ACR/EULAR 2010 RA classification (Aletaha Table 3). Required: clinical_synovitis true. Score: joint involvement 0/1/2/3/5 + serology 0/2/3 + abnormal CRP or ESR 1 + duration ≥6 weeks 1; max 10. classified false if another disease better explains synovitis; else true if typical RA erosions or score ≥6. Not a diagnosis. Not 1987 ACR RA. Not SDAI. magent does not examine joints or titer RF/ACPA.

Samuel 2002 pediatric appendicitis score from eight caller-assigned findings, ages 4-15, max 10. Cough/percussion/hopping RLQ and RIF tenderness score 2; anorexia, pyrexia, nausea/emesis, leukocytosis, neutrophilia, and migration score 1. Fever and labs are flags, not unpublished cutoffs. Not Alvarado (MANTRELS). Not a diagnosis or operating-room protocol. magent does not examine the abdomen.

Hohol 1995 Disease Steps: caller-assigned integer 0-6 (normal; mild disability, mild symptoms or signs; moderate disability, visible abnormality of gait; early cane, intermittent use of cane; late cane, cane-dependent; bilateral support; confined to wheelchair). Not 0.5 half-steps, not later Disease Steps Plus, and not EDSS. magent does not examine gait. Not a medical device and not an MS diagnosis.

van der Helm-van Mil 2007 Leiden prediction rule for recent-onset undifferentiated arthritis (Figure 1 / Table 2). Age × 0.02, female 1, small joints 0.5, symmetric 0.5, upper 1 or upper-and-lower 1.5, 100-mm morning-stiffness VAS 0/1/2, tender and swollen counts 0/0.5/1, CRP mg/L 0/0.5/1.5, RF 1, anti-CCP 2. Round to nearest 0.0 or 0.5. Bands ≤6 low, >6 and <8 indeterminate, ≥8 high. Not a diagnosis of RA. magent does not examine joints or assay serologies.

ESUS (Hart 2014). True iff all five caller-assigned flags hold: non-lacunar infarct on CT/MRI; no ≥50% stenosis supplying the ischaemia; no major-risk cardioembolic source; no other specific cause; and Panel 3 minimum workup (brain CT/MRI, 12-lead ECG, precordial echo, ≥24 h automated rhythm monitoring (not telemetry), extra- and intracranial arterial imaging). Incomplete workup is not ESUS. Not a stroke diagnosis, not TOAST cryptogenic, not anticoagulation. magent does not read imaging.

2024 McDonald typical CIS (Montalban 2025). mcdonald_ms iff typical_cis and no_better_explanation and: ≥2 attacks with ≥2 lesions; or ≥2 attacks with one lesion plus DIS; or CIS with ≥2 lesions plus MRI DIT, OCB, or kappa FLC; or CIS with one lesion plus DIS and MRI DIT, OCB, or kappa FLC. DIS = ≥2 of 5 topographies (optic nerve added). Kappa FLC interchangeable with OCB. Not 2017 McDonald. Not PPMS. Not RIS. magent does not read MRI or assay CSF. Not a diagnosis. Not a medical device.

Frontera 2006 modified Fisher CT grades 0-4 for subarachnoid hemorrhage as assigned from CT by the caller. Returns the published grade and description (0 no SAH or IVH; 1 focal or diffuse thin SAH, no IVH; 2 thin SAH with IVH; 3 thick SAH, no IVH; 4 thick SAH with IVH). Not Fisher 1980 original groups 1-4. Not a diagnosis; magent does not read the CT or invent a millimetre cutoff or vasospasm percentage.

Kothari 1996 ABC/2 ellipsoid ICH volume: (a_cm × b_cm × c_cm) / 2, rounded to one decimal milliliter. A is the largest diameter, B the perpendicular diameter, and C slice count times thickness; the caller supplies all three in cm. Volume estimate only; not the ICH Score. magent does not read imaging or choose A/B/C on the scan. Not a diagnosis.

2017 McDonald typical CIS (Thompson 2018). mcdonald_ms iff typical_cis and no_better_explanation and: ≥2 attacks with ≥2 lesions; or ≥2 attacks with one lesion plus DIS; or CIS with ≥2 lesions plus MRI DIT or CSF-specific OCB; or CIS with one lesion plus DIS and MRI DIT or CSF-specific OCB. 2017: OCB may substitute for DIT when DIS is met. Symptomatic and cortical lesions count. Not PPMS. Not 2010 McDonald. magent does not read MRI or CSF. Not a diagnosis. Not a medical device.

