Entangled Health
Quantum reservoir computing, classical echo state networks, and the future of AI in healthcare — explored in this Entangled Health Experiments episode. We trace the path from biological time series (ECG, EEG, glucose, voice) through reservoir computing, NG-RC, and quantum dynamics, asking whether intelligence is something we build or something matter already does. Part of the Entangled Health Experiments mini-series, with a companion six-part Substack series. This episode is part of Entangled Health Experiments — a special mini-series where we deliberately stress-test ideas in public with tools like Perplexity and NotebookLM as thinking partners. No polished keynote, no finished answers, just live exploration of hard questions at the intersection of quantum science, computation, and health. In this session, we follow one of the most underappreciated threads in computational medicine: reservoir computing from its classical origins to its quantum frontier and end on a question that refuses to go away: if a bucket of water, an optical fibre loop, a neuromorphic chip, or a handful of quantum spins can process information and remember the past, are we building intelligence, or just learning how to tap into something matter already does? Read more on substack 6-Part Article: https://thomasehmer1.substack.com/p/i-built-something-here-is-what-it What we cover: Why biological time series are so hard. ECG, EEG, continuous glucose monitoring, voice, multi-omics — they are nonstationary, multiscale, noisy, and label-poor. Standard deep learning architectures (LSTM, Transformer) were not designed for this regime. Reservoir computing as an alternative. Echo state networks, liquid state machines, and next-generation reservoir computing (NG-RC) separate dynamics from learning: fix the reservoir, train only a linear readout. Fast, small-data-friendly, interpretable. Quantum reservoir computing (QRC). What the exponential Hilbert space actually buys you — and where the encoding bottleneck and dissipation requirement quietly constrain it. Seven qubits matching hundreds of classical neurons is real. But the "2ⁿ" story requires careful unpacking. The dissipation paradox. Why today's noisy NISQ hardware is sometimes structurally better for reservoir computing than tomorrow's fault-tolerant machines — and what that means for the hardware road Tools used in this experiment: Guests are "Sheela and Himbert, the famous voices form NotebookLM - as I did not find names anywhere, I named them. Content prepared and fed to Sheela and Himbert by perplexity.ai Sheela and Himbert generated via notebooklm.google.com Mixed with Logic on Mac. Host & Guests * Thomas Ehmer [https://entangled-health.transistor.fm/people/thomas-ehmer] - Host We don't simplify for comfort. We entangle. Submit your questions or propose a guest in the comments, or join the discussion on Substack: https://thomasehmer1.substack.com/ If you like it — thanks for subscribing and leaving a thumbs up wherever you listen 👍
27 episodios
Comentarios
0Sé la primera persona en comentar
¡Regístrate ahora y únete a la comunidad de Entangled Health!