AI-based coronary artery disease risk assessment

The Cardio Explorer® was developed in collaboration with the cardiology department at the University Hospital Basel. The algorithm has been generated with advanced AI methods, combining associative mathematical structures (e.g., neural networks, self-organizing maps) with evolutionary models. This approach captures complex, non-linear relationships between biomarkers and clinical outcomes.

From an initial pool of over 100 factors, 32 predictive biomarkers were identified and integrated into the final algorithm. The result is a multi-marker solution with superior accuracy compared to single parameters, validated in peer-reviewed clinical studies with more than 4,500 patients.

Cardio Explorer® uses 15 blood factors, including:

• hs-troponin T

• Mean corpuscular hemoglobin concentration (MCHC)

• White blood cells (leucocytes)

• Urea

• Uric acid

• Glucose

• Cholesterol (total)

• LDL cholesterol

• HDL cholesterol

• Liver enzyme glutamate pyruvate transaminase (GPT or ALAT)

• Alkaline phosphatase

• Pancreatic amylase

• Protein (total)

• Albumin

• Bilirubin (total)

In addition to:

• Age

• Sex

• Height

• Weight

• Blood pressure

• Nicotine Status

• Chest Pain Status

• Medication

• Resting ECG

The Cardio Explorer® is an in-vitro diagnostic Software as a Medical Device (SaMD) intended to support healthcare professionals in the diagnostic assessment and risk stratification of adult patients being evaluated for suspected chronic coronary syndrome (CCS). It is specifically designed for use by licensed physicians who are responsible for evaluating such patients, typically within a professional healthcare environment. No special training is required beyond basic IT knowledge and familiarity with the instructions for use.

The software is intended for patients aged 18 years or older in whom CCS is suspected. It is not validated for use in patients with acute coronary syndrome, pediatric populations, or those with prior revascularization. Importantly, there are no known contraindications for its use.

Cardio Explorer® functions as an adjunct diagnostic tool and must not replace the clinical judgment of a qualified physician. The results provide a numerical estimate of the likelihood of obstructive coronary artery disease (CAD), aligned with ESC guidelines for the management of CCS, but should never serve as the sole basis for diagnosis or clinical decision-making.

The Cardio Explorer® achieves its high accuracy by going far beyond single-parameter assessment. Instead of relying on isolated values such as cholesterol or blood pressure alone, the algorithm simultaneously evaluates 32 routinely available clinical and laboratory parameters. This creates billions of potential interactions, allowing the system to recognize complex, non-linear relationships that are not accessible to the human eye or conventional statistical models.

The underlying AI-X-Engine, which was used to generate the algorithm of the Cardio Explorer®, applies an evolution-based approach that captures dynamic dependencies between input variables and the clinical endpoint (obstructive CAD). Unlike adaptive algorithms, it is fixed and does not change over time, ensuring reproducibility and control. Each calculation results in a personalized probability tailored to the specific patient context.

Accuracy is further supported by extensive clinical validation across independent cohorts with different prevalence levels, age groups and both sexes. In these studies, the algorithm consistently demonstrated strong discriminative performance (AUC 0.84–0.88) and excellent calibration, leading to high sensitivity and negative predictive values in lower-risk populations.

In summary, Cardio Explorer®’s high accuracy results from its ability to integrate and interpret the interplay of dozens of parameters simultaneously, translating complex real-world data into a robust, guideline-aligned risk estimate for each individual patient.

Cardio Explorer® operationalizes contemporary ESC Guidelines for the Diagnosis and Management of Chronic Coronary Syndrome through an AI-based multi-marker algorithm that quantifies the clinical likelihood of obstructive coronary artery disease (CAD) and recommends the next guideline-directed diagnostic step. Establishing clinical likelihood is a cornerstone of evidence-based CAD assessment, serving as the foundation for determining which patients should undergo further non-invasive or invasive testing according to international recommendations.

With a negative predictive value exceeding >95%, Cardio Explorer® enables safe and efficient exclusion of CAD, substantially reducing unnecessary diagnostic procedures while preserving diagnostic accuracy even in atypical or asymptomatic presentations.

Clinically, it can be applied across the full spectrum of care – from primary and specialty cardiology to emergency medicine – and also serves as a robust tool for proactive cardiovascular risk stratification. Its simplicity, rapid turnaround and cost-effectiveness support broad implementation and equitable access to precision cardiovascular diagnostics in routine practice.