Skip to content

Neurology Intelligence Agent -- Learning Guide: Advanced Topics

Version: 1.0.0 Date: 2026-03-22 Author: Adam Jones


Purpose

This advanced guide covers the specialized clinical frameworks, trial criteria, assessment methodologies, and pattern recognition skills implemented within the Neurology Intelligence Agent. It is intended for those who have completed the Foundations guide and want to understand the clinical reasoning encoded in the agent's workflows and scale calculators.


1. The ATN Biomarker Framework in Detail

1.1 Biological Definition of Alzheimer's Disease

The 2018 NIA-AA Research Framework fundamentally shifted Alzheimer's from a clinical diagnosis to a biological one. AD is now defined by the presence of biomarkers, not symptoms alone.

1.2 Biomarker Modalities

Marker Fluid-Based Imaging-Based Cut-offs (Approximate)
A (Amyloid) CSF Abeta42 < 500 pg/mL; Abeta42/40 ratio < 0.058 Amyloid PET (Florbetapir, Florbetaben, Flutemetamol): visual read positive Plasma p-tau217 emerging as blood-based screen
T (Tau) CSF p-tau181 > 27 pg/mL; p-tau217 > 0.38 pg/mL Tau PET (Flortaucipir): Braak stage I-VI uptake pattern Regional tau correlates with symptom domain
N (Neurodegeneration) CSF total tau > 400 pg/mL; serum NfL > 19 pg/mL FDG-PET (temporo-parietal hypometabolism); MRI volumetrics (hippocampal volume) NfL is non-specific (elevated in ALS, MS, stroke)

1.3 Clinical Staging Integration

ATN Profile Biological Interpretation Clinical Implication
A-T-N- Normal AD biomarkers No AD pathology; evaluate other causes
A+T-N- Alzheimer's pathologic change Preclinical AD; biomarker-only stage
A+T+N- AD (early biological) Biological AD; may be cognitively normal or MCI
A+T+N+ AD (full biological) Full AD continuum; correlates with clinical dementia
A+T-N+ Alzheimer's + non-tau neurodegeneration AD amyloid with concomitant pathology (vascular, Lewy body)
A-T+N- Non-AD pathologic change Primary tauopathy (PSP, CBD, FTD-tau); SNAP
A-T-N+ Non-AD neurodegeneration Vascular, Lewy body, hippocampal sclerosis
A-T+N+ Non-AD with neurodegeneration Primary tauopathy with advanced disease

1.4 Anti-Amyloid Therapy Eligibility

Current eligibility criteria (based on CLARITY AD and TRAILBLAZER-ALZ 2): - Age 50-90 (some variation by trial) - MoCA typically 18-26 (MCI to mild dementia) - Amyloid-positive on PET or CSF - No more than 4 microhemorrhages on MRI - No macrohemorrhage history - APOE e4/e4 carriers have higher ARIA risk -- requires informed consent

1.5 ARIA Monitoring Protocol

  • Baseline MRI before treatment initiation
  • MRI at weeks 7, 14, 52 (minimum for lecanemab)
  • ARIA-E (edema): Usually asymptomatic; hold treatment until resolved
  • ARIA-H (hemorrhage): Microhemorrhages may or may not require treatment hold
  • Symptomatic ARIA: Hold treatment, close follow-up

2. Extended Stroke Treatment Windows: DAWN and DEFUSE-3

2.1 The Paradigm Shift

Before 2018, mechanical thrombectomy was limited to 6 hours from symptom onset. DAWN and DEFUSE-3 trials extended the window to 24 hours using imaging selection, saving thousands of additional patients annually.

2.2 DAWN Trial Criteria (6-24 hours)

Key principle: Clinical-imaging mismatch -- the clinical deficit is disproportionately worse than the infarct core size.

Patient Group NIHSS Requirement Infarct Core Limit
Age >= 80 NIHSS >= 10 Core volume < 21 mL
Age < 80, moderate deficit NIHSS >= 10 Core volume < 31 mL
Age < 80, severe deficit NIHSS >= 20 Core volume 31-51 mL

Additional requirements: - ICA or M1 MCA occlusion - Pre-stroke mRS 0-1 - Core volume measured by CTP or DWI-MRI

Results: 49% vs 13% functional independence (mRS 0-2) at 90 days. NNT = 2.8.

2.3 DEFUSE-3 Trial Criteria (6-16 hours)

Key principle: Perfusion-diffusion mismatch -- significant salvageable tissue (penumbra) exists beyond the infarct core.

