PW15 - NEUROSCENT: A DEEP LEARNING EEG PLATFORM FOR OBJECTIVE OLFACTORY PROFILING IN CLINICAL NUTRITION

PW15

NEUROSCENT: A DEEP LEARNING EEG PLATFORM FOR OBJECTIVE OLFACTORY PROFILING IN CLINICAL NUTRITION

T. Marchetti^1,*, C. Serantoni¹, M. M. De Giulio², A. Riente², D. Hatem², M. Santantonio¹, G. C. Passali^1,2, J. Galli^1,2, M. De Spirito^1,2, G. Maulucci^1,2

¹Fondazione Policlinico Universitario Agostino Gemelli IRCCS, ²Università Cattolica del Sacro Cuore, Rome, Italy

Rationale: Olfactory function is tightly linked to appetite regulation and  metabolic processes. However, current clinical methods for olfactory assessment are subjective and lack consistency.

Methods: We present NEUROSCENT, a novel deep learning-based framework designed to objectively evaluate olfactory function using electroencephalographic (EEG) measurements. Our system employs a long short-term memory module and a multi-head self-attention layer to capture the temporal dynamics of odour-induced brain activity and classify with high accuracy neural patterns associated with specific odours. 

Results: In a study involving 75 recordings from healthy volunteers, the model achieved over 90% accuracy in binary classification tasks, distinguishing between the presence or absence of olfactory stimuli. Additionally, it demonstrated high performance in multi-class classification, discerning between distinct odor types, highlighting its potential for fine-grained olfactory profiling.

Conclusion: With NEUROSCENT, we propose a novel EEG-based framework for objective monitoring of olfactory function. These encouraging preliminary results support potential, yet to be validated, applications in the nutritional field—such as investigating appetite regulation and sensory-driven eating behavior. 

References: Funded by the European Union – Next Generation EU – PNRR M6C2 - Investment 2.1 Enhancement and Strengthening of Biomedical Research in the National Health Service.

Project code: PNRR-POC-2023-12377964

Disclosure of Interest: None declared