In the global healthcare landscape, paper electrocardiograms (ECGs) remain a daily tool in rural and remote clinics where connectivity and computational capacity are limited. These physical records, however, are excluded from modern artificial intelligence (AI)-assisted diagnostic systems, causing delays in detecting critical conditions such as acute myocardial infarction (AMI). To address this gap, ECGLight emerges as a lightweight end-to-end framework that digitizes paper ECGs from a smartphone photo and, in under 30 seconds, reconstructs a calibrated 12-lead signal and detects infarction pathologies, all running exclusively on CPU without internet connectivity. This breakthrough represents a true democratization of AI in cardiology, enabling any facility with a mobile device to access high-precision diagnosis.
The architecture of ECGLight combines image processing techniques with machine learning models optimized for low-resource devices. The digitization module extracts traces from each lead, corrects distortions, and scales voltages using printed references, generating a high-fidelity numerical signal. Next, a lightweight classifier based on convolutional networks trained on 21,799 ECGs from the PTB-XL dataset identifies signs of myocardial infarction (MI). To provide transparency, the system incorporates SHAP (SHapley Additive exPlanations) that highlight the signal regions most influential in the decision. Reported results are notable: 95.51% accuracy (F1=0.9519) on PTB-XL and 88.89% accuracy (F1=0.8862) on the hospital-acquired ECG-Matrix dataset, outperforming many cloud-based solutions that require high bandwidth.
From a technical and business perspective, ECGLight exemplifies how software engineering can optimize complex models for hardware-constrained environments. Techniques such as quantization, pruning, and knowledge distillation reduce model size and computational consumption without sacrificing accuracy. This is crucial in healthcare, where devices often have limited resources and response speed can save lives. Development companies like Q2BSTUDIO apply these strategies to create custom software applications that integrate lightweight AI into mobile devices, embedded systems, or diagnostic equipment. The ability to customize every component—from image capture to clinical interpretation—ensures solutions aligned with the real workflows of hospitals and clinics.
Implementing a system like ECGLight also requires secure management of patient data. Digitized clinical information is sensitive and must comply with regulations such as HIPAA or GDPR. Therefore, Q2BSTUDIO offers specialized cybersecurity services that protect both transmission and storage of ECGs, whether on local infrastructure or in the cloud. Deploying the solution on cloud platforms like AWS or Azure allows processing to scale on demand, using serverless functions and containers to maintain efficiency, while applying encryption and strict access controls. This combination of local lightness and cloud power ensures that even clinics without their own servers can benefit from secure assisted diagnosis.
Beyond point-of-care diagnosis, the data generated from digitizations can be integrated into Business Intelligence (BI) systems using tools like Power BI. Hospitals can build dashboards to monitor infarction incidence by region, treatment effectiveness, or response times. It is even possible to incorporate intelligent agents (AI agents) that automate early alerts when high-risk patterns are detected in real time, speeding communication with the medical team. Q2BSTUDIO develops these integrations, combining predictive analytics with interactive visualization to turn raw data into actionable clinical decisions.
The case of ECGLight demonstrates that healthcare innovation does not always require large infrastructure. With a focus on optimization and user-centered design, it is possible to bring artificial intelligence to the most disconnected places. For technology companies, this represents an opportunity to collaborate with healthcare institutions to build lightweight, secure, and scalable software. At Q2BSTUDIO, we believe that the union of custom software, cloud computing, and cybersecurity is the key to extending the reach of digital health. If your organization needs to implement similar solutions, we invite you to explore how we can help through our custom software development and cloud solutions on AWS and Azure.
