For Dr. Craig J. Gardner, DO

AI-Powered Clinical
Decision Support for Neurology

What if every medication decision, every risk score, and every drug interaction check had a safety net — one that catches what the human eye misses, deterministically, every time?

Tailored for general neurology, Parkinson's, epilepsy, dementia, and stroke/TIA
2:1
EHR hours per patient hour
8-15+
meds per elderly neuro patient
42%
of elderly have a drug interaction
The Problem
The daily reality of neurology practice
These aren't hypothetical. These are the things that eat your day and keep you up at night.
500K+

Known Drug Interactions

Your Parkinson's patient is on levodopa, a MAO-B inhibitor, an antidepressant, a blood pressure med, and a sleep aid. Checking every pairwise interaction manually is impossible at scale.

2 hours

Documentation Per Hour of Care

For every hour you spend with a patient, you spend two on EHR documentation. Clinical notes, medication reconciliation, referral letters — all manual, all error-prone.

Complex

Dosing is Trial-and-Error

Levodopa timing, wearing-off windows, dyskinesia thresholds — every Parkinson's patient is different. Adjustments are educated guesses informed by experience, not data.

The Solution
AI that checks its own work — with rules you control
Not another chatbot. A clinical decision support layer that uses AI for analysis and deterministic rules for safety.
 Patient Data
 AI Analysis
 Safety Rules
 Verified Output
The AI does the heavy lifting — analyzing patterns, drafting notes, suggesting adjustments. But before anything reaches you, deterministic rules check it against real medical databases. Drug interactions verified against DrugBank. Scores calculated with published algorithms. Dosages checked against formulary limits. No hallucinations. No guessing.
Workflows
Six tools built for your practice
Each one addresses a real clinical workflow. Scroll down for detailed mockups of the three highest-impact tools.

Polypharmacy Safety Checker

Elderly neuro patients average 8-15+ medications. Anticholinergics worsen dementia. Enzyme inducers affect anti-epileptic levels. MAO-B inhibitors interact with everything.
AI reviews the full medication list → DrugBank API checks all pairwise interactions → anticholinergic burden score calculated → Beers Criteria flagged for inappropriate meds in elderly.
Fully Deterministic — Layer 1-3

Parkinson's Medication Optimizer

Levodopa dosing is incredibly complex. Wearing-off effects, dyskinesia windows, timing matters hugely. Each patient responds differently.
AI analyzes symptom patterns + current dosing → suggests adjustment → deterministic rules enforce: max daily levodopa dose, interaction checks with MAO-B/COMT inhibitors, minimum dosing intervals.
Deterministic — Layer 2 + 3

Dementia Screening Triage

MoCA and MMSE scores need context — age-adjusted norms, medication review for cognitive side effects, differential between depression, delirium, and dementia.
AI interprets scores + history → deterministic rules: age-adjusted score thresholds, anticholinergic burden check, Beers Criteria for cognition-impairing medications.
Deterministic — Layer 2 + 3

Epilepsy Drug Interaction Guard

Anti-epileptic drugs are enzyme inducers/inhibitors with massive interaction profiles. Narrow therapeutic windows. Pregnancy teratogenicity risks.
AI manages the medication plan → deterministic rules: therapeutic level ranges, interaction checks (carbamazepine metabolism), pregnancy category warnings, serum level validation.
Deterministic — Layer 2 + 3

Stroke / TIA Risk Calculator

After TIA, you need rapid stroke risk assessment and anticoagulation decisions. Multiple validated scoring systems: ABCD2, CHA₂DS₂-VASc, HAS-BLED.
AI collects patient data → scores calculated deterministically using published algorithms → anticoagulation recommendation with bleeding risk balance. Pure arithmetic.
Pure Layer 2 — Validated Algorithms

