AI-supported system for detecting heart abnormalities

Abstract

A system for detecting cardiac anomalies, consisting of a user interface for inputting health-related data, a processor communicatively connected to the user interface, and a trained artificial intelligence model executed by the processor to analyze the inputs and predict potential cardiac anomalies, wherein the processor generates warning messages on the user interface when an anomaly is detected.

Description

AREA OF INVENTION

[0001] The invention relates to medical monitoring devices with integrated user interfaces, biosignal inputs and AI models for analyzing cardiovascular measurements and providing real-time anomaly alerts and long-term tracking. BACKGROUND OF THE INVENTION

[0002] Cardiovascular diseases remain among the leading causes of morbidity and mortality. Early detection of risks or arrhythmias can improve treatment outcomes through timely intervention and risk modification. However, current screening methods are episodic and often fail to detect intermittent abnormalities. AI models trained on large ECG and clinical datasets can enhance prediction and detection beyond clinical interpretation, identifying risks and arrhythmias even in the presence of a normal 12-lead ECG. This demonstrates additional prognostic value and higher sensitivity for subtle patterns. Approvals for AI cardiology platforms that analyze ECG data to detect atrial fibrillation and other arrhythmias demonstrate the clinical applicability of software-assisted detection in combination with wearable or ambulatory devices in everyday practice.A device-centric system that combines user-generated health data, sensor data, AI inference, alerting, and the visualization of historical trends enables accessible, proactive cardiovascular monitoring for diverse user profiles while supporting clinical workflows. SUMMARY OF THE INVENTION

[0003] The invention relates to a system with a user interface for inputting health data and displaying results, a processor for aggregating multiparameter inputs (including demographic data and biosignals), and a trained AI model for analyzing the inputs and predicting cardiac anomalies. When thresholds are exceeded, the system issues warnings and recommendations via the user interface. The system supports secure data storage for long-term monitoring, visual feedback for improved comprehension, and adaptive functionality for different health profiles. This enables proactive management and facilitates early medical assessment upon detection of anomalies. DETAILED DESCRIPTION

[0004] The system includes a user interface for mobile devices, web clients, or dedicated displays. This allows for the input and verification of demographic data, laboratory values ​​such as cholesterol, and vital parameters such as heart rate. If required, it can also connect external sensors or ECG wearables for biosignal acquisition. A processor controls data acquisition, preprocessing, and feature extraction from structured fields and time series, normalizes values, and derives risk features compatible with modern AI-powered cardiology procedures for risk assessment and arrhythmia detection. The AI ​​model is trained on historical datasets of ECG recordings and clinical outcomes to identify patterns indicative of cardiac abnormalities.Models validated using large ECG corpora demonstrated improved prediction of adverse events and detection of silent arrhythmias, thus supporting their use in proactive monitoring. The processor applies the trained model to the user's current input to generate a probability assessment for anomalies and a classification (e.g., suspected atrial fibrillation). When thresholds are exceeded, alerts are issued via the user interface to enable timely attention and, optionally, a physician review. The user interface's visual elements provide clear summaries and recommendations, including risk trends and action plans, adhering to best practices for patient engagement and understanding in AI-assisted cardiovascular monitoring.The system stores historical data on metrics and model outputs, enabling trend analysis to monitor changes over time and identify emerging risks earlier than with episodic screenings—aligned with longitudinal AI approaches in cardiology. Connectivity to compatible ECG patches or wearables allows for continuous or intermittent data collection. Some FDA-approved platforms offer clinically relevant arrhythmia detection capabilities that can be integrated for enhanced anomaly detection. The architecture ensures data security and access control to protect health information while supporting the export of clinical application summaries and model updates based on post-market insights, in accordance with applicable regulatory frameworks for AI in medical software.The system is adaptable to various user scenarios and health profiles through configurable data inputs, thresholds, and device integrations, thus expanding its application range from preventive monitoring to physician-supervised programs. In use, the system functions as a combined software-hardware ecosystem that utilizes AI to improve the early detection and ongoing management of cardiovascular risks, enhance user-friendliness, and enable preventive strategies.

Claims

[1] A system for detecting cardiac anomalies, comprising a user interface for inputting health-related data, a processor communicatively connected to the user interface, and a trained artificial intelligence model executed by the processor to analyze the inputs and predict potential cardiac anomalies, wherein the processor generates warning messages on the user interface when an anomaly is detected. [2] System according to claim 1, wherein the input variables include demographic factors, laboratory values ​​including cholesterol, vital parameters including heart rate and optionally biosignals from connected ECG wearables and the processor performs preprocessing and feature extraction prior to model inference. [3] System according to claim 1, wherein the user interface displays visual feedback including risk assessments, recommendations and trend graphs, and the system stores historical health data and model outputs to track changes over time. [4] System according to claim 1, wherein the processor integrates outputs from portable ECG devices enabling arrhythmia detection and synchronizes alerts with the workflows of medical assessment, utilizing validated AI ECG approaches for improved sensitivity.