Late Breaking Abstract - Engineering an AI-base mesh nebulizer system for personalized respiratory drug delivery.

Aerosol therapy is widely used in pulmonary disease treatments. However, around 70% of patients showed insufficient drug intake, resulting in frequent hospitalizations and unexpected emergencies. We engineered an AI-base nebulization system with a commercial mesh nebulizer (MicroBase Technology, Taiwan) to precisely deliver the required dosage. The Mel-spectrogram of breathing nasal sounds was analyzed using convolutional neural network architecture for activating nebulization during inhalation. Compared to regular nebulization, our drug delivery efficiency increased approximately 250% under various breathing conditions, such as chronic obstructive pulmonary disease, asthma, interstitial lung disease and normal status. Clinical results (ClinicalTrials.gov <ext-link>NCT06157515</ext-link>) showed the AI-based nebulizer provided averaged 225% improved delivery efficiency (n = 24) under 80dB environmental noise. Moreover, using the Saldolin® (salbutamol sulfate) reagent, we tuned the aerosol diameter from fine (~5 µm) to extra fine particle (~2 µm), which suggested the manipulation of the aerosol deposition sites in the respiratory system. Our AI nebulizer shows the potential to optimize aerosol delivery according to each patient's unique breathing conditions, which is expected to contribute to personalized aerosol therapies in the future. <fig><object-id>erj;64/suppl_68/PA2339/F1</object-id><object-id>F1</object-id><object-id>F1</object-id><graphic></graphic></fig>