VECTORYX Medical AI, Understanding the Heart in 3D

The Hidden Problem

The ECG is the most used diagnostic tool in medicine, and the least understood.

A landmark meta-analysis in JAMA Internal Medicine laid bare an uncomfortable truth. ECG interpretation accuracy plateaus well before competence. After a century of surface-first teaching, even cardiologists top out around three-quarters. The ceiling is not talent. It is pedagogy.

Med students

42%

Residents

56%

Practicing physicians

69%

Cardiologists

75%

A ceiling built by pedagogy, not by science. Cook, Oh and Pusic (JAMA Intern Med, 2020) showed systematic variability across all levels of training, a signal that ECG education itself is the problem, not the learner. The science that governs ECG interpretation (cardiac action potential, three-dimensional anatomy, conduction system propagation, volume conduction biophysics) has been understood with precision for decades. What has not been adequately resolved is how to teach it.

What sits beneath every ECG

Five causal layers. Four of them invisible to the student.

1Cellular ElectrophysiologyHidden

23D Cardiac Anatomy in situHidden

3Conduction System PropagationHidden

4Volume Conduction, Solid AngleHidden

⎯⎯⎯⎯⎯⎯⎯⎯⎯ the waterline of memorization ⎯⎯⎯⎯⎯⎯⎯⎯⎯

5. Surface ECG Expression

All students see

A Century of Conceptual Foundations

Five paradigm shifts in ECG education, 1906 to 2026.

Over the past 120 years, ECG education has evolved through five identifiable paradigms, each a genuine conceptual advance. Yet none of them inverted the teaching sequence. Every paradigm introduced learners from the surface inward. The symptom was taught. The mechanism was not.

1906 to 1930 Era I · Discovery

Einthoven

Original description of the electrocardiogram. Waveform identification. The Einthoven triangle.

1912 to 1950s Era II · Physiological

Lewis, Wilson, Goldberger

Physiological interpretation. Precordial leads and augmented leads. Cardiac electrical mechanisms and vectors.

1950s to 1980s Era III · Structured

Marriott, Wagner

Structured educational frameworks. Systematic, step-by-step ECG analysis.

1990s to 2020s Era IV · Mechanistic Clinical

Bayés de Luna

Mechanistic clinical ECG interpretation, electrophysiology, and clinical correlation.

2020s to 2026 Era V · Cognitive AI

VECTORYX, the Five-Layer Framework

AI-enhanced adaptive learning. Physics-based 3D modeling. Immersive VR. Cause before effect.

The Five-Layer Framework

We reversed the teaching logic. ECG waves emerge last, not first.

Rather than starting with the surface tracing, VECTORYX begins inside the thorax, with the biology of the cardiac cell, the anatomy of the heart in its actual 3D position, and the physics of how electrical activity reaches the skin. The 12 leads are the predictable consequence, not a pattern to memorize.

Figure 1. The Five-Layer Conceptual Model. Each layer is scientifically necessary to understand the one that follows it.

Cellular Electrophysiology

Ion channels, action potentials, refractory periods. The molecular foundation of every deflection on the tracing. Before the P wave exists, sodium has already flooded in. At rest, the cardiomyocyte sits at approximately negative 90 millivolts, actively maintained by inward rectifier K⁺ channels and the sodium-potassium ATPase pump.

Layer 1 · Molecular foundation

3D Cardiac Anatomy in situ

The heart does not hang apex down like the Valentine's symbol. In 67% of individuals it lies compressed against the diaphragm in a horizontal C-shape. The apex points anteriorly and to the left, not inferiorly. The atria are posterior structures. This spatial reality is invisible in flat ECG pattern teaching.

Layer 2 · Spatial, attitudinal

Impulse Formation and Conduction

SA node, AV node, bundle of His, Purkinje network. Wilhelm His Jr. described the AV bundle in 1893. Sunao Tawara elucidated the AV node and Purkinje system in 1906. Keith and Flack identified the SA node in 1907. The sequential activation generates the QRS as five discrete biventricular vectors, each with a specific anatomical address.

