All Case Studies
Ab
Abstrabit Technologies
AI Process Automation
Case Study
🏥 Ogden Clinic · Multi-Specialty · 500+ Staff
AI Triage · Chain of Thought · Urgency Classification

A Cardiac Emergency
Sitting in the Waiting Room.
We Built the AI That
Fixed the Queue.

15–20 min
Manual assessment
per patient
→
Seconds
AI pre-assessment
with reasoning trace
⚡ IMMEDIATE
See now
⚠ URGENT
Prioritized
✓ ROUTINE
Standard slot
15–20 min
Manual assessment replaced by AI in seconds
3 Tiers
Urgency classification at intake
Chain of Thought
Visible reasoning — not a black box
5 Records
Historical context ingested per patient
HIPAA
Compliant audit trail for every action
Client
Ogden Clinic
Type
Multi-Specialty
Staff
500+
AI Model
GPT-4o
Framework
Med-Prompt
Infra
AWS · HIPAA
abstrabit.com
01 / 04
Ab
Abstrabit Technologies
AI Process Automation
The Problem
The Problem

First-Come, First-Served.
Critical Patients Waited Behind Routine Ones.

Before — Arrival Order ⚠ Clinical risk
1
Patient A · Routine follow-upRoutineseen 1st
2
Patient B · Medication checkRoutineseen 2nd
3
Patient C · Chest pain, mildUrgentseen 3rd
4
Patient D · Annual physicalRoutineseen 4th
5
Patient E · Radiating chest pain, sweating⚡ CARDIACwaiting 47 min
After — AI Urgency Queue ✓ Critical first
1
Patient E · Radiating chest pain, diaphoresis⚡ IMMEDIATEseen 1st
2
Patient C · Chest pain, mild onsetUrgentseen 2nd
3
Patient A · Routine follow-upRoutineseen 3rd
4
Patient B · Medication checkRoutineseen 4th
5
Patient D · Annual physicalRoutineseen 5th
⏱
15–20 min Manual Review Per Patient
all cases treated equally
🔀
No Prioritization Without a Physician
first-come-first-served
📂
Each Visit Evaluated Without History
no context at triage point
📋
No Structured AI Audit Trail
HIPAA documentation gap
abstrabit.com
02 / 04
Ab
Abstrabit Technologies
AI Process Automation
Architecture & Proof
System Architecture

Not a Black Box. A Reasoned Argument.
Every Recommendation Comes With Visible Reasoning.

🧠 Chain of Thought — 4-Step Visible Reasoning
01
Demographics
Age, sex, BMI, risk factors as baseline context
→
02
Vitals
BP, HR, SpO₂ read against patient baseline
→
03
Symptom Clusters
Evaluated together — not individually
→
04
Medical History
5 prior records — chronic patterns detected
Output: Urgency + Confidence-Scored Diagnosis + Reasoning Trace
Physician edits, confirms, or overrides. Every action logged.
visible to physician
↓ same reading · different history · different triage
📂 Why History Changes the Decision
same BP · two conclusions
No prior records
148/92
First visit. No hypertension history. No baseline to compare.
✓ Routine — monitor and recheck
5 records ingested
148/92
Hypertensive × 4 yrs. Last reading 112/74. Today: headache + blurred vision.
⚡ Immediate — acute spike + neuro symptoms
↓ classification + confidence scoring
🚨 Three-Tier Urgency
Front desk acts before physician triage
⚡ IMMEDIATE
See now — physician directed immediately
⚠ URGENT
Time-sensitive — expedited in queue
✓ ROUTINE
Standard visit — safely deferrable
📊 Confidence-Scored Differential
Male · 58 · chest pain + diaphoresis
Unstable angina
82%
NSTEMI
11%
Musculoskeletal
5%
Anxiety / GERD
2%
What Physicians Actually See

82% Likelihood. Reasoning Visible.
The AI Argues Its Case. The Physician Evaluates It.

ogden-triage-portal · physician-view 🔒 HIPAA Active
Robert H. Vasquez
Male · 58 · MRN: OCL-44821 · Arrived 10:14am
Chest tightness, left arm pain, diaphoresis — sudden onset
AI Urgency
IMMEDIATE
🧠 Chain of Thought — visible to physician
Step 1 · DemographicsHigh-risk profile
Male, 58. BMI 29.4. Type 2 diabetes, hypercholesterolaemia, former smoker. Elevated cardiac risk before symptom analysis.
Step 2 · VitalsBaseline deviation detected
BP 158/101 (baseline 124/78). HR 104. SpO₂ 94%. 27% spike from controlled baseline.
Step 3–4 · Symptoms + HistoryACS presentation
Chest tightness + arm radiation + diaphoresis. Classic ACS triad in 58M with cardiac risk factors.
▸ Differential — confidence scored
82%
Unstable Angina
ACS pattern · risk factors present · symptom triad
ECG immediately · troponin panel · cardiology consult
11%NSTEMI — pending troponin
5%Musculoskeletal — inconsistent with diaphoresis
2%Anxiety / GERD — inconsistent with vital deviation
Physician override logged · HIPAA audited
✓ Accept Override
🚨 Live Queue — 10:15am
sorted by AI urgency
1
Robert Vasquez
Chest tightness, arm pain, sweating
⚡ IMMEDIATE
2
Dana Kim
Severe headache, neck stiffness
⚠ URGENT
3
Marcus Lee
Shortness of breath, 2-day onset
⚠ URGENT
4
Patricia Chen
Medication follow-up
✓ Routine
5
Thomas Grant
Annual physical
✓ Routine
Robert arrived 4th. Under first-come-first-served, he'd wait behind 3 routine appointments. The AI placed him first at intake.
📂 Same Reading, Different Triage
History changes the conclusion
No prior records
148/92
First visit. No hypertension history.
✓ Routine — monitor
5 records ingested
148/92
Controlled at 112/74 three weeks ago. Today: headache + visual changes.
⚡ Immediate — acute spike
abstrabit.com
03 / 04
Ab
Abstrabit Technologies
AI Process Automation
Results
Business Impact

Right Patient Seen First. Every Time.
With a Reasoning Trace They Can Examine.

Seconds
AI pre-assessment — replaced 15–20 min manual review
Chain of
Thought
Visible reasoning — physician examines the argument
HIPAA
Every recommendation logged · every override captured
Assessment Time
Before
15–20 min
After (AI)
Seconds
Critical Patient ID
Before
During physician review
After
At intake
Physician Decision Support
Before
Start from scratch
After
CoT diagnosis + scores
Tech Stack
GPT-4o Clinical Reasoning Med-Prompt Framework Chain of Thought (CoT) 5-Record History Ingestion Urgency Classification React · Vite · TypeScript Python Flask Backend AWS DynamoDB · EC2 AWS Cognito (HIPAA) CloudWatch Audit Log
Metric
Before
After
Initial assessment
15–20 min manual
Seconds (AI)
Critical identification
During physician review
Flagged at intake
Prioritization
First-come, first-served
Urgency-based queue
Decision support
Start from scratch
CoT diagnosis + scores
History at intake
Not surfaced
5 records integrated
Audit trail
Manual documentation
Automated · every action
Why Visible Reasoning Is the Product
An AI that returns "82% unstable angina" without showing its work is a liability. The Chain of Thought trace lets physicians examine the argument — demographics, vitals deviation, symptom clusters, historical patterns — and decide whether they agree. That's the product: not a prediction, but a reasoned argument a physician can evaluate.
abstrabit.com
04 / 04