Greving 2014 PHASES 5-year rupture risk for an unruptured intracranial aneurysm. Caller assigns population (north_america_europe 0, japanese 3, finnish 5), hypertension (1), age >=70 (1), size mm (<7 0, 7.0-9.9 3, 10.0-19.9 6, >=20 10), earlier SAH (1), and site (ica 0, mca 2, aca_pcom_posterior 4). Returns points (max 22) and grouped 5-year risk percent. Not a treatment order or PHASES of care. magent does not assign ancestry or read imaging.

Fisher 1980 original CT groups 1-4 for subarachnoid hemorrhage as assigned from CT by the caller. Returns the published group and description. Group 3 (thick SAH) was the high-vasospasm group; group 4 is ICH/IVH without thick SAH, not a worse grade than 3. Not modified Fisher 2006. Not a diagnosis; magent does not read the CT.

Hunt and Hess 1968 SAH surgical-risk grade from a caller-assigned grade 1-5. Serious systemic disease (as assigned by the caller) places the patient in the next less favorable category, capped at 5. Deterministic published grade with citation. Not a diagnosis; not WFNS; not Hunt-Kosnik 1974 grade 0/1a.

PASCAL (PFO-Associated Stroke Causal Likelihood; Kent 2021 JAMA Table 1) from a caller-assigned RoPE total (0-10; magent does not recompute RoPE) plus large-shunt and atrial-septal-aneurysm flags. high_rope is RoPE ≥7 (not >7). high_risk_pfo is large shunt and/or ASA. probable if both, possible if exactly one, unlikely if neither. Not a closure order, not RoPE, not ESUS. Trial ages 18-60 are a limitation. magent does not read echocardiography.

Graham 2006 CTS-6 from six caller-assigned booleans with published discrete weights (median-territory numbness 3.5, nocturnal numbness 4, thenar atrophy or weakness 5, Phalen 5, two-point discrimination loss 4.5, Tinel 4; max 26). Score >5 is associated with ≈25% probability of CTS; score >12 with ≈80%. magent does not examine the patient. Deterministic published arithmetic with citation; not a CTS diagnosis, not an EMG order, not a Boston questionnaire, and not a medical device.

PLAN score (O'Donnell 2012 Arch Intern Med Table 4): preadmission dependence 1.5, cancer 1.5, CHF 1, atrial fibrillation 1, reduced LOC 5, age 1 point per decade (max 10), significant/total arm weakness 2, leg weakness 2, aphasia or neglect 1. Max 25. Bands <10, 10-15, >15. Not a tPA decision and not iScore. magent does not examine the patient.

Douglas 2013 AWOL: four caller-assigned yes/no flags (age ≥80, unable to spell WORLD backward, not fully oriented to place, nurse-rated illness moderately ill/severely ill/moribund), 1 point each, max 4. Combined-population observed delirium 2/4/14/20/64% at scores 0–4. Deterministic published score with citation; not a delirium diagnosis, not CAM-ICU, not 4AT, and not a medical device. magent does not interview the patient or collect an integer age.

PRE (Pittsburgh Response to Endovascular therapy) score: age + 2×NIHSS - 10×ASPECTS (Rangaraju 2015). Integer PRE, quartiles <0 / 0-24 / 25-49 / >=50, and recanalization-benefit strata <-24 / -24 to 49 / >=50. Not a thrombectomy recommendation. The caller assigns NIHSS and ASPECTS; magent does not assign NIHSS or read CT.

Traube 2014 CAPD: 8 caller-assigned integers 0-4. eye_contact, purposeful_actions, awareness, and communication are reverse-scored in that paper (never contributes 4; 0 is the never-delirium pole after that reverse key). Total 0-32. Screen-positive iff score ≥9. magent does not examine the child or invent a developmental-delay cutoff. Not a psychiatric diagnosis.

Farris 2017 brain metastasis velocity: bmv = new_brain_metastases / years_since_first_srs, 1 decimal. new_brain_metastases integer 0-200; years_since_first_srs 0.01-30. Low if BMV <4 (median OS 12.4 months after initial distant brain failure), intermediate 4-13 (8.2), high >13 (4.3). Does not floor BMV (NRG-BN009). Not time since last SRS. magent does not count metastases on imaging. Not a diagnosis. Not a medical device.

4AT (Bellelli 2014): alertness (normal 0, abnormal 4) + AMT4 mistake count 0/1/2 + attention 0/1/2 + acute change or fluctuation (false 0, true 4). Max 12. Band 0 unlikely, 1–3 possible cognitive impairment, ≥4 possible delirium. Not a delirium or dementia diagnosis. magent does not examine the patient. AMT4 is the caller-assigned mistake count only; this API does not collect age, date of birth, place, or year.