Criterion Requirement
Vessel occlusion ICA or M1 MCA
Age 18-90
NIHSS >= 6
Pre-stroke mRS 0-2
Infarct core volume < 70 mL
Mismatch ratio >= 1.8
Mismatch volume >= 15 mL
Tmax > 6 seconds Defines penumbra on CTP/MR perfusion

Results: 45% vs 17% functional independence. Mismatch volume predicted treatment benefit.

2.4 Practical Integration in the Agent

The stroke triage workflow evaluates: 1. Time from last known well 2. NIHSS score (calculated or provided) 3. ASPECTS (from CT) or core volume (from CTP/DWI) 4. Vessel occlusion status (from CTA) 5. Pre-stroke functional status (mRS) 6. Maps patient to appropriate protocol (standard window, DAWN, DEFUSE-3, or not eligible)


3. NEDA-3 and MS Treatment Targets

3.1 NEDA-3 Definition

"No Evidence of Disease Activity" is the composite treatment target for relapsing MS:

Component Measurement Definition of "Activity"
Clinical relapses Clinician-assessed Any new or recurrent neurological symptom lasting > 24 hours
MRI activity Annual brain MRI Any new or enlarging T2 lesion OR Gd-enhancing lesion
Disability progression EDSS >= 1.0 point increase sustained at 3-6 months (baseline <= 5.5) OR >= 0.5 point increase (baseline >= 6.0)

3.2 NEDA-4 Extension

Adds brain volume loss (annualized brain volume loss > 0.4% is abnormal). More sensitive but less routinely measured.

3.3 DMT Escalation Decision Framework

Patient on DMT
  |
  +-- Meeting NEDA-3? ---> Yes: Continue current DMT
  |
  +-- No: Evidence of disease activity
       |
       +-- Relapses? ------> 1 mild relapse: Monitor closely
       |                      2+ relapses or severe: Escalate
       |
       +-- MRI activity? --> 1-2 new T2 lesions: May monitor
       |                      3+ new lesions or Gd+: Escalate
       |
       +-- EDSS progression? -> Confirmed: Escalate
       |
       +-- Combined: Any 2+ of above: Strong escalation indication
       |
       +-- Escalation path:
            Platform --> Moderate efficacy --> High efficacy
            Consider: JCV status, pregnancy planning, comorbidities

3.4 JCV/PML Risk Stratification (Natalizumab)

JCV Antibody Index Prior Immunosuppression PML Risk (per 1000)
Negative No ~0.07
Low (< 0.9) No ~0.1
Moderate (0.9-1.5) No ~1.0
High (> 1.5) No ~6.0
High (> 1.5) Yes ~11.0

Risk increases significantly after 24 months of treatment. Extended interval dosing (every 6 weeks instead of 4) may reduce risk.


4. ACLS Stroke Algorithm Integration

4.1 Prehospital Stroke Recognition

BE-FAST mnemonic: - Balance: Sudden loss of balance - Eyes: Sudden vision change - Face: Facial droop - Arm: Arm weakness - Speech: Speech difficulty - Time: Time to call 911

4.2 Emergency Department Stroke Protocol

Stroke Alert Activation
  |
  +-- Time zero: Last known well (LKW) documented
  |
  +-- 10 min: CT head (rule out hemorrhage)
  |
  +-- 15 min: NIHSS scored
  |
  +-- 20 min: Labs drawn (glucose, coags, CBC)
  |
  +-- 25 min: CT angiography (if NIHSS >= 6 or LVO suspected)
  |
  +-- tPA Decision:
  |     LKW < 4.5h + NIHSS >= 1 + no contraindications
  |     --> IV alteplase 0.9 mg/kg (max 90 mg)
  |     --> 10% bolus, 90% over 60 minutes
  |     --> Target door-to-needle < 60 minutes
  |
  +-- Thrombectomy Decision:
  |     LVO confirmed (ICA/M1) + NIHSS >= 6 + ASPECTS >= 6
  |     --> Mechanical thrombectomy
  |     --> Target door-to-groin < 90 minutes
  |
  +-- Extended Window (6-24h):
        DAWN or DEFUSE-3 criteria met
        --> CTP or MRI perfusion for core/penumbra
        --> Thrombectomy if eligible

4.3 Post-tPA Monitoring

  • Blood pressure: < 180/105 for 24 hours
  • Neurological checks: Every 15 minutes for 2 hours, then q30min for 6 hours, then hourly for 16 hours
  • No antiplatelets or anticoagulants for 24 hours
  • CT head at 24 hours before starting antithrombotic therapy
  • Watch for signs of hemorrhagic transformation: sudden neurological worsening, headache, nausea/vomiting

5. EMG/NCS Pattern Recognition

5.1 Fundamentals

Nerve Conduction Studies (NCS): Electrical stimulation of a nerve with recording of the response. Measures: - Amplitude: Reflects number of functioning axons (reduced in axonal disease) - Conduction velocity: Reflects myelin integrity (slowed in demyelinating disease) - Distal latency: Time from stimulus to response - F-wave latency: Proximal nerve conduction (tests nerve roots)

Electromyography (EMG): Needle electrode inserted into muscle. Evaluates: - Insertional activity: Normal, increased (denervation), decreased (end-stage) - Spontaneous activity: Fibrillation potentials and positive sharp waves (active denervation) - Motor unit potential (MUP) morphology: Large/polyphasic (neurogenic), small/short (myopathic) - Recruitment pattern: Reduced (neurogenic), early (myopathic)

5.2 Pattern Recognition Guide

Pattern NCS Findings EMG Findings Differential
Axonal neuropathy Low amplitudes, normal velocities Fibrillations in distal muscles, large MUAPs Diabetic, toxic, hereditary (CMT2)
Demyelinating neuropathy Slow velocities, prolonged distal latencies, conduction block May have secondary axonal changes CIDP, GBS (AIDP), CMT1
Motor neuron disease Normal sensory NCS; may have low CMAPs Diffuse active denervation, fasciculations, large MUAPs ALS, SMA, Kennedy disease
NMJ (postsynaptic) Decremental response on RNS at 2-3 Hz Normal or minimal changes Myasthenia gravis (AChR+, MuSK+)
NMJ (presynaptic) Low CMAP amplitudes; incremental response on rapid RNS -- Lambert-Eaton (VGCC antibodies)
Myopathy Normal NCS or mildly reduced CMAPs Small, short, polyphasic MUAPs; early recruitment Inflammatory myopathy, dystrophy
Radiculopathy Normal NCS (or low CMAP if severe) Fibrillations in myotomal distribution, paraspinal muscles Disc herniation, spondylosis

5.3 Critical Diagnostic Distinctions

GBS vs. CIDP: - GBS: Acute onset (< 4 weeks to nadir), monophasic, post-infectious - CIDP: Chronic (> 8 weeks), relapsing or progressive, no trigger - NCS: Both demyelinating, but CIDP shows more uniform slowing

ALS Diagnosis (El Escorial Criteria): - Definite: UMN + LMN signs in 3+ regions (bulbar, cervical, thoracic, lumbosacral) - Probable: UMN + LMN signs in 2+ regions with UMN rostral to LMN - EMG: Active denervation in 3+ regions with normal sensory NCS is highly suggestive


6. DBS Candidacy Assessment

6.1 Indications for DBS in Parkinson's Disease

DBS is considered when medical therapy is inadequate despite optimization:

CAPSIT-PD (Core Assessment Program for Surgical Interventional Therapies) criteria:

Criterion Requirement
Diagnosis Idiopathic PD (not atypical parkinsonism)
Disease duration >= 5 years
Levodopa response >= 30% improvement in UPDRS Part III on vs. off
Motor complications Disabling dyskinesias OR motor fluctuations
Cognition MoCA >= 26 (no dementia)
Psychiatric No active untreated psychiatric disease
Age Typically < 70-75 (relative, not absolute)
MRI No structural lesions contraindicating surgery

6.2 Target Selection

Target Best For Advantages Considerations
STN (subthalamic nucleus) Broadly effective PD Reduces medication needs by ~50% May affect mood, speech
GPi (globus pallidus interna) Dyskinesia-predominant Directly suppresses dyskinesias Less medication reduction
VIM thalamus Essential tremor, PD tremor-dominant Most effective for tremor Does not help bradykinesia

6.3 Emerging Alternatives

  • Focused ultrasound thalamotomy: Non-invasive, unilateral tremor control. MRI-guided.
  • Adaptive DBS: Closed-loop stimulation that adjusts in real-time based on brain signals.

7. Epilepsy Surgical Evaluation

7.1 Pre-Surgical Workup

Drug-Resistant Epilepsy (failed 2+ ASMs)
  |
  +-- Phase I: Non-invasive evaluation
  |     +-- Video-EEG monitoring (capture habitual seizures)
  |     +-- 3T MRI with epilepsy protocol
  |     +-- Neuropsychological testing
  |     +-- PET (FDG): Interictal hypometabolism
  |     +-- ictal SPECT (if available)
  |     +-- MEG (if available)
  |
  +-- Concordance analysis:
  |     EEG focus + MRI lesion + PET hypometabolism + semiology
  |     All point to same region? --> Proceed to surgery
  |
  +-- Discordant or non-lesional?
        +-- Phase II: Invasive monitoring
              +-- Stereo-EEG (SEEG): Depth electrodes
              +-- Subdural grid electrodes (less common now)
              +-- Functional mapping (language, motor)
              +-- Identify seizure onset zone

7.2 Surgical Options and Outcomes

Procedure Indication Seizure Freedom Rate Engel Class I
Anterior temporal lobectomy TLE with MTS 60-80% 65-75%
Selective amygdalohippocampectomy TLE, sparing lateral temporal 50-70% 55-65%
Lesionectomy Focal lesion (FCD, cavernoma, tumor) 60-80% 60-75%
LITT (laser ablation) MTS, focal lesions, hypothalamic hamartoma 50-65% 50-60%
Corpus callosotomy Generalized/drop attacks (LGS) Palliative; reduces drop attacks 60-80% Rarely class I
Hemispherectomy Hemispheric pathology (Rasmussen, Sturge-Weber) 60-80% 60-70%

7.3 Engel Classification (Surgical Outcomes)

Class Description
I Free of disabling seizures
II Rare disabling seizures
III Worthwhile improvement (> 75% reduction)
IV No worthwhile improvement

8. Neuro-Oncology Molecular Integration

8.1 WHO 2021 CNS Tumor Classification

The 2021 WHO classification requires integrated molecular diagnosis for many tumor types:

Tumor Type Required Molecular Treatment Implications
Glioblastoma, IDH-wildtype IDH1/2 wt, +TERT mutation OR +EGFR amp OR +7/−10 Stupp protocol: maximal resection + RT 60Gy + TMZ
Astrocytoma, IDH-mutant IDH1/2 mut, ATRX loss, no 1p19q codel Grade-dependent; RT + TMZ for grade 3-4
Oligodendroglioma, IDH-mutant, 1p19q-codeleted IDH1/2 mut + 1p19q codel PCV or TMZ + RT; better prognosis
H3 K27-altered diffuse midline glioma H3K27M mutation Radiation; poor prognosis
Medulloblastoma WNT, SHH, Group 3, Group 4 Subgroup-specific therapy

8.2 MGMT Methylation

O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation is the strongest predictive biomarker for temozolomide response in glioblastoma: - Methylated (~40%): Better response to TMZ, median OS ~21 months - Unmethylated (~60%): Poorer response, median OS ~14 months - Testing: Methylation-specific PCR or pyrosequencing

8.3 RANO Criteria (Response Assessment in Neuro-Oncology)

Response Criteria
Complete response No enhancing disease, stable/improved FLAIR, no new lesions
Partial response >= 50% decrease in enhancing tumor area
Stable disease < 50% decrease and < 25% increase
Progressive disease >= 25% increase or new enhancing lesion

9. Advanced EEG Pattern Recognition

9.1 ACNS Standardized Terminology (2021)

Pattern Description Clinical Significance
LPDs (lateralized periodic discharges) Periodic sharp waves, one hemisphere Acute structural lesion, may be ictal
GPDs (generalized periodic discharges) Bilateral periodic discharges Metabolic encephalopathy, CJD, anoxia
LRDA (lateralized rhythmic delta) Rhythmic delta, one hemisphere Strongly associated with seizures
GRDA (generalized rhythmic delta) Bilateral rhythmic delta Encephalopathy, increased ICP
Burst suppression Bursts of activity alternating with suppression Severe encephalopathy, anesthesia, hypothermia
Triphasic waves Anterior-predominant, positive-negative-positive Hepatic encephalopathy (classic), also metabolic, CJD
3 Hz spike-and-wave Generalized, rhythmic Typical absence seizures
Hypsarrhythmia Chaotic, high-amplitude, multifocal spikes West syndrome (infantile spasms)
PLEDs (now LPDs) Periodic lateralized epileptiform discharges Herpes encephalitis, acute stroke, tumor
Alpha coma Unreactive alpha, diffuse, non-posterior Brainstem lesion, anoxia (poor prognosis)

9.2 Ictal vs. Interictal Patterns

Interictal: Epileptiform discharges (spikes, sharp waves) between seizures. Localizing but not diagnostic of ongoing seizure.

Ictal: Evolving rhythmic activity representing an electrographic seizure. Requires: - Definite evolution in frequency, morphology, and distribution - Duration typically > 10 seconds - Clinical correlation (may be subclinical/NCSE)

9.3 Non-Convulsive Status Epilepticus (NCSE)

NCSE is a medical emergency that is underdiagnosed because the patient may not convulse. Suspect in: - Unexplained altered mental status - Subtle motor signs (eye deviation, nystagmus, facial twitching) - Failure to improve after convulsive status epilepticus

Salzburg Criteria for NCSE: - >= 2.5 Hz epileptiform discharges, OR - Epileptiform discharges < 2.5 Hz + clinical improvement with IV benzodiazepine, OR - Subtle clinical signs + EEG correlate


10. Neuromuscular Junction: Advanced Diagnostics

10.1 Repetitive Nerve Stimulation (RNS)

Disease Stimulation Rate Response
Myasthenia gravis 2-3 Hz (low rate) Decremental response > 10%
Lambert-Eaton 2-3 Hz then 50 Hz Decrement at low rate; > 100% increment at high rate
Botulism 2-3 Hz then 50 Hz Similar to LEMS pattern

10.2 Single-Fiber EMG (SFEMG)

Most sensitive test for NMJ disorders: - Jitter: Variability in time between two muscle fiber potentials from same motor unit - Blocking: Failure of one fiber to fire - Abnormal jitter + blocking = NMJ dysfunction - 95% sensitivity for MG (even in ocular MG)

10.3 Antibody Panels

Antibody Disease Frequency Treatment Implication
AChR Generalized MG 85% Cholinesterase inhibitors, immunosuppression, thymectomy
MuSK MG (often bulbar-predominant) 5-10% Rituximab preferred; avoid cholinesterase inhibitors in some
LRP4 MG 1-3% Similar to AChR-MG management
VGCC Lambert-Eaton >90% 3,4-DAP; screen for SCLC
Ganglioside (GM1, GQ1b) GBS subtypes Variable IVIg or PLEX
SOD1 Familial ALS ~20% of fALS Tofersen (ASO therapy)

11. Sleep-Movement Disorder Connection

11.1 RBD as a Prodromal Marker

REM sleep behavior disorder (dream enactment behavior with loss of REM atonia) is the strongest prodromal marker for synucleinopathies: - Conversion rate: ~75-90% will develop PD, DLB, or MSA within 10-15 years - Confirmed by polysomnography: REM without atonia (RWA) - Clinical significance: Patients with isolated RBD should be counseled and monitored - Alpha-synuclein seed amplification assay (SAA): Can detect alpha-synuclein pathology in CSF of RBD patients before motor onset

11.2 Sleep Disorders in Parkinson's Disease

Disorder Prevalence in PD Mechanism
RBD 25-50% Brainstem nuclei degeneration
Excessive daytime sleepiness 30-50% Medication effect + hypothalamic involvement
Insomnia 30-80% Multifactorial (motor, nocturia, depression)
RLS/PLMD 15-30% Dopaminergic dysfunction
Sleep apnea 20-60% Upper airway dysfunction, autonomic

12. Precision Medicine in Neurology

12.1 Gene-Specific Therapies (Approved and Pipeline)

Gene/Target Disease Therapy Mechanism
SOD1 Familial ALS Tofersen (Qalsody) Antisense oligonucleotide
SMN1/2 SMA Nusinersen (Spinraza), Risdiplam (Evrysdi), Onasemnogene (Zolgensma) ASO, SMN2 splicing modifier, gene replacement
SCN1A Dravet syndrome Fenfluramine, cannabidiol Serotonergic, multi-mechanism
HTT Huntington disease Tominersen (suspended), other ASOs in development Huntingtin lowering
APOE Alzheimer's Gene editing approaches (preclinical) Risk modification
LRRK2 Parkinson's LRRK2 kinase inhibitors (Phase II) Kinase inhibition
GBA1 Parkinson's/GD Venglustat, ambroxol (Phase II/III) GCase enhancement

12.2 Biomarker-Driven Treatment Selection

The agent encodes biomarker-to-treatment pathways: - Amyloid PET positive --> lecanemab/donanemab eligibility - SOD1 mutation confirmed --> tofersen initiation - MGMT methylated GBM --> temozolomide benefit predicted - JCV antibody positive --> natalizumab risk stratification - SMN2 copy number --> SMA therapy selection and prognosis


Neurology Intelligence Agent -- Learning Guide: Advanced Topics v1.0.0 HCLS AI Factory / GTC Europe 2026