equivalent as inputs to refine the target.
Finding: awaits post-op outcomes to fit against.
These signatures update as post-op outcomes accumulate — divergent and untested items are the priority list for the Friday head-to-head.
* Cross-validated agreement with the operating surgeon's programmed values on eyes the
model had not seen. Segmented myope/hyperope engine fit on 717 EX500 eyes. Decision support only —
not a substitute for surgeon judgment, not medical advice. Internal Bimini AI tool.