Flexible AI-powered wearable patch for non-invasive medication compliance monitoring

Abstract

We provide a flexible, AI-powered wearable patch system for non-invasive medication compliance monitoring. [Solution] This system 100 integrates physiological sensors 101 to detect heart rate variability, skin temperature, electrocutaneous activity, and swallowing-related muscle signals. A drug detection sensor 102 identifies interactions with drug containers, and a medication event sensor 103 detects gestures or physiological signals indicating medication. An embedded AI processor 104 receives multimodal sensor data and performs digital filtering, normalization, artifact removal, and machine learning-based classification to determine medication confirmation events, missed doses, or irregular medication patterns. A power management unit 105 controls the power supply to ensure continuous operation. A wireless communication module 106 transmits processed medication data to an external device. This enables reliable, objective, and real-time medication confirmation without relying on invasive ingestion sensors or patient reports.

Description

Technical Field

[0001] The present invention relates to a wearable healthcare monitoring system, and more particularly, to a flexible AI - equipped wearable patch configured to non - invasively and multimodally monitor medication compliance and autonomously verify it using embedded artificial intelligence and integrated sensor technology.

Background Art

[0002] The background description includes information that may be useful in understanding the present invention. Any information provided in this description is not admitted to be prior art or relevant to the present invention. Also, any publications specifically or implicitly cited are not admitted to be prior art.

[0003] Medication non - compliance remains a significant issue in the management of chronic and acute diseases. Patients often forget to take their medications, take them late, or inaccurately report their medication status. Such discrepancies reduce treatment effectiveness, increase hospitalization rates, and raise healthcare costs. Therefore, a reliable and objective monitoring solution is essential to ensure treatment compliance and improve overall patient outcomes in various healthcare settings.

[0004] Many existing medication compliance monitoring technologies rely on patient operation. For example, smart pill bottles, reminder applications, or manual recording systems. These methods are highly dependent on user self - reporting and continuous operation, making inaccuracies likely. Furthermore, these systems cannot independently verify whether the drug has actually been taken, so it is impossible to definitely confirm whether the drug has actually been administered.

[0005] Invasive approaches, such as ingestible sensors, attempt to confirm ingestion events with greater reliability. However, such systems come with increased costs, regulatory complexities, and patient discomfort. Furthermore, their reliance on ingestible components limits their widespread adoption and may make them unsuitable for routine or long-term monitoring in diverse healthcare settings.

[0006] Furthermore, many existing wearable solutions heavily rely on cloud-based analytics for data interpretation. This continued reliance on remote processing introduces delays, potential connectivity issues, and privacy risks. Delays in analysis can hinder timely intervention, and the transmission of sensitive health data raises security concerns. These issues highlight the need for secure analytics solutions with on-device processing. [Overview of the project]

[0007] This invention provides a flexible AI-powered wearable patch system (100) for non-invasive medication compliance monitoring. The system integrates a physiological sensor (101), a drug detection sensor (102), and a medication event sensor (103) to acquire multimodal signals related to medication intake. An embedded AI processor (104) analyzes the fused sensor data to autonomously classify medication compliance events in real time.

[0008] The system further includes a power management unit (105) to ensure continuous operation and a wireless communication module (106) for securely transmitting data to an external device. By combining multimodal sensing and on-device artificial intelligence, the invention enables reliable, objective, and real-time medication compliance without relying on invasive intake sensors or patient reports. [Brief explanation of the drawing]

[0009] [Figure 1]The diagram shows a flexible AI-powered wearable patch system (100) that integrates a physiological sensor (101), a drug detection sensor (102), and a medication event sensor (103) within a patch structure. An embedded AI processor (104) is operablely connected to each sensor for multimodal data analysis. A power management unit (105) provides stable power, and a wireless communication module (106) transmits processed medication data to an external device. [Modes for carrying out the invention]

[0010] The embodiments shown in the attached drawings will be described in detail below.

[0011] This invention relates to a flexible AI-powered wearable patch system (100) designed for non-invasive medication compliance monitoring. The system (100) is designed for continuous wear and is configured to autonomously verify medication events. By integrating multimodal sensing and embedded artificial intelligence, it is possible to objectively and in real time evaluate medication compliance without relying on patient self-reporting or invasive intake sensors.

[0012] The system (100) includes a physiological sensor (101) configured to detect biological parameters such as heart rate variability, skin temperature, electrocutaneous activity, and muscle signals related to swallowing. These physiological signals provide basic health data and capture subtle biological responses related to drug intake. By continuously acquiring these signals, it becomes possible to distinguish between normal physiological fluctuations and intake-related changes.

[0013] A drug detection sensor (102) is integrated into the system (100) and detects interaction with drug containers. This drug detection sensor (102) can detect proximity or handling events using radio frequency, volume detection, or optical sensing technology. This sensing layer provides contextual validation and improves the reliability of ingestion confirmation by correlating with physiological and behavioral signals.

[0014] The medication event sensor (103) is configured to detect gestures or biomechanical signals indicating medication intake. This medication event sensor (103) acquires motion patterns such as hand-to-mouth movements and swallowing-related signals. By monitoring temporal patterns related to medication behavior, the system improves detection accuracy and reduces false detections caused by unrelated movements.

[0015] Signals generated by the physiological sensor (101), drug detection sensor (102), and medication event sensor (103) are received by the embedded AI processor (104). The embedded AI processor (104) performs digital filtering, normalization, and artifact removal to eliminate noise and irrelevant data. This preprocessing ensures that only significant sensor information is used for medication determination.

[0016] Furthermore, the embedded AI processor (104) runs machine learning algorithms trained to recognize temporal and multimodal patterns associated with medication intake. By fusing data between sensor inputs, the processor classifies events as “medication confirmed,” “possible medication,” “medication missed,” or “irregular medication pattern.” On-device processing reduces latency and improves data privacy by minimizing reliance on external cloud systems.

[0017] The power management unit (105) is integrated into the system (100) and controls the power supply to all functional components. The power management unit (105) may include a thin-film rechargeable battery and energy optimization circuitry. By controlling power distribution, it ensures stable operation during prolonged wear while maintaining efficient energy consumption throughout sensing, processing, and communication operations.

[0018] The wireless communication module (106) is operationally connected to the embedded AI processor (104) and transmits processed medication data to an external device. The wireless communication module (106) can exchange data securely and with low power consumption using the Bluetooth Low Energy protocol. This communication capability enables remote monitoring, clinical evaluation, and integration with digital health management platforms. [Explanation of Symbols]

[0019] 100 Systems 101 Physiological Sensor 102 Drug detection sensor 103 Medication Event Sensor 104 AI Processors 105 Power Management Unit 106 Wireless Communication Module

Claims

1. A flexible AI-powered wearable patch system (100) for non-invasive medication compliance monitoring, It includes the following: a. A physiological sensor (101) configured to detect physiological parameters including heart rate variability, skin temperature, electrocutaneous activity, and swallowing-related muscle signals; b. A drug detection sensor (102) configured to detect the handling or presence of a drug container; c. A medication event sensor (103) configured to detect a signal indicating a drug ingestion event; d. An embedded AI processor (104) operably connected to a physiological sensor (101), a drug detection sensor (102), and a medication event sensor (103), configured to perform digital filtering, normalization, artifact removal, and multimodal data fusion, and to classify the detected signals into medication confirmation events, missed medications, or irregular medication patterns; e. A power management unit (105) electrically connected to the embedded AI processor (104) and configured to control the power supply; f. A wireless communication module (106) operably connected to an embedded AI processor (104) and configured to transmit processed medication data to an external device.

2. The system (100) according to claim 1, wherein the physiological sensor (101) includes a sensor that detects heart rate variability, skin temperature, electrocutaneous activity, and swallowing-related muscle activity.

3. A system (100) according to claim 1, wherein the drug detection sensor (102) is a system that identifies a drug container using radio frequency, volume detection, or optical detection.

4. The system (100) according to claim 1, wherein the medication event sensor (103) is a system for detecting a temporal pattern corresponding to a hand-to-mouth gesture or swallowing signal.

5. A system (100) according to claim 1, wherein the embedded AI processor (104) includes a low-power microcontroller that executes a machine learning algorithm trained to recognize temporal patterns related to intake.

6. A system (100) according to claim 1, wherein the power management unit (105) includes a thin-film rechargeable battery and / or an energy harvesting circuit.

7. A system (100) according to claim 1, wherein the wireless communication module (106) operates using the Bluetooth Low Energy protocol.

Images