Intelligent Inhaler: Revolutionizing Respiratory Care with Smart Technology

Abstract

Respiratory diseases such as asthma and Chronic Obstructive Pulmonary Disease (COPD) impact over 300 million individuals globally. Despite the availability of effective pharmacological treatments, poor medication adherence and improper inhaler technique remain critical barriers to successful disease management. Studies have shown that a significant percentage of patients misuse inhalers, leading to ineffective therapy, increased hospitalization, and elevated healthcare costs.

This paper presents the design and evaluation of an intelligent inhaler system that leverages embedded sensing technologies, Internet of Things (IoT) connectivity, and on-device machine learning to deliver real-time monitoring and feedback to users. Drawing inspiration from the design concepts outlined in US Patent 11090449 and grounded in empirical findings from recent research on digital inhalers, the proposed system offers a clinically viable solution that combines hardware innovation with software intelligence.

The system comprises three key sensors—flow rate, pressure, and humidity—that are integrated with an ESP32 microcontroller. The ESP32 executes a quantized classification model built using TensorFlow Lite Micro, enabling the system to categorize inhalation attempts as either correct or incorrect based on sensor data patterns. Unlike commercial devices that rely on cloud-based analytics, this architecture supports edge processing, which significantly reduces latency and ensures uninterrupted performance even in low-connectivity settings.

Users receive immediate feedback via a 0.96” OLED screen and a piezo buzzer, both of which provide intuitive prompts designed to improve adherence and technique. Simultaneously, all sensor data and classification results are uploaded to a Firebase Realtime Database for persistent storage and remote access. A companion Android application allows patients to view their inhalation history and receive dosage reminders, while a clinician-facing web dashboard—developed using React.js—supports long-term treatment tracking and real-time patient monitoring.

In contrast to existing digital inhalers, the proposed system offers real-time feedback, on-device intelligence, open-source hardware compatibility, and enhanced sensor fusion. Evaluations demonstrated 85.3% average classification accuracy, sub-second latency, and operational reliability over multiple days on a single battery charge. These results validate the feasibility of the system as a low-cost, extensible, and effective platform for modern respiratory disease management.