- Customer Stories
- /
- AI Cardiac Diagnostics Software
AI Cardiac Diagnostics Software
AI cardiac diagnostics software using real-time ECG, PCG, and PPG signal processing.
Accuracy – exceeded legacy PoC performance
To validate the startup’s business hypothesis
Reliable extraction of key cardiovascular metrics (PEP, LVET, IVCT) from noisy ECG, PCG, and PPG signals gathered by wearable devices.
They always delivered on time and showed 100% transparency in the project development.
Alejandro Gonzalez
Product manager, Monitoring LifeThe primary challenge was handling unstable input: missing peaks, poor signal quality, and hardware inconsistency.
Solution
We developed an ML-based application that integrated directly with the client’s sensors. The system performed denoising, fiducial point detection, and signal segmentation. It delivered over 70% accuracy and sensitivity — significantly outperforming their legacy PoC — and included a fully automated pipeline for real-time metric extraction and analysis.
To enable real-time cardiovascular signal processing and robust ML performance, Blackthorn AI applied:
Project duration
02 Weeks
Review of legacy PoC architecture, requirements gathering from early clinical trials and sensor data analysis & feasibility study
6.5 Weeks
Preprocessing and denoising modules, ECG/PCG/PPG signal peak detection and feature extraction and metric derivation (PEP, LVET)
04 Weeks
Infrastructure & scalable backend design and MVP UI & end-user reporting integration
02 Weeks
Developer-friendly feedback loop and final deployment & delivery
Team Size
Delivering Impact
Beyond the values already highlighted, there’s even more to discover. Our commitment to innovation, client success, and impactful results sets us apart
99%
Signal processing accuracyDelivered over 70% accuracy on ECG/PCG/PPG data vs. 50–55% from legacy PoC.
4
WeeksPrototype validated core clinical assumption within first month.
3x
Faster metric extractionNew pipeline delivers features (PEP, LVET) ~3× faster than previous setup.
100%
Sensor compatibilityFully integrated with client’s existing wearable hardware, no redesign needed.
<5%
Manual intervention neededMajority of processing is now automated, with <5% of cases requiring review.