Layer 3 · Dynamic propagation

Volume Conduction and the Solid Angle

Dipoles do not reach electrodes directly. They propagate through a heterogeneous thoracic medium. The solid angle theorem (Ep = Ω / 4π · k · ΔVm) provides a rigorous framework. Each lead "sees" the heart from a specific vantage. The ECG waveform is a predictable geometric consequence, not an arbitrary electrical signature.

Layer 4 · Biophysical geometry

Surface ECG Expression

Only here does the 12-lead tracing appear. Twelve photographers, one event. Six limb leads sample the frontal plane (superior, inferior). Six precordial leads sample the horizontal plane (anterior, posterior). Each wave is the scalar projection of a 3D vector field onto 12 spatial axes. Predictable, not memorized.

Layer 5 · Clinical expression

"Instead of connecting waveforms in silos, learners can interact with dynamic simulations that begin at the cellular level, simulate impulse propagation through the conduction system, and demonstrate how vectors project onto the 12 leads to produce the surface tracing."

Uribe, van Dam, Proniewska, Nalepa. A Five-Layer Conceptual Framework for Understanding the Electrocardiogram, 2026.

Inside Layer 2, 3D Anatomy in situ

Three cardiac orientations, not one 'normal.'

The heart taught in most textbooks (the so-called Valentine position, heart on its apex) is not how the heart actually sits in the living thorax. Virtual dissection of computed tomographic datasets reveals three orientations. The most common is not the one you were taught.

Heart in normal G-shape orientation (28% of population) with hexaxial reference system in frontal plane

G shape · Mesomorphic

"Normal" orientation (minority)

The orientation most commonly depicted in textbooks. Moderately tilted heart. Intermediate electrical axis (approximately +60°). This is actually a minority presentation.

Heart in horizontal C-shape orientation — most common (67% of population) with hexaxial reference system

C shape · Brevilineal (Sancho Panza)

The most common orientation

Heart lies flat against the diaphragm. Drives a leftward electrical axis. Produces ECG patterns often mislabeled as "variants" in classic teaching. This is the majority, yet rarely taught as such.

Heart in vertical U-shape orientation (5% of population) with hexaxial reference system in frontal plane

U shape · Longilineal (Don Quixote)

Vertical orientation

Heart appears suspended, apex pointing more inferiorly. Slender body habitus. Inferior electrical axis. Tall, narrow chest radiograph profile.

Why this changes everything. The attitudinal orientation of the heart determines the electrical axis, the polarity of each lead, and the interpretation of every waveform. When students learn ECG on the Valentine heart, they learn a geometry that describes less than one in three patients. The Five-Layer Framework teaches the living heart, in its real thoracic position, with VR navigation and vectorcardiographic projection at every step.

The Medical Cognition Operating System

AI does not diagnose. AI personalizes how you learn to reason.

VECTORYX is built on a simple but radical commitment. The AI does not give learners answers. It gives them better questions. It identifies where reasoning has collapsed, which assumption was wrong, which of the five layers was skipped, and returns the learner to the point of genuine understanding, not the point of a correct guess.

Socratic, Not Prescriptive

The AI identifies where your reasoning broke.

Every answer you give creates a trace. The engine does not reward correct guesses. It reads the path you took to get there, flags where a layer was skipped, and returns you to the layer that matters. Bayesian Knowledge Tracing models each learner's mastery state at every layer. Content is gated by demonstrated understanding, not by time spent.

Learns knowledge gaps, misconceptions that keep reappearing
Detects reasoning errors, not just wrong answers
Maps cognitive pathways, how you think, not the average student
Builds adaptive mental models, the scaffold is removed when you no longer need it

Scaffolding, Not Crutches

The visualization is present during construction, and removed when the understanding is solid.

3D modeling, vector animation, VR immersion, adaptive sequencing. All of it fades as mastery builds. What remains is a clinician who can reason correctly at 3 AM, from cellular electrophysiology to surface waveform, with no technology available except their own mind.

Input. Interaction patterns, response accuracy, reasoning paths, time on concept, repeated misconceptions
Prediction. The optimal next learning step for this specific learner
Output. Adapted sequence, difficulty, visualization type, guidance level
Goal. Mechanistic understanding, not a pattern library dressed in 3D

Five-Layer Framework visualization showing stacked translucent disks for Cellular Physiology, Action Potential Dynamics, 3D Anatomy, Impulse Propagation, and Surface ECG Manifestation, with adaptive AI cognitive inputs on the left including Knowledge Gap detection, Enhance Reasoning, Learner Behavior tracking, Always Available support, and Adaptive Logic, plus a machine learning caption describing how the agentic AI software gets to know the student and builds adaptive progression based on learner data. — Figure. *The Agentic AI Engine across the Five-Layer Framework.* Cognitive inputs (left) feed adaptive logic that operates simultaneously on every layer of the curriculum (right), from cellular physiology upward to the surface ECG.

↓

Cognitive Load

We reduce this.

The friction of memorizing surface patterns whose origin was never explained. Rote recall of rules without understanding. Scattered, disconnected waveform facts that collapse under clinical pressure.

Reduced by adaptive scaffolding, 3D visualization, and causal sequencing.

≠

↑

Cognitive Depth

We never reduce this.

The analytical depth required to understand why a waveform is aberrant, not just that it is. The ability to trace a surface finding back through five causal layers to its molecular origin.

Strengthened by cross-layer reasoning, Socratic AI, and mechanistic questioning.

The Adaptive Assessment, Live Demo

Two axes. Two kinds of intelligence. One personalized pathway.

Every VECTORYX session begins by measuring you along two orthogonal axes. What you know (knowledge concepts) and how you reason (cognitive concepts). Try a condensed version of the assessment below. In 6 questions, the engine will place you on the cognitive map and generate the pathway only you need.

⏱ Approximately 3 minutes · 6 questions

This is not a quiz. It is how the AI meets you where you are.

You will see 3 knowledge questions (what the ECG is) and 3 cognitive questions (how the ECG is reasoned about). Your answers, and the path you take to them, plot you on a 2D cognitive map and determine where the platform will begin teaching.

Axis X · Knowledge

Do you know the facts?

Action potential phases, vector directions, lead orientations, wave definitions. The declarative content of electrocardiography.

Axis Y · Cognition

Can you reason across layers?

When shown an abnormal tracing, can you walk the five layers backward to identify which cellular, anatomical, or biophysical process changed?

Knowledge Cellular Electrophysiology

Question 1 / 6

Reasoning trace

Select an option to continue

AI Placement Complete

Your placement is not a grade.
It is a starting coordinate.

The scatter map plots your knowledge score (horizontal) against your cognitive reasoning score (vertical). The quadrant you land in dictates the shape of your pathway, not the content itself. Everyone reaches Layer 5. The route is what we personalize.

AI Reasoning Trace

Knowledge score…

Cognitive score…

Layer of first gap…

Pathway archetype…

Visualization level…

Independence coefficient…

Reasoning strong, facts thin

Mastery, reinforce and extend

Rebuild from Layer 1

Pattern trained, no scaffold

Knowledge →

Cognition →

◉ Recommended Pathway

Your adaptive pathway

ⓘ This is a condensed demo. The live platform uses 1,000+ annotated ECG cases and continuously recalibrates.

Why ECG is the Entry Wedge

For many diseases, the ECG is not the first test. It is the only one.

No other diagnostic tool combines physiology, physics, and clinical reasoning in the same way. This is what makes ECG the ideal entry point for an AI cognitive medical learning platform. In a large class of conditions, the diagnosis cannot be made without the tracing.

Tier 1 · Primary Pillar

Acute Ischemic Syndromes

ST-elevation myocardial infarction (STEMI)
Non-ST-elevation myocardial infarction (NSTEMI)
Posterior and right ventricular MI

Without ECG, acute coronary care algorithms cannot function.

Tier 2 · Gold Standard

Cardiac Arrhythmias

Atrial fibrillation, atrial flutter
Ventricular tachycardia, fibrillation
Supraventricular tachycardia

No imaging modality diagnoses rhythm without ECG evidence.

Tier 3 · Defined by ECG

Conduction Disease

Atrioventricular block (first, second, third degree)
Right bundle branch block
Left bundle branch block, fascicular blocks

These disorders are defined electrophysiologically.

Tier 4 · Only ECG Can Define

Channelopathies (electrical disorders)

Long QT syndrome (QTc > 500 ms)
Brugada syndrome (Type 1 pattern)
Short QT syndrome, early repolarization

Genetic testing confirms subtype, but ECG establishes the phenotype.

Tier 5 · Syndromic

Wolff, Parkinson, White syndrome

Delta wave
Short PR interval (< 120 ms)
Preexcitation pattern

EP study confirms the accessory pathway, but ECG establishes the syndrome.

Tier 5 · Syndromic

Acute Pericarditis

Diffuse ST elevation
PR segment depression
Spodick's sign

These findings are the diagnostic criteria.

Tier 6 · Clinical Clues

Pulmonary Embolism

S1Q3T3 pattern
Right heart strain (T inversion V1 to V4)
New right bundle branch block

ECG cannot confirm PE but provides classic diagnostic clues.

Tier 7 · Continuous Monitoring

Emergency monitoring

Torsades de pointes
Complete heart block
Ventricular fibrillation

In emergency settings, ECG monitoring is the only immediate diagnostic method.

65M+

The ECG is everywhere healthcare is.

The global healthcare workforce is approximately 65 million professionals (WHO / BMJ Global Health), expanding toward 104 million by broader definitions. Every physician, resident, nurse, paramedic, and ECG technician needs to read and reason about the electrocardiogram. VECTORYX addresses a market where the skill is universal, the pedagogy is broken, and the consequences of misinterpretation are measured in lives.

Sources. WHO Global Health Workforce Statistics · BMJ Global Health (Boniol et al., 2022) · AHA STEMI Guidelines · ESC Clinical Practice Guidelines.

Longitudinal Follow up

The AI doesn't just place you. It watches you grow.

After placement, VECTORYX tracks layer by layer mastery, catalogs persistent misconceptions, and validates durability with "AI off" retention checks at 3 and 6 months. Faculty see the cohort level view. Learners see their own trajectory.

Layer Mastery · Learner #0001

Updated 14 min ago

L1Cellular Electrophys.

82%

L23D Anatomy in situ

68%

L3Conduction System

44%

L4Volume Conduction

31%

L5Surface ECG

58%

+14%

7 day improvement

▲ ahead of cohort

L3, L4

Current focus zone

Misconception Heatmap · 28 day window

Per concept density

Low

High

91%

Retention at 3 months (AI off)

▲ vs 68% in control

Persistent gaps flagged

Visualization · Inside the Platform

VR immersion. 3D cardiac modeling. Dynamic vectocardiography.

The visualization layer is where the five-layer framework becomes tactile. Students navigate inside an anatomically accurate heart, watch impulse propagation in real time, and observe the solid angle projection that turns a 3D dipole into a 2D waveform. Powered by CineECG (Dr. Peter van Dam), integrated WebXR immersive modules, and the HoloMed platform (Dr. Klaudia Proniewska).

VR headset demonstration at cardiology conference

VR Anatomy · Heart in Situ

Walk inside the thorax. See the conduction system from the valve plane downward.

Global Reach · Medical Cognition OS

Clinicians everywhere. One mechanistic foundation.

Cardiac conduction system rendered in 3D heart model

Layer 3

Conduction system, animated.

Solid angle theorem diagram showing electrode field of view

Layer 4

Solid angle. Geometry before waveform.

Surface Expression

Waves as vector shadows, not patterns.

The Founding Team

Built by clinicians, biophysicists, XR engineers, and AI researchers.

Four co-founders, four complementary disciplines. Clinical cardiology and electrophysiology meet physics-based cardiac modeling, immersive medical XR, and explainable human-centered artificial intelligence. CES University · Jagiellonian University Medical College · UMC Utrecht · AGH University.

William Uribe, MD

Founder, Chief Medical Officer

Cardiologist and electrophysiologist. 30+ years of clinical and academic expertise in ECG interpretation, arrhythmias, and cardiac electrophysiology. Co-author of landmark HRS, EHRA, APHRS expert consensus statements. Originator of the Five-Layer Pedagogical Framework. Harvard Medical School Executive Education, AI in Health Care.

PvD

Peter van Dam, PhD

Co-Founder, Chief Science Officer

Inventor of CineECG. Cardiac biophysicist at UMC Utrecht and Jagiellonian University Medical College. Founder of ECG Excellence BV (formerly Peacs BV). Leading international expert in computational cardiac modeling, volume conduction theory, and the inverse problem in electrocardiography. Approximately 3,250 citations.

Klaudia Proniewska, PhD, Eng.

Co-Founder, Chief Innovation Officer

Deputy Director, Center for Digital Medicine and Robotics, Jagiellonian University Medical College. Head of the 3D Functional and Virtual Medical Imaging Laboratory. Pioneer of the HoloMed project (Microsoft HoloLens 2 for holographic visualization of medical data). Polish Society of Cardiology member.

Grzegorz J. Nalepa, PhD, DSc

Co-Founder, Chief Technology Officer

Full Professor of Artificial Intelligence, Jagiellonian University. Coordinator of the Jagiellonian Human-Centered AI Laboratory (JAHCAI). Professor of Machine Learning, Halmstad University (Sweden). Leading Polish researcher in Explainable AI (XAI), affective computing, knowledge engineering, and human-centered AI. 150+ publications.

Full Profiles and CVs

Dig deeper. Download the full research profiles.

Each co-founder brings decades of specialized expertise. Access complete professional profiles, publication histories, academic credentials, and collaboration networks.

The ECG is the most used diagnostic tool in medicine, and the least understood.

Five causal layers. Four of them invisible to the student.

Five paradigm shifts in ECG education, 1906 to 2026.

Einthoven

Lewis, Wilson, Goldberger

Marriott, Wagner

Bayés de Luna

VECTORYX, the Five-Layer Framework

We reversed the teaching logic. ECG waves emerge last, not first.

Cellular Electrophysiology

3D Cardiac Anatomy in situ

Impulse Formation and Conduction

Volume Conduction and the Solid Angle

Surface ECG Expression

Three cardiac orientations, not one 'normal.'

"Normal" orientation (minority)

The most common orientation

Vertical orientation

AI does not diagnose. AI personalizes how you learn to reason.

The AI identifies where your reasoning broke.

The visualization is present during construction, and removed when the understanding is solid.

We reduce this.

We never reduce this.

Two axes. Two kinds of intelligence. One personalized pathway.

This is not a quiz. It is how the AI meets you where you are.

Do you know the facts?

Can you reason across layers?

Your placement is not a grade.It is a starting coordinate.

Your adaptive pathway

For many diseases, the ECG is not the first test. It is the only one.

Acute Ischemic Syndromes

Cardiac Arrhythmias

Conduction Disease

Channelopathies (electrical disorders)

Wolff, Parkinson, White syndrome

Acute Pericarditis

Pulmonary Embolism

Emergency monitoring

The ECG is everywhere healthcare is.

The AI doesn't just place you. It watches you grow.

Layer Mastery · Learner #0001

Misconception Heatmap · 28 day window

VR immersion. 3D cardiac modeling. Dynamic vectocardiography.

Walk inside the thorax. See the conduction system from the valve plane downward.

Clinicians everywhere. One mechanistic foundation.

Conduction system, animated.

Solid angle. Geometry before waveform.

Waves as vector shadows, not patterns.

Built by clinicians, biophysicists, XR engineers, and AI researchers.

William Uribe, MD

Peter van Dam, PhD

Klaudia Proniewska, PhD, Eng.

Grzegorz J. Nalepa, PhD, DSc

Dig deeper. Download the full research profiles.

Learner sign in

Admin edit mode

Your placement is not a grade.
It is a starting coordinate.