Clinical Note Safety Net

AI-generated clinical notes risk hallucinating medications or diagnoses. Documentation consumes 2x more time than patient care.
AI drafts the note → deterministic rules: medication list must match EHR, diagnosis codes validated against ICD-10, dosages checked against formulary ranges, allergies cross-checked.
Deterministic — Layer 2 + 3
Detailed Mockups
See it work, step by step
Three workflows with real neurology data. Every drug name, dose, and interaction is medically accurate.
Workflow 1
Polypharmacy Safety Checker
Patient with Parkinson's and depression. 7 active medications. AI reviews the list, DrugBank catches two dangerous interactions.
IN
Input — patient medication list
72-year-old, Parkinson's + depression + hypertension
patient.json
{
  "age": 72,
  "conditions": ["parkinson's", "depression", "hypertension"],
  "medications": [
    { "name": "carbidopa-levodopa", "dose": "25/100mg", "freq": "TID" },
    { "name": "selegiline", "dose": "5mg", "freq": "BID" },
    { "name": "sertraline", "dose": "100mg", "freq": "daily" },
    { "name": "amlodipine", "dose": "5mg", "freq": "daily" },
    { "name": "omeprazole", "dose": "20mg", "freq": "daily" },
    { "name": "diphenhydramine", "dose": "25mg", "freq": "PRN" },
    { "name": "atorvastatin", "dose": "20mg", "freq": "daily" }
  ]
}
1
Step 1 — DrugBank API: pairwise interaction check
21 pairwise checks (7 medications × 6 ÷ 2)
Every combination checked against DrugBank's 500K+ interaction database. Results are deterministic — same meds, same result, every time.
selegiline + sertraline MAJOR — serotonin syndrome risk
omeprazole + carbidopa-levodopa MODERATE — may reduce levodopa absorption
19 other pairs no significant interactions
2
Step 2 — Anticholinergic burden score
Calculated from standardized ACB scale
Anticholinergic medications worsen cognitive decline in elderly patients — especially critical in someone already at risk for dementia with Parkinson's.
diphenhydramine — ACB score: 3 (high) ALERT — strong anticholinergic in a 72yo with Parkinson's
3
Step 3 — Beers Criteria check (age ≥ 65)
AGS Beers Criteria for potentially inappropriate medications in elderly
diphenhydramine — Beers listed avoid in elderly — confusion, falls, urinary retention
omeprazole > 8 weeks — Beers caution long-term PPI use — fracture risk, C. diff
OUT
Verified output
Actionable safety report
Safety Report — 3 Issues Found
Critical
Selegiline + Sertraline — serotonin syndrome risk. Consider switching SSRI or MAO-B inhibitor. DrugBank
Recommend
Discontinue diphenhydramine — high anticholinergic burden (ACB 3), Beers listed, worsens Parkinson's cognition. Suggest melatonin for sleep.
Monitor Review omeprazole duration — consider step-down if >8 weeks. May affect levodopa absorption.
Audit 21 interaction checks · 7 ACB scores · 7 Beers checks · all deterministic
Workflow 2
Parkinson's Medication Optimizer
Patient experiencing wearing-off between doses. AI suggests a timing adjustment, deterministic rules enforce safety limits on total daily levodopa.
IN
Input — current regimen + symptom report
Wearing-off 90 min before next dose
symptoms.json
{
  "current_regimen": {
    "drug": "carbidopa-levodopa",
    "dose": "25/100mg",
    "schedule": ["7:00", "12:00", "17:00"],
    "daily_levodopa": 300
  },
  "symptoms": {
    "wearing_off": true,
    "wearing_off_minutes_before_dose": 90,
    "dyskinesia": false,
    "peak_dose_tremor_control": "good"
  }
}
1
Step 1 — AI analysis + consensus
AI suggests: increase frequency to QID, keep per-dose at 25/100mg
Rationale: patient has good peak-dose control but wearing off too early. Adding a 4th dose (every 3.5hrs instead of 5hrs) extends coverage without increasing per-dose amount. No dyskinesia means room for more total levodopa.
2
Step 2 — Deterministic safety rules
Levodopa limits, interaction checks, timing validation
Max daily levodopa ≤ 800mg 400mg proposed → well under limit
Min dose interval ≥ 2 hours 3.5hr intervals → PASS
MAO-B interaction check selegiline + levodopa 400mg → within safe range
Dyskinesia risk threshold no current dyskinesia, 33% dose increase → low risk
OUT
Verified output
Suggested regimen change with safety clearance
Recommendation — All Safety Checks Passed
Suggestion Increase to QID: 25/100mg at 7:00, 10:30, 14:00, 17:30
New daily total 400mg levodopa (was 300mg) — within safe range
Rationale Shorter intervals (3.5hr vs 5hr) address wearing-off without increasing per-dose amount. Monitor for dyskinesia at new total.
Confidence
0.89
Follow-up Re-evaluate in 2 weeks. If dyskinesia emerges, consider adding entacapone (COMT inhibitor) instead of 4th dose.
Workflow 3
Stroke / TIA Risk Calculator
Patient presents with suspected TIA. AI collects data, validated scoring algorithms calculate risk deterministically, anticoagulation decision is data-driven.
IN
Input — patient presentation
68-year-old, transient right-arm weakness, resolved in 40 minutes
tia-assessment.json
{
  "age": 68,
  "sex": "male",
  "event": "transient right arm weakness, 40 min",
  "bp_at_presentation": "158/92",
  "history": {
    "hypertension": true,
    "diabetes": true,
    "chf": false,
    "prior_stroke": false,
    "vascular_disease": true,
    "atrial_fibrillation": true
  }
}
1
Step 1 — ABCD2 Score (7-day stroke risk after TIA)
Deterministic algorithm — published scoring, zero ambiguity
CriteriaDetailsPoints
Age ≥ 6068 years old1
Blood pressure ≥ 140/90158/92 mmHg1
Clinical featuresUnilateral weakness2
Duration ≥ 60 min40 minutes (10-59 min range)1
DiabetesYes1
ABCD2 Total6 / 7
HIGH RISK — Score ≥ 6 8.1% 7-day stroke risk → urgent workup required
2
Step 2 — CHA₂DS₂-VASc (annual stroke risk with AFib)
Patient has atrial fibrillation — anticoagulation decision needed
CriteriaPresent?Points
CHFNo0
HypertensionYes1
Age ≥ 75No (68)0
DiabetesYes1
Stroke/TIACurrent TIA2
Vascular diseaseYes1
Age 65-74Yes (68)1
Sex (female)No0
CHA₂DS₂-VASc Total6 / 9
Score ≥ 2 in males → Anticoagulation recommended ~9.8% annual stroke risk without anticoagulation
3
Step 3 — HAS-BLED (bleeding risk assessment)
Before starting anticoagulation, check bleeding risk
HAS-BLED Score: 2 HTN (1) + Age >65 (1) = moderate risk → proceed with caution
Stroke risk (9.8%) >> bleeding risk (1.88%) benefit clearly outweighs risk → anticoagulation recommended
OUT
Verified output — fully deterministic
Risk-stratified recommendation with scoring breakdown
Risk Assessment Complete
ABCD2 6/7 — HIGH RISK (8.1% 7-day stroke risk). Urgent neurovascular workup.
CHA₂DS₂-VASc 6/9 — Anticoagulation strongly recommended (9.8% annual stroke risk).
HAS-BLED 2 — Moderate bleeding risk. Benefit outweighs risk (9.8% vs 1.88%).
Recommendation Start DOAC (apixaban 5mg BID or rivaroxaban 20mg daily). Recheck renal function. Schedule carotid imaging within 24hrs.
Validation 3 scoring systems · all published algorithms · 0 NLP involved · fully reproducible
Your Input
What would actually help?
This isn't a finished product — it's a question. I want to know which of these would make the biggest difference in your day.