Breath analysis is an emerging technique in the field of diagnostics. The presence of thousands of gases and volatile organic compounds (VOCs), many of them at part per billion (ppb) concentration levels, require the development of ultrasensitive and selective detection approaches, which pose challenges still trying to be addressed by the scientific community. Here, we describe two approaches that provide a substantial contribution to the development of gas sensors. The first one is based on modifications of the used sensing material, namely a specific surface functionalization based on gold nanoparticles of carbon nanotubes to achieve selectivity toward hydrogen sulfide, together with the implementation of multiple sensors for self-validation. The second one focuses on the analysis method, implementing machine learning algorithms to maximize the information obtained from each single sensor to distinguish gases based on their interaction kinetics with the sensor. The combination of both approaches is foreseen as a powerful tool for the development of new smart sensing platforms with high potential in terms of analytical efficiency.