The methodological challenges and research demands at Neuroscape require new computational frameworks for rapid processing and integration of real-time recordings from the brain and the body. Our computational platforms are based on advanced machine learning and statistical techniques (e.g. deep learning models) that reliably predict cognitive and affective states from multimodal biosensing (MMBS) data. The current focus is on emotion and attention, as these states are among the most informative types of experiential states during inaccessible experiences (e.g, meditation, video game play, and psychedelic treatments). Understanding biomarkers of emotion and attention with MMBS data will enable us to automatically predict experiential states during closed-loop video game treatments and will be used to modify treatments tailored to an individual’s state in real-time to maximize therapeutic